diff --git a/assets/i18n/de.yaml b/assets/i18n/de.yaml
index d6b9fd366..78d5fc778 100644
--- a/assets/i18n/de.yaml
+++ b/assets/i18n/de.yaml
@@ -171,6 +171,380 @@
 "multicam.switch_9": "Zu Kamera 9 wechseln"
 "effect_library.hint": "Doppelklicken, um zum ausgewählten Clip hinzuzufügen"
 "effect_library.group.builtin": "Integriert"
+"effect_library.group.color": "Color"
+"effect_library.group.filter": "Filter"
+"effect_library.group.distort": "Distort"
+"effect_library.group.keying": "Keying"
+"effect_library.group.generator": "Generator"
+"effect_library.group.math": "Math"
+"effect_library.group.time": "Time"
+"effect_library.group.channel": "Channel"
+"effect_library.group.transition": "Transition"
+"effect_library.group.general": "General"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "Polygon"
+"node.polygon.input.base_in": "Base"
+"node.polygon.input.points_in": "Points"
+"node.polygon.input.color_in": "Color"
+"node.ortho.name": "Orthographic Matrix"
+"node.ortho.input.pos_in": "Position"
+"node.ortho.input.rot_in": "Rotation"
+"node.ortho.input.scale_in": "Scale"
+"node.ortho.input.uniform_scale_in": "Uniform Scale"
+"node.ortho.input.anchor_in": "Anchor Point"
+"node.transform.name": "Transform"
+"node.transform.input.tex_in": "Texture"
+"node.transform.input.parent_in": "Parent"
+"node.transform.input.autoscale_in": "Auto-Scale"
+"node.transform.input.interpolation_in": "Interpolation"
+"node.transform.input.pos_in": "Position"
+"node.transform.input.rot_in": "Rotation"
+"node.transform.input.scale_in": "Scale"
+"node.transform.input.uniform_scale_in": "Uniform Scale"
+"node.transform.input.anchor_in": "Anchor Point"
+"node.volume.name": "Volume"
+"node.volume.input.samples_in": "Samples"
+"node.volume.input.volume_in": "Volume"
+"node.pan.name": "Pan"
+"node.pan.input.samples_in": "Samples"
+"node.pan.input.panning_in": "Pan"
+"node.math.name": "Math"
+"node.math.input.method_in": "Method"
+"node.math.input.param_a_in": "Value"
+"node.math.input.param_b_in": "Value"
+"node.time.name": "Time"
+"node.time.input.enabled_in": "Enabled"
+"node.trigonometry.name": "Trigonometry"
+"node.trigonometry.input.method_in": "Method"
+"node.trigonometry.input.x_in": "Value"
+"node.blur.name": "Blur"
+"node.blur.input.tex_in": "Input"
+"node.blur.input.method_in": "Method"
+"node.blur.input.radius_in": "Radius"
+"node.blur.input.horiz_in": "Horizontal"
+"node.blur.input.vert_in": "Vertical"
+"node.blur.input.directional_degrees_in": "Direction"
+"node.blur.input.radial_center_in": "Center"
+"node.blur.input.repeat_edge_pixels_in": "Repeat Edge Pixels"
+"node.solidgenerator.name": "Solid"
+"node.solidgenerator.input.color_in": "Color"
+"node.merge.name": "Merge"
+"node.merge.input.base_in": "Base"
+"node.merge.input.blend_in": "Blend"
+"node.stroke.name": "Stroke"
+"node.stroke.input.tex_in": "Input"
+"node.stroke.input.color_in": "Color"
+"node.stroke.input.radius_in": "Radius"
+"node.stroke.input.opacity_in": "Opacity"
+"node.stroke.input.inner_in": "Inner"
+"node.textgenerator.name": "Text (Legacy)"
+"node.textgenerator.input.text_in": "Text"
+"node.textgenerator.input.html_in": "Enable HTML"
+"node.textgenerator.input.color_in": "Color"
+"node.textgenerator.input.valign_in": "Vertical Align"
+"node.textgenerator.input.font_in": "Font"
+"node.textgenerator.input.font_size_in": "Font Size"
+"node.text2.name": "Text (Legacy)"
+"node.text2.input.base_in": "Base"
+"node.text2.input.pos_in": "Position"
+"node.text2.input.size_in": "Size"
+"node.text2.input.color_in": "Color"
+"node.text2.input.text_in": "Text"
+"node.text2.input.html_in": "Enable HTML"
+"node.text2.input.valign_in": "Vertical Align"
+"node.text2.input.font_in": "Font"
+"node.text2.input.font_size_in": "Font Size"
+"node.text3.name": "Text"
+"node.text3.input.base_in": "Base"
+"node.text3.input.pos_in": "Position"
+"node.text3.input.size_in": "Size"
+"node.text3.input.plain_text_in": "Text"
+"node.text3.input.text_in": "Text"
+"node.text3.input.font_family_in": "Font Family"
+"node.text3.input.font_size_in": "Font Size"
+"node.text3.input.outline_enabled_in": "Outline"
+"node.text3.input.outline_color_in": "Outline Color"
+"node.text3.input.outline_width_in": "Outline Width"
+"node.text3.input.glow_enabled_in": "Glow"
+"node.text3.input.glow_color_in": "Glow Color"
+"node.text3.input.glow_radius_in": "Glow Radius"
+"node.text3.input.color_in": "Color"
+"node.text3.input.valign_in": "Vertical Alignment"
+"node.text3.input.args_in": "Arguments"
+"node.mosaicfilter.name": "Mosaic"
+"node.mosaicfilter.input.tex_in": "Texture"
+"node.mosaicfilter.input.horiz_in": "Horizontal"
+"node.mosaicfilter.input.vert_in": "Vertical"
+"node.crop.name": "Crop"
+"node.crop.input.tex_in": "Texture"
+"node.crop.input.left_in": "Left"
+"node.crop.input.top_in": "Top"
+"node.crop.input.right_in": "Right"
+"node.crop.input.bottom_in": "Bottom"
+"node.crop.input.feather_in": "Feather"
+"node.value.name": "Value"
+"node.value.input.type_in": "Type"
+"node.value.input.value_in": "Value"
+"node.timeremap.name": "Time Remap"
+"node.timeremap.input.time_in": "Time"
+"node.timeremap.input.input_in": "Input"
+"node.shape.name": "Shape"
+"node.shape.input.type_in": "Type"
+"node.shape.input.base_in": "Base"
+"node.shape.input.pos_in": "Position"
+"node.shape.input.size_in": "Size"
+"node.shape.input.color_in": "Color"
+"node.shape.input.radius_in": "Radius"
+"node.colordifferencekey.name": "Color Difference Key"
+"node.colordifferencekey.input.tex_in": "Input"
+"node.colordifferencekey.input.garbage_in": "Garbage Matte"
+"node.colordifferencekey.input.core_in": "Core Matte"
+"node.colordifferencekey.input.color_in": "Key Color"
+"node.colordifferencekey.input.highlights_in": "Highlights"
+"node.colordifferencekey.input.shadows_in": "Shadows"
+"node.colordifferencekey.input.mask_only_in": "Show Mask Only"
+"node.despill.name": "Despill"
+"node.despill.input.tex_in": "Input"
+"node.despill.input.color_in": "Key Color"
+"node.despill.input.method_in": "Method"
+"node.despill.input.preserve_luminance_input": "Preserve Luminance"
+"node.group.name": "Group"
+"node.opacity.name": "Opacity"
+"node.opacity.input.tex_in": "Texture"
+"node.opacity.input.opacity_in": "Opacity"
+"node.flip.name": "Flip"
+"node.flip.input.tex_in": "Input"
+"node.flip.input.horiz_in": "Horizontal"
+"node.flip.input.vert_in": "Vertical"
+"node.noise.name": "Noise"
+"node.noise.input.base_in": "Base"
+"node.noise.input.strength_in": "Strength"
+"node.noise.input.color_in": "Color"
+"node.timeoffset.name": "Time Offset"
+"node.timeoffset.input.time_in": "Time"
+"node.timeoffset.input.input_in": "Input"
+"node.transition.name": "Transition"
+"node.transition.input.tex_in": "From"
+"node.transition.input.blend_in": "To"
+"node.transition.input.progress_in": "Progress"
+"node.transition.input.type_in": "Type"
+"node.transitionfx.name": "Transition FX"
+"node.transitionfx.input.tex_in": "From"
+"node.transitionfx.input.blend_in": "To"
+"node.transitionfx.input.progress_in": "Progress"
+"node.transitionfx.input.color_in": "Color"
+"node.transitionfx.input.direction_in": "Direction"
+"node.transitionfx.input.type_in": "Type"
+"node.cornerpin.name": "Corner Pin"
+"node.cornerpin.input.tex_in": "Texture"
+"node.cornerpin.input.perspective_in": "Perspective"
+"node.cornerpin.input.top_left_in": "Top Left"
+"node.cornerpin.input.top_right_in": "Top Right"
+"node.cornerpin.input.bottom_right_in": "Bottom Right"
+"node.cornerpin.input.bottom_left_in": "Bottom Left"
+"node.displaytransform.name": "Display Transform"
+"node.displaytransform.input.tex_in": "Input"
+"node.displaytransform.input.display_in": "Display"
+"node.displaytransform.input.view_in": "View"
+"node.displaytransform.input.dir_in": "Direction"
+"node.ociogradingtransformlinear.name": "OCIO Color Grading (Linear)"
+"node.ociogradingtransformlinear.input.tex_in": "Input"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "Contrast"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "Offset"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "Exposure"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "Saturation"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "Pivot"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "Black Clamp"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "Enable White Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO Color Grading (Log)"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "Input"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "Lift"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "Gain"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "Gamma"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "Saturation"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "Pivot"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "Enable White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "White Clamp"
+"node.whitebalance.name": "White Balance"
+"node.whitebalance.input.tex_in": "Input"
+"node.whitebalance.input.temperature_in": "Temperature (K)"
+"node.whitebalance.input.tint_in": "Tint"
+"node.ociolut.name": "OCIO LUT"
+"node.ociolut.input.tex_in": "Input"
+"node.ociolut.input.lut_file_in": "LUT File"
+"node.ociolut.input.lut_dir_in": "Direction"
+"node.threewaycolor.name": "Three-Way Color"
+"node.threewaycolor.input.tex_in": "Input"
+"node.threewaycolor.input.shadows_color_in": "Shadows"
+"node.threewaycolor.input.midtones_color_in": "Midtones"
+"node.threewaycolor.input.highlights_color_in": "Highlights"
+"node.threewaycolor.input.shadows_amount_in": "Shadows Amount"
+"node.threewaycolor.input.midtones_amount_in": "Midtones Amount"
+"node.threewaycolor.input.highlights_amount_in": "Highlights Amount"
+"node.chromakey.name": "Chroma Key"
+"node.chromakey.input.tex_in": "Input"
+"node.chromakey.input.color_key": "Key Color"
+"node.chromakey.input.lower_tolerance_in": "Lower Tolerance"
+"node.chromakey.input.upper_tolerance_in": "Upper Tolerance"
+"node.chromakey.input.garbage_in": "Garbage Matte"
+"node.chromakey.input.core_in": "Core Matte"
+"node.chromakey.input.highlights_in": "Highlights"
+"node.chromakey.input.shadows_in": "Shadows"
+"node.chromakey.input.invert_in": "Invert Mask"
+"node.chromakey.input.mask_only_in": "Show Mask Only"
+"node.mask.name": "Mask"
+"node.mask.input.base_in": "Texture"
+"node.mask.input.points_in": "Points"
+"node.mask.input.color_in": "Color"
+"node.mask.input.invert_in": "Invert"
+"node.mask.input.feather_in": "Feather"
+"node.dropshadow.name": "Drop Shadow"
+"node.dropshadow.input.tex_in": "Texture"
+"node.dropshadow.input.color_in": "Color"
+"node.dropshadow.input.distance_in": "Distance"
+"node.dropshadow.input.angle_in": "Angle"
+"node.dropshadow.input.radius_in": "Softness"
+"node.dropshadow.input.opacity_in": "Opacity"
+"node.dropshadow.input.fast_in": "Faster (Lower Quality)"
+"node.timeformat.name": "Time Format"
+"node.timeformat.input.time_in": "Time"
+"node.timeformat.input.format_in": "Format"
+"node.timeformat.input.localtime_in": "Interpret time as local time"
+"node.wave.name": "Wave"
+"node.wave.input.tex_in": "Input"
+"node.wave.input.frequency_in": "Frequency"
+"node.wave.input.intensity_in": "Intensity"
+"node.wave.input.evolution_in": "Evolution"
+"node.wave.input.vertical_in": "Direction"
+"node.tile.name": "Tile"
+"node.tile.input.tex_in": "Input"
+"node.tile.input.scale_in": "Scale"
+"node.tile.input.position_in": "Position"
+"node.tile.input.anchor_in": "Anchor"
+"node.tile.input.mirrorx_in": "Mirror Horizontally"
+"node.tile.input.mirrory_in": "Mirror Vertically"
+"node.swirl.name": "Swirl"
+"node.swirl.input.tex_in": "Input"
+"node.swirl.input.radius_in": "Radius"
+"node.swirl.input.angle_in": "Angle"
+"node.swirl.input.pos_in": "Position"
+"node.ripple.name": "Ripple"
+"node.ripple.input.tex_in": "Input"
+"node.ripple.input.evolution_in": "Evolution"
+"node.ripple.input.intensity_in": "Intensity"
+"node.ripple.input.frequency_in": "Frequency"
+"node.ripple.input.position_in": "Position"
+"node.ripple.input.stretch_in": "Stretch"
+"node.colormatrix.name": "Color Matrix"
+"node.colormatrix.input.tex_in": "Input"
+"node.colormatrix.input.m0": "R <- R"
+"node.colormatrix.input.m1": "R <- G"
+"node.colormatrix.input.m2": "R <- B"
+"node.colormatrix.input.m3": "R <- A"
+"node.colormatrix.input.m4": "G <- R"
+"node.colormatrix.input.m5": "G <- G"
+"node.colormatrix.input.m6": "G <- B"
+"node.colormatrix.input.m7": "G <- A"
+"node.colormatrix.input.m8": "B <- R"
+"node.colormatrix.input.m9": "B <- G"
+"node.colormatrix.input.m10": "B <- B"
+"node.colormatrix.input.m11": "B <- A"
+"node.colormatrix.input.m12": "A <- R"
+"node.colormatrix.input.m13": "A <- G"
+"node.colormatrix.input.m14": "A <- B"
+"node.colormatrix.input.m15": "A <- A"
+"node.dilate.name": "Dilate"
+"node.dilate.input.tex_in": "Input"
+"node.dilate.input.radius_in": "Radius"
+"node.edgedetect.name": "Edge Detect"
+"node.edgedetect.input.tex_in": "Input"
+"node.edgedetect.input.threshold_in": "Threshold"
+"node.erode.name": "Erode"
+"node.erode.input.tex_in": "Input"
+"node.erode.input.radius_in": "Radius"
+"node.colorcorrect.name": "Color Correct"
+"node.colorcorrect.input.tex_in": "Input"
+"node.colorcorrect.input.saturation_in": "Saturation"
+"node.colorcorrect.input.contrast_in": "Contrast"
+"node.colorcorrect.input.gamma_in": "Gamma"
+"node.colorcorrect.input.gain_in": "Gain"
+"node.colorcorrect.input.offset_in": "Offset"
+"node.gamma.name": "Gamma"
+"node.gamma.input.tex_in": "Input"
+"node.gamma.input.gamma_in": "Gamma"
+"node.saturation.name": "Saturation"
+"node.saturation.input.tex_in": "Input"
+"node.saturation.input.saturation_in": "Saturation"
+"node.invert.name": "Invert"
+"node.invert.input.tex_in": "Input"
+"node.invert.input.invert_r_in": "Invert Red"
+"node.invert.input.invert_g_in": "Invert Green"
+"node.invert.input.invert_b_in": "Invert Blue"
+"node.invert.input.invert_a_in": "Invert Alpha"
+"node.clamp.name": "Clamp"
+"node.clamp.input.tex_in": "Input"
+"node.clamp.input.min_in": "Min"
+"node.clamp.input.max_in": "Max"
+"node.grade.name": "Grade"
+"node.grade.input.tex_in": "Input"
+"node.grade.input.blackpoint_in": "Black Point"
+"node.grade.input.whitepoint_in": "White Point"
+"node.grade.input.black_in": "Black Output"
+"node.grade.input.white_in": "White Output"
+"node.grade.input.gamma_in": "Gamma"
+"node.dirblur.name": "Directional Blur"
+"node.dirblur.input.tex_in": "Input"
+"node.dirblur.input.amount_in": "Amount"
+"node.dirblur.input.angle_in": "Angle"
+"node.sharpen.name": "Sharpen"
+"node.sharpen.input.tex_in": "Input"
+"node.sharpen.input.amount_in": "Amount"
+"node.dissolve.name": "Dissolve"
+"node.dissolve.input.tex_in": "Input"
+"node.dissolve.input.blend_in": "Blend"
+"node.dissolve.input.mix_in": "Mix"
+"node.keymix.name": "KeyMix"
+"node.keymix.input.tex_in": "Input"
+"node.keymix.input.blend_in": "Blend"
+"node.keymix.input.mask_in": "Mask"
+"node.premult.name": "Premultiply"
+"node.premult.input.tex_in": "Input"
+"node.premult.input.premult_channel_in": "Channel"
+"node.unpremult.name": "Unpremultiply"
+"node.unpremult.input.tex_in": "Input"
+"node.unpremult.input.unpremult_channel_in": "Channel"
+"node.position.name": "Position"
+"node.position.input.tex_in": "Input"
+"node.position.input.offset_in": "Translate"
+"node.mirror.name": "Mirror"
+"node.mirror.input.tex_in": "Input"
+"node.mirror.input.horizontal_in": "Horizontal (flop)"
+"node.mirror.input.vertical_in": "Vertical (flip)"
+"node.checkerboard.name": "Checkerboard"
+"node.checkerboard.input.base_in": "Base"
+"node.checkerboard.input.size_in": "Box Size"
+"node.checkerboard.input.color1_in": "Color 1"
+"node.checkerboard.input.color2_in": "Color 2"
+"node.colorbars.name": "Color Bars"
+"node.colorbars.input.base_in": "Base"
+"node.colorbars.input.standard_in": "Standard"
+"node.ramp.name": "Ramp"
+"node.ramp.input.base_in": "Base"
+"node.ramp.input.point0_in": "Point 0"
+"node.ramp.input.point1_in": "Point 1"
+"node.ramp.input.color0_in": "Color 0"
+"node.ramp.input.color1_in": "Color 1"
+"node.multicam.name": "Multi-Cam"
+"node.multicam.input.current_in": "Current"
+"node.multicam.input.sources_in": "Sources"
+"node.multicam.input.sequence_in": "Sequence"
+"node.multicam.input.sequence_type_in": "Sequence Type"
 "status.ready": "Bereit"
 "status.cache": "Cache: Aktiviert"
 "status.proxy.on": "Proxy: Ein"
@@ -481,6 +855,8 @@
 "node.category.group": "Gruppe"
 "project.context.new": "Neu"
 "project.new_sequence": "Neue Sequenz"
+"project.add_text_footage": "Add Text Footage"
+"timeline.context.add_adjustment_layer": "Add Adjustment Layer"
 "project.context.reveal_in_finder": "Im Finder anzeigen"
 "project.context.replace_footage": "Material ersetzen"
 "project.context.rename": "Umbenennen"
diff --git a/assets/i18n/en-US.yaml b/assets/i18n/en-US.yaml
index 206d1edfe..e5387a059 100644
--- a/assets/i18n/en-US.yaml
+++ b/assets/i18n/en-US.yaml
@@ -169,6 +169,380 @@
 "multicam.switch_9": "Switch to Camera 9"
 "effect_library.hint": "Double-click to add to the selected clip"
 "effect_library.group.builtin": "Built-in"
+"effect_library.group.color": "Color"
+"effect_library.group.filter": "Filter"
+"effect_library.group.distort": "Distort"
+"effect_library.group.keying": "Keying"
+"effect_library.group.generator": "Generator"
+"effect_library.group.math": "Math"
+"effect_library.group.time": "Time"
+"effect_library.group.channel": "Channel"
+"effect_library.group.transition": "Transition"
+"effect_library.group.general": "General"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "Polygon"
+"node.polygon.input.base_in": "Base"
+"node.polygon.input.points_in": "Points"
+"node.polygon.input.color_in": "Color"
+"node.ortho.name": "Orthographic Matrix"
+"node.ortho.input.pos_in": "Position"
+"node.ortho.input.rot_in": "Rotation"
+"node.ortho.input.scale_in": "Scale"
+"node.ortho.input.uniform_scale_in": "Uniform Scale"
+"node.ortho.input.anchor_in": "Anchor Point"
+"node.transform.name": "Transform"
+"node.transform.input.tex_in": "Texture"
+"node.transform.input.parent_in": "Parent"
+"node.transform.input.autoscale_in": "Auto-Scale"
+"node.transform.input.interpolation_in": "Interpolation"
+"node.transform.input.pos_in": "Position"
+"node.transform.input.rot_in": "Rotation"
+"node.transform.input.scale_in": "Scale"
+"node.transform.input.uniform_scale_in": "Uniform Scale"
+"node.transform.input.anchor_in": "Anchor Point"
+"node.volume.name": "Volume"
+"node.volume.input.samples_in": "Samples"
+"node.volume.input.volume_in": "Volume"
+"node.pan.name": "Pan"
+"node.pan.input.samples_in": "Samples"
+"node.pan.input.panning_in": "Pan"
+"node.math.name": "Math"
+"node.math.input.method_in": "Method"
+"node.math.input.param_a_in": "Value"
+"node.math.input.param_b_in": "Value"
+"node.time.name": "Time"
+"node.time.input.enabled_in": "Enabled"
+"node.trigonometry.name": "Trigonometry"
+"node.trigonometry.input.method_in": "Method"
+"node.trigonometry.input.x_in": "Value"
+"node.blur.name": "Blur"
+"node.blur.input.tex_in": "Input"
+"node.blur.input.method_in": "Method"
+"node.blur.input.radius_in": "Radius"
+"node.blur.input.horiz_in": "Horizontal"
+"node.blur.input.vert_in": "Vertical"
+"node.blur.input.directional_degrees_in": "Direction"
+"node.blur.input.radial_center_in": "Center"
+"node.blur.input.repeat_edge_pixels_in": "Repeat Edge Pixels"
+"node.solidgenerator.name": "Solid"
+"node.solidgenerator.input.color_in": "Color"
+"node.merge.name": "Merge"
+"node.merge.input.base_in": "Base"
+"node.merge.input.blend_in": "Blend"
+"node.stroke.name": "Stroke"
+"node.stroke.input.tex_in": "Input"
+"node.stroke.input.color_in": "Color"
+"node.stroke.input.radius_in": "Radius"
+"node.stroke.input.opacity_in": "Opacity"
+"node.stroke.input.inner_in": "Inner"
+"node.textgenerator.name": "Text (Legacy)"
+"node.textgenerator.input.text_in": "Text"
+"node.textgenerator.input.html_in": "Enable HTML"
+"node.textgenerator.input.color_in": "Color"
+"node.textgenerator.input.valign_in": "Vertical Align"
+"node.textgenerator.input.font_in": "Font"
+"node.textgenerator.input.font_size_in": "Font Size"
+"node.text2.name": "Text (Legacy)"
+"node.text2.input.base_in": "Base"
+"node.text2.input.pos_in": "Position"
+"node.text2.input.size_in": "Size"
+"node.text2.input.color_in": "Color"
+"node.text2.input.text_in": "Text"
+"node.text2.input.html_in": "Enable HTML"
+"node.text2.input.valign_in": "Vertical Align"
+"node.text2.input.font_in": "Font"
+"node.text2.input.font_size_in": "Font Size"
+"node.text3.name": "Text"
+"node.text3.input.base_in": "Base"
+"node.text3.input.pos_in": "Position"
+"node.text3.input.size_in": "Size"
+"node.text3.input.plain_text_in": "Text"
+"node.text3.input.text_in": "Text"
+"node.text3.input.font_family_in": "Font Family"
+"node.text3.input.font_size_in": "Font Size"
+"node.text3.input.outline_enabled_in": "Outline"
+"node.text3.input.outline_color_in": "Outline Color"
+"node.text3.input.outline_width_in": "Outline Width"
+"node.text3.input.glow_enabled_in": "Glow"
+"node.text3.input.glow_color_in": "Glow Color"
+"node.text3.input.glow_radius_in": "Glow Radius"
+"node.text3.input.color_in": "Color"
+"node.text3.input.valign_in": "Vertical Alignment"
+"node.text3.input.args_in": "Arguments"
+"node.mosaicfilter.name": "Mosaic"
+"node.mosaicfilter.input.tex_in": "Texture"
+"node.mosaicfilter.input.horiz_in": "Horizontal"
+"node.mosaicfilter.input.vert_in": "Vertical"
+"node.crop.name": "Crop"
+"node.crop.input.tex_in": "Texture"
+"node.crop.input.left_in": "Left"
+"node.crop.input.top_in": "Top"
+"node.crop.input.right_in": "Right"
+"node.crop.input.bottom_in": "Bottom"
+"node.crop.input.feather_in": "Feather"
+"node.value.name": "Value"
+"node.value.input.type_in": "Type"
+"node.value.input.value_in": "Value"
+"node.timeremap.name": "Time Remap"
+"node.timeremap.input.time_in": "Time"
+"node.timeremap.input.input_in": "Input"
+"node.shape.name": "Shape"
+"node.shape.input.type_in": "Type"
+"node.shape.input.base_in": "Base"
+"node.shape.input.pos_in": "Position"
+"node.shape.input.size_in": "Size"
+"node.shape.input.color_in": "Color"
+"node.shape.input.radius_in": "Radius"
+"node.colordifferencekey.name": "Color Difference Key"
+"node.colordifferencekey.input.tex_in": "Input"
+"node.colordifferencekey.input.garbage_in": "Garbage Matte"
+"node.colordifferencekey.input.core_in": "Core Matte"
+"node.colordifferencekey.input.color_in": "Key Color"
+"node.colordifferencekey.input.highlights_in": "Highlights"
+"node.colordifferencekey.input.shadows_in": "Shadows"
+"node.colordifferencekey.input.mask_only_in": "Show Mask Only"
+"node.despill.name": "Despill"
+"node.despill.input.tex_in": "Input"
+"node.despill.input.color_in": "Key Color"
+"node.despill.input.method_in": "Method"
+"node.despill.input.preserve_luminance_input": "Preserve Luminance"
+"node.group.name": "Group"
+"node.opacity.name": "Opacity"
+"node.opacity.input.tex_in": "Texture"
+"node.opacity.input.opacity_in": "Opacity"
+"node.flip.name": "Flip"
+"node.flip.input.tex_in": "Input"
+"node.flip.input.horiz_in": "Horizontal"
+"node.flip.input.vert_in": "Vertical"
+"node.noise.name": "Noise"
+"node.noise.input.base_in": "Base"
+"node.noise.input.strength_in": "Strength"
+"node.noise.input.color_in": "Color"
+"node.timeoffset.name": "Time Offset"
+"node.timeoffset.input.time_in": "Time"
+"node.timeoffset.input.input_in": "Input"
+"node.transition.name": "Transition"
+"node.transition.input.tex_in": "From"
+"node.transition.input.blend_in": "To"
+"node.transition.input.progress_in": "Progress"
+"node.transition.input.type_in": "Type"
+"node.transitionfx.name": "Transition FX"
+"node.transitionfx.input.tex_in": "From"
+"node.transitionfx.input.blend_in": "To"
+"node.transitionfx.input.progress_in": "Progress"
+"node.transitionfx.input.color_in": "Color"
+"node.transitionfx.input.direction_in": "Direction"
+"node.transitionfx.input.type_in": "Type"
+"node.cornerpin.name": "Corner Pin"
+"node.cornerpin.input.tex_in": "Texture"
+"node.cornerpin.input.perspective_in": "Perspective"
+"node.cornerpin.input.top_left_in": "Top Left"
+"node.cornerpin.input.top_right_in": "Top Right"
+"node.cornerpin.input.bottom_right_in": "Bottom Right"
+"node.cornerpin.input.bottom_left_in": "Bottom Left"
+"node.displaytransform.name": "Display Transform"
+"node.displaytransform.input.tex_in": "Input"
+"node.displaytransform.input.display_in": "Display"
+"node.displaytransform.input.view_in": "View"
+"node.displaytransform.input.dir_in": "Direction"
+"node.ociogradingtransformlinear.name": "OCIO Color Grading (Linear)"
+"node.ociogradingtransformlinear.input.tex_in": "Input"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "Contrast"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "Offset"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "Exposure"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "Saturation"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "Pivot"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "Black Clamp"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "Enable White Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO Color Grading (Log)"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "Input"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "Lift"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "Gain"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "Gamma"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "Saturation"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "Pivot"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "Enable White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "White Clamp"
+"node.whitebalance.name": "White Balance"
+"node.whitebalance.input.tex_in": "Input"
+"node.whitebalance.input.temperature_in": "Temperature (K)"
+"node.whitebalance.input.tint_in": "Tint"
+"node.ociolut.name": "OCIO LUT"
+"node.ociolut.input.tex_in": "Input"
+"node.ociolut.input.lut_file_in": "LUT File"
+"node.ociolut.input.lut_dir_in": "Direction"
+"node.threewaycolor.name": "Three-Way Color"
+"node.threewaycolor.input.tex_in": "Input"
+"node.threewaycolor.input.shadows_color_in": "Shadows"
+"node.threewaycolor.input.midtones_color_in": "Midtones"
+"node.threewaycolor.input.highlights_color_in": "Highlights"
+"node.threewaycolor.input.shadows_amount_in": "Shadows Amount"
+"node.threewaycolor.input.midtones_amount_in": "Midtones Amount"
+"node.threewaycolor.input.highlights_amount_in": "Highlights Amount"
+"node.chromakey.name": "Chroma Key"
+"node.chromakey.input.tex_in": "Input"
+"node.chromakey.input.color_key": "Key Color"
+"node.chromakey.input.lower_tolerance_in": "Lower Tolerance"
+"node.chromakey.input.upper_tolerance_in": "Upper Tolerance"
+"node.chromakey.input.garbage_in": "Garbage Matte"
+"node.chromakey.input.core_in": "Core Matte"
+"node.chromakey.input.highlights_in": "Highlights"
+"node.chromakey.input.shadows_in": "Shadows"
+"node.chromakey.input.invert_in": "Invert Mask"
+"node.chromakey.input.mask_only_in": "Show Mask Only"
+"node.mask.name": "Mask"
+"node.mask.input.base_in": "Texture"
+"node.mask.input.points_in": "Points"
+"node.mask.input.color_in": "Color"
+"node.mask.input.invert_in": "Invert"
+"node.mask.input.feather_in": "Feather"
+"node.dropshadow.name": "Drop Shadow"
+"node.dropshadow.input.tex_in": "Texture"
+"node.dropshadow.input.color_in": "Color"
+"node.dropshadow.input.distance_in": "Distance"
+"node.dropshadow.input.angle_in": "Angle"
+"node.dropshadow.input.radius_in": "Softness"
+"node.dropshadow.input.opacity_in": "Opacity"
+"node.dropshadow.input.fast_in": "Faster (Lower Quality)"
+"node.timeformat.name": "Time Format"
+"node.timeformat.input.time_in": "Time"
+"node.timeformat.input.format_in": "Format"
+"node.timeformat.input.localtime_in": "Interpret time as local time"
+"node.wave.name": "Wave"
+"node.wave.input.tex_in": "Input"
+"node.wave.input.frequency_in": "Frequency"
+"node.wave.input.intensity_in": "Intensity"
+"node.wave.input.evolution_in": "Evolution"
+"node.wave.input.vertical_in": "Direction"
+"node.tile.name": "Tile"
+"node.tile.input.tex_in": "Input"
+"node.tile.input.scale_in": "Scale"
+"node.tile.input.position_in": "Position"
+"node.tile.input.anchor_in": "Anchor"
+"node.tile.input.mirrorx_in": "Mirror Horizontally"
+"node.tile.input.mirrory_in": "Mirror Vertically"
+"node.swirl.name": "Swirl"
+"node.swirl.input.tex_in": "Input"
+"node.swirl.input.radius_in": "Radius"
+"node.swirl.input.angle_in": "Angle"
+"node.swirl.input.pos_in": "Position"
+"node.ripple.name": "Ripple"
+"node.ripple.input.tex_in": "Input"
+"node.ripple.input.evolution_in": "Evolution"
+"node.ripple.input.intensity_in": "Intensity"
+"node.ripple.input.frequency_in": "Frequency"
+"node.ripple.input.position_in": "Position"
+"node.ripple.input.stretch_in": "Stretch"
+"node.colormatrix.name": "Color Matrix"
+"node.colormatrix.input.tex_in": "Input"
+"node.colormatrix.input.m0": "R <- R"
+"node.colormatrix.input.m1": "R <- G"
+"node.colormatrix.input.m2": "R <- B"
+"node.colormatrix.input.m3": "R <- A"
+"node.colormatrix.input.m4": "G <- R"
+"node.colormatrix.input.m5": "G <- G"
+"node.colormatrix.input.m6": "G <- B"
+"node.colormatrix.input.m7": "G <- A"
+"node.colormatrix.input.m8": "B <- R"
+"node.colormatrix.input.m9": "B <- G"
+"node.colormatrix.input.m10": "B <- B"
+"node.colormatrix.input.m11": "B <- A"
+"node.colormatrix.input.m12": "A <- R"
+"node.colormatrix.input.m13": "A <- G"
+"node.colormatrix.input.m14": "A <- B"
+"node.colormatrix.input.m15": "A <- A"
+"node.dilate.name": "Dilate"
+"node.dilate.input.tex_in": "Input"
+"node.dilate.input.radius_in": "Radius"
+"node.edgedetect.name": "Edge Detect"
+"node.edgedetect.input.tex_in": "Input"
+"node.edgedetect.input.threshold_in": "Threshold"
+"node.erode.name": "Erode"
+"node.erode.input.tex_in": "Input"
+"node.erode.input.radius_in": "Radius"
+"node.colorcorrect.name": "Color Correct"
+"node.colorcorrect.input.tex_in": "Input"
+"node.colorcorrect.input.saturation_in": "Saturation"
+"node.colorcorrect.input.contrast_in": "Contrast"
+"node.colorcorrect.input.gamma_in": "Gamma"
+"node.colorcorrect.input.gain_in": "Gain"
+"node.colorcorrect.input.offset_in": "Offset"
+"node.gamma.name": "Gamma"
+"node.gamma.input.tex_in": "Input"
+"node.gamma.input.gamma_in": "Gamma"
+"node.saturation.name": "Saturation"
+"node.saturation.input.tex_in": "Input"
+"node.saturation.input.saturation_in": "Saturation"
+"node.invert.name": "Invert"
+"node.invert.input.tex_in": "Input"
+"node.invert.input.invert_r_in": "Invert Red"
+"node.invert.input.invert_g_in": "Invert Green"
+"node.invert.input.invert_b_in": "Invert Blue"
+"node.invert.input.invert_a_in": "Invert Alpha"
+"node.clamp.name": "Clamp"
+"node.clamp.input.tex_in": "Input"
+"node.clamp.input.min_in": "Min"
+"node.clamp.input.max_in": "Max"
+"node.grade.name": "Grade"
+"node.grade.input.tex_in": "Input"
+"node.grade.input.blackpoint_in": "Black Point"
+"node.grade.input.whitepoint_in": "White Point"
+"node.grade.input.black_in": "Black Output"
+"node.grade.input.white_in": "White Output"
+"node.grade.input.gamma_in": "Gamma"
+"node.dirblur.name": "Directional Blur"
+"node.dirblur.input.tex_in": "Input"
+"node.dirblur.input.amount_in": "Amount"
+"node.dirblur.input.angle_in": "Angle"
+"node.sharpen.name": "Sharpen"
+"node.sharpen.input.tex_in": "Input"
+"node.sharpen.input.amount_in": "Amount"
+"node.dissolve.name": "Dissolve"
+"node.dissolve.input.tex_in": "Input"
+"node.dissolve.input.blend_in": "Blend"
+"node.dissolve.input.mix_in": "Mix"
+"node.keymix.name": "KeyMix"
+"node.keymix.input.tex_in": "Input"
+"node.keymix.input.blend_in": "Blend"
+"node.keymix.input.mask_in": "Mask"
+"node.premult.name": "Premultiply"
+"node.premult.input.tex_in": "Input"
+"node.premult.input.premult_channel_in": "Channel"
+"node.unpremult.name": "Unpremultiply"
+"node.unpremult.input.tex_in": "Input"
+"node.unpremult.input.unpremult_channel_in": "Channel"
+"node.position.name": "Position"
+"node.position.input.tex_in": "Input"
+"node.position.input.offset_in": "Translate"
+"node.mirror.name": "Mirror"
+"node.mirror.input.tex_in": "Input"
+"node.mirror.input.horizontal_in": "Horizontal (flop)"
+"node.mirror.input.vertical_in": "Vertical (flip)"
+"node.checkerboard.name": "Checkerboard"
+"node.checkerboard.input.base_in": "Base"
+"node.checkerboard.input.size_in": "Box Size"
+"node.checkerboard.input.color1_in": "Color 1"
+"node.checkerboard.input.color2_in": "Color 2"
+"node.colorbars.name": "Color Bars"
+"node.colorbars.input.base_in": "Base"
+"node.colorbars.input.standard_in": "Standard"
+"node.ramp.name": "Ramp"
+"node.ramp.input.base_in": "Base"
+"node.ramp.input.point0_in": "Point 0"
+"node.ramp.input.point1_in": "Point 1"
+"node.ramp.input.color0_in": "Color 0"
+"node.ramp.input.color1_in": "Color 1"
+"node.multicam.name": "Multi-Cam"
+"node.multicam.input.current_in": "Current"
+"node.multicam.input.sources_in": "Sources"
+"node.multicam.input.sequence_in": "Sequence"
+"node.multicam.input.sequence_type_in": "Sequence Type"
 "status.ready": "Ready"
 "status.cache": "Cache: Enabled"
 "status.proxy.on": "Proxy: On"
@@ -479,6 +853,8 @@
 "node.category.group": "Group"
 "project.context.new": "New"
 "project.new_sequence": "New Sequence"
+"project.add_text_footage": "Add Text Footage"
+"timeline.context.add_adjustment_layer": "Add Adjustment Layer"
 "project.context.reveal_in_finder": "Reveal in Finder"
 "project.context.replace_footage": "Replace Footage"
 "project.context.rename": "Rename"
diff --git a/assets/i18n/es.yaml b/assets/i18n/es.yaml
index fb824f7aa..41188c622 100644
--- a/assets/i18n/es.yaml
+++ b/assets/i18n/es.yaml
@@ -171,6 +171,380 @@
 "multicam.switch_9": "Cambiar a cámara 9"
 "effect_library.hint": "Haz doble clic para añadir al clip seleccionado"
 "effect_library.group.builtin": "Integradas"
+"effect_library.group.color": "Color"
+"effect_library.group.filter": "Filter"
+"effect_library.group.distort": "Distort"
+"effect_library.group.keying": "Keying"
+"effect_library.group.generator": "Generator"
+"effect_library.group.math": "Math"
+"effect_library.group.time": "Time"
+"effect_library.group.channel": "Channel"
+"effect_library.group.transition": "Transition"
+"effect_library.group.general": "General"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "Polygon"
+"node.polygon.input.base_in": "Base"
+"node.polygon.input.points_in": "Points"
+"node.polygon.input.color_in": "Color"
+"node.ortho.name": "Orthographic Matrix"
+"node.ortho.input.pos_in": "Position"
+"node.ortho.input.rot_in": "Rotation"
+"node.ortho.input.scale_in": "Scale"
+"node.ortho.input.uniform_scale_in": "Uniform Scale"
+"node.ortho.input.anchor_in": "Anchor Point"
+"node.transform.name": "Transform"
+"node.transform.input.tex_in": "Texture"
+"node.transform.input.parent_in": "Parent"
+"node.transform.input.autoscale_in": "Auto-Scale"
+"node.transform.input.interpolation_in": "Interpolation"
+"node.transform.input.pos_in": "Position"
+"node.transform.input.rot_in": "Rotation"
+"node.transform.input.scale_in": "Scale"
+"node.transform.input.uniform_scale_in": "Uniform Scale"
+"node.transform.input.anchor_in": "Anchor Point"
+"node.volume.name": "Volume"
+"node.volume.input.samples_in": "Samples"
+"node.volume.input.volume_in": "Volume"
+"node.pan.name": "Pan"
+"node.pan.input.samples_in": "Samples"
+"node.pan.input.panning_in": "Pan"
+"node.math.name": "Math"
+"node.math.input.method_in": "Method"
+"node.math.input.param_a_in": "Value"
+"node.math.input.param_b_in": "Value"
+"node.time.name": "Time"
+"node.time.input.enabled_in": "Enabled"
+"node.trigonometry.name": "Trigonometry"
+"node.trigonometry.input.method_in": "Method"
+"node.trigonometry.input.x_in": "Value"
+"node.blur.name": "Blur"
+"node.blur.input.tex_in": "Input"
+"node.blur.input.method_in": "Method"
+"node.blur.input.radius_in": "Radius"
+"node.blur.input.horiz_in": "Horizontal"
+"node.blur.input.vert_in": "Vertical"
+"node.blur.input.directional_degrees_in": "Direction"
+"node.blur.input.radial_center_in": "Center"
+"node.blur.input.repeat_edge_pixels_in": "Repeat Edge Pixels"
+"node.solidgenerator.name": "Solid"
+"node.solidgenerator.input.color_in": "Color"
+"node.merge.name": "Merge"
+"node.merge.input.base_in": "Base"
+"node.merge.input.blend_in": "Blend"
+"node.stroke.name": "Stroke"
+"node.stroke.input.tex_in": "Input"
+"node.stroke.input.color_in": "Color"
+"node.stroke.input.radius_in": "Radius"
+"node.stroke.input.opacity_in": "Opacity"
+"node.stroke.input.inner_in": "Inner"
+"node.textgenerator.name": "Text (Legacy)"
+"node.textgenerator.input.text_in": "Text"
+"node.textgenerator.input.html_in": "Enable HTML"
+"node.textgenerator.input.color_in": "Color"
+"node.textgenerator.input.valign_in": "Vertical Align"
+"node.textgenerator.input.font_in": "Font"
+"node.textgenerator.input.font_size_in": "Font Size"
+"node.text2.name": "Text (Legacy)"
+"node.text2.input.base_in": "Base"
+"node.text2.input.pos_in": "Position"
+"node.text2.input.size_in": "Size"
+"node.text2.input.color_in": "Color"
+"node.text2.input.text_in": "Text"
+"node.text2.input.html_in": "Enable HTML"
+"node.text2.input.valign_in": "Vertical Align"
+"node.text2.input.font_in": "Font"
+"node.text2.input.font_size_in": "Font Size"
+"node.text3.name": "Text"
+"node.text3.input.base_in": "Base"
+"node.text3.input.pos_in": "Position"
+"node.text3.input.size_in": "Size"
+"node.text3.input.plain_text_in": "Text"
+"node.text3.input.text_in": "Text"
+"node.text3.input.font_family_in": "Font Family"
+"node.text3.input.font_size_in": "Font Size"
+"node.text3.input.outline_enabled_in": "Outline"
+"node.text3.input.outline_color_in": "Outline Color"
+"node.text3.input.outline_width_in": "Outline Width"
+"node.text3.input.glow_enabled_in": "Glow"
+"node.text3.input.glow_color_in": "Glow Color"
+"node.text3.input.glow_radius_in": "Glow Radius"
+"node.text3.input.color_in": "Color"
+"node.text3.input.valign_in": "Vertical Alignment"
+"node.text3.input.args_in": "Arguments"
+"node.mosaicfilter.name": "Mosaic"
+"node.mosaicfilter.input.tex_in": "Texture"
+"node.mosaicfilter.input.horiz_in": "Horizontal"
+"node.mosaicfilter.input.vert_in": "Vertical"
+"node.crop.name": "Crop"
+"node.crop.input.tex_in": "Texture"
+"node.crop.input.left_in": "Left"
+"node.crop.input.top_in": "Top"
+"node.crop.input.right_in": "Right"
+"node.crop.input.bottom_in": "Bottom"
+"node.crop.input.feather_in": "Feather"
+"node.value.name": "Value"
+"node.value.input.type_in": "Type"
+"node.value.input.value_in": "Value"
+"node.timeremap.name": "Time Remap"
+"node.timeremap.input.time_in": "Time"
+"node.timeremap.input.input_in": "Input"
+"node.shape.name": "Shape"
+"node.shape.input.type_in": "Type"
+"node.shape.input.base_in": "Base"
+"node.shape.input.pos_in": "Position"
+"node.shape.input.size_in": "Size"
+"node.shape.input.color_in": "Color"
+"node.shape.input.radius_in": "Radius"
+"node.colordifferencekey.name": "Color Difference Key"
+"node.colordifferencekey.input.tex_in": "Input"
+"node.colordifferencekey.input.garbage_in": "Garbage Matte"
+"node.colordifferencekey.input.core_in": "Core Matte"
+"node.colordifferencekey.input.color_in": "Key Color"
+"node.colordifferencekey.input.highlights_in": "Highlights"
+"node.colordifferencekey.input.shadows_in": "Shadows"
+"node.colordifferencekey.input.mask_only_in": "Show Mask Only"
+"node.despill.name": "Despill"
+"node.despill.input.tex_in": "Input"
+"node.despill.input.color_in": "Key Color"
+"node.despill.input.method_in": "Method"
+"node.despill.input.preserve_luminance_input": "Preserve Luminance"
+"node.group.name": "Group"
+"node.opacity.name": "Opacity"
+"node.opacity.input.tex_in": "Texture"
+"node.opacity.input.opacity_in": "Opacity"
+"node.flip.name": "Flip"
+"node.flip.input.tex_in": "Input"
+"node.flip.input.horiz_in": "Horizontal"
+"node.flip.input.vert_in": "Vertical"
+"node.noise.name": "Noise"
+"node.noise.input.base_in": "Base"
+"node.noise.input.strength_in": "Strength"
+"node.noise.input.color_in": "Color"
+"node.timeoffset.name": "Time Offset"
+"node.timeoffset.input.time_in": "Time"
+"node.timeoffset.input.input_in": "Input"
+"node.transition.name": "Transition"
+"node.transition.input.tex_in": "From"
+"node.transition.input.blend_in": "To"
+"node.transition.input.progress_in": "Progress"
+"node.transition.input.type_in": "Type"
+"node.transitionfx.name": "Transition FX"
+"node.transitionfx.input.tex_in": "From"
+"node.transitionfx.input.blend_in": "To"
+"node.transitionfx.input.progress_in": "Progress"
+"node.transitionfx.input.color_in": "Color"
+"node.transitionfx.input.direction_in": "Direction"
+"node.transitionfx.input.type_in": "Type"
+"node.cornerpin.name": "Corner Pin"
+"node.cornerpin.input.tex_in": "Texture"
+"node.cornerpin.input.perspective_in": "Perspective"
+"node.cornerpin.input.top_left_in": "Top Left"
+"node.cornerpin.input.top_right_in": "Top Right"
+"node.cornerpin.input.bottom_right_in": "Bottom Right"
+"node.cornerpin.input.bottom_left_in": "Bottom Left"
+"node.displaytransform.name": "Display Transform"
+"node.displaytransform.input.tex_in": "Input"
+"node.displaytransform.input.display_in": "Display"
+"node.displaytransform.input.view_in": "View"
+"node.displaytransform.input.dir_in": "Direction"
+"node.ociogradingtransformlinear.name": "OCIO Color Grading (Linear)"
+"node.ociogradingtransformlinear.input.tex_in": "Input"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "Contrast"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "Offset"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "Exposure"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "Saturation"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "Pivot"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "Black Clamp"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "Enable White Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO Color Grading (Log)"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "Input"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "Lift"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "Gain"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "Gamma"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "Saturation"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "Pivot"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "Enable White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "White Clamp"
+"node.whitebalance.name": "White Balance"
+"node.whitebalance.input.tex_in": "Input"
+"node.whitebalance.input.temperature_in": "Temperature (K)"
+"node.whitebalance.input.tint_in": "Tint"
+"node.ociolut.name": "OCIO LUT"
+"node.ociolut.input.tex_in": "Input"
+"node.ociolut.input.lut_file_in": "LUT File"
+"node.ociolut.input.lut_dir_in": "Direction"
+"node.threewaycolor.name": "Three-Way Color"
+"node.threewaycolor.input.tex_in": "Input"
+"node.threewaycolor.input.shadows_color_in": "Shadows"
+"node.threewaycolor.input.midtones_color_in": "Midtones"
+"node.threewaycolor.input.highlights_color_in": "Highlights"
+"node.threewaycolor.input.shadows_amount_in": "Shadows Amount"
+"node.threewaycolor.input.midtones_amount_in": "Midtones Amount"
+"node.threewaycolor.input.highlights_amount_in": "Highlights Amount"
+"node.chromakey.name": "Chroma Key"
+"node.chromakey.input.tex_in": "Input"
+"node.chromakey.input.color_key": "Key Color"
+"node.chromakey.input.lower_tolerance_in": "Lower Tolerance"
+"node.chromakey.input.upper_tolerance_in": "Upper Tolerance"
+"node.chromakey.input.garbage_in": "Garbage Matte"
+"node.chromakey.input.core_in": "Core Matte"
+"node.chromakey.input.highlights_in": "Highlights"
+"node.chromakey.input.shadows_in": "Shadows"
+"node.chromakey.input.invert_in": "Invert Mask"
+"node.chromakey.input.mask_only_in": "Show Mask Only"
+"node.mask.name": "Mask"
+"node.mask.input.base_in": "Texture"
+"node.mask.input.points_in": "Points"
+"node.mask.input.color_in": "Color"
+"node.mask.input.invert_in": "Invert"
+"node.mask.input.feather_in": "Feather"
+"node.dropshadow.name": "Drop Shadow"
+"node.dropshadow.input.tex_in": "Texture"
+"node.dropshadow.input.color_in": "Color"
+"node.dropshadow.input.distance_in": "Distance"
+"node.dropshadow.input.angle_in": "Angle"
+"node.dropshadow.input.radius_in": "Softness"
+"node.dropshadow.input.opacity_in": "Opacity"
+"node.dropshadow.input.fast_in": "Faster (Lower Quality)"
+"node.timeformat.name": "Time Format"
+"node.timeformat.input.time_in": "Time"
+"node.timeformat.input.format_in": "Format"
+"node.timeformat.input.localtime_in": "Interpret time as local time"
+"node.wave.name": "Wave"
+"node.wave.input.tex_in": "Input"
+"node.wave.input.frequency_in": "Frequency"
+"node.wave.input.intensity_in": "Intensity"
+"node.wave.input.evolution_in": "Evolution"
+"node.wave.input.vertical_in": "Direction"
+"node.tile.name": "Tile"
+"node.tile.input.tex_in": "Input"
+"node.tile.input.scale_in": "Scale"
+"node.tile.input.position_in": "Position"
+"node.tile.input.anchor_in": "Anchor"
+"node.tile.input.mirrorx_in": "Mirror Horizontally"
+"node.tile.input.mirrory_in": "Mirror Vertically"
+"node.swirl.name": "Swirl"
+"node.swirl.input.tex_in": "Input"
+"node.swirl.input.radius_in": "Radius"
+"node.swirl.input.angle_in": "Angle"
+"node.swirl.input.pos_in": "Position"
+"node.ripple.name": "Ripple"
+"node.ripple.input.tex_in": "Input"
+"node.ripple.input.evolution_in": "Evolution"
+"node.ripple.input.intensity_in": "Intensity"
+"node.ripple.input.frequency_in": "Frequency"
+"node.ripple.input.position_in": "Position"
+"node.ripple.input.stretch_in": "Stretch"
+"node.colormatrix.name": "Color Matrix"
+"node.colormatrix.input.tex_in": "Input"
+"node.colormatrix.input.m0": "R <- R"
+"node.colormatrix.input.m1": "R <- G"
+"node.colormatrix.input.m2": "R <- B"
+"node.colormatrix.input.m3": "R <- A"
+"node.colormatrix.input.m4": "G <- R"
+"node.colormatrix.input.m5": "G <- G"
+"node.colormatrix.input.m6": "G <- B"
+"node.colormatrix.input.m7": "G <- A"
+"node.colormatrix.input.m8": "B <- R"
+"node.colormatrix.input.m9": "B <- G"
+"node.colormatrix.input.m10": "B <- B"
+"node.colormatrix.input.m11": "B <- A"
+"node.colormatrix.input.m12": "A <- R"
+"node.colormatrix.input.m13": "A <- G"
+"node.colormatrix.input.m14": "A <- B"
+"node.colormatrix.input.m15": "A <- A"
+"node.dilate.name": "Dilate"
+"node.dilate.input.tex_in": "Input"
+"node.dilate.input.radius_in": "Radius"
+"node.edgedetect.name": "Edge Detect"
+"node.edgedetect.input.tex_in": "Input"
+"node.edgedetect.input.threshold_in": "Threshold"
+"node.erode.name": "Erode"
+"node.erode.input.tex_in": "Input"
+"node.erode.input.radius_in": "Radius"
+"node.colorcorrect.name": "Color Correct"
+"node.colorcorrect.input.tex_in": "Input"
+"node.colorcorrect.input.saturation_in": "Saturation"
+"node.colorcorrect.input.contrast_in": "Contrast"
+"node.colorcorrect.input.gamma_in": "Gamma"
+"node.colorcorrect.input.gain_in": "Gain"
+"node.colorcorrect.input.offset_in": "Offset"
+"node.gamma.name": "Gamma"
+"node.gamma.input.tex_in": "Input"
+"node.gamma.input.gamma_in": "Gamma"
+"node.saturation.name": "Saturation"
+"node.saturation.input.tex_in": "Input"
+"node.saturation.input.saturation_in": "Saturation"
+"node.invert.name": "Invert"
+"node.invert.input.tex_in": "Input"
+"node.invert.input.invert_r_in": "Invert Red"
+"node.invert.input.invert_g_in": "Invert Green"
+"node.invert.input.invert_b_in": "Invert Blue"
+"node.invert.input.invert_a_in": "Invert Alpha"
+"node.clamp.name": "Clamp"
+"node.clamp.input.tex_in": "Input"
+"node.clamp.input.min_in": "Min"
+"node.clamp.input.max_in": "Max"
+"node.grade.name": "Grade"
+"node.grade.input.tex_in": "Input"
+"node.grade.input.blackpoint_in": "Black Point"
+"node.grade.input.whitepoint_in": "White Point"
+"node.grade.input.black_in": "Black Output"
+"node.grade.input.white_in": "White Output"
+"node.grade.input.gamma_in": "Gamma"
+"node.dirblur.name": "Directional Blur"
+"node.dirblur.input.tex_in": "Input"
+"node.dirblur.input.amount_in": "Amount"
+"node.dirblur.input.angle_in": "Angle"
+"node.sharpen.name": "Sharpen"
+"node.sharpen.input.tex_in": "Input"
+"node.sharpen.input.amount_in": "Amount"
+"node.dissolve.name": "Dissolve"
+"node.dissolve.input.tex_in": "Input"
+"node.dissolve.input.blend_in": "Blend"
+"node.dissolve.input.mix_in": "Mix"
+"node.keymix.name": "KeyMix"
+"node.keymix.input.tex_in": "Input"
+"node.keymix.input.blend_in": "Blend"
+"node.keymix.input.mask_in": "Mask"
+"node.premult.name": "Premultiply"
+"node.premult.input.tex_in": "Input"
+"node.premult.input.premult_channel_in": "Channel"
+"node.unpremult.name": "Unpremultiply"
+"node.unpremult.input.tex_in": "Input"
+"node.unpremult.input.unpremult_channel_in": "Channel"
+"node.position.name": "Position"
+"node.position.input.tex_in": "Input"
+"node.position.input.offset_in": "Translate"
+"node.mirror.name": "Mirror"
+"node.mirror.input.tex_in": "Input"
+"node.mirror.input.horizontal_in": "Horizontal (flop)"
+"node.mirror.input.vertical_in": "Vertical (flip)"
+"node.checkerboard.name": "Checkerboard"
+"node.checkerboard.input.base_in": "Base"
+"node.checkerboard.input.size_in": "Box Size"
+"node.checkerboard.input.color1_in": "Color 1"
+"node.checkerboard.input.color2_in": "Color 2"
+"node.colorbars.name": "Color Bars"
+"node.colorbars.input.base_in": "Base"
+"node.colorbars.input.standard_in": "Standard"
+"node.ramp.name": "Ramp"
+"node.ramp.input.base_in": "Base"
+"node.ramp.input.point0_in": "Point 0"
+"node.ramp.input.point1_in": "Point 1"
+"node.ramp.input.color0_in": "Color 0"
+"node.ramp.input.color1_in": "Color 1"
+"node.multicam.name": "Multi-Cam"
+"node.multicam.input.current_in": "Current"
+"node.multicam.input.sources_in": "Sources"
+"node.multicam.input.sequence_in": "Sequence"
+"node.multicam.input.sequence_type_in": "Sequence Type"
 "status.ready": "Listo"
 "status.cache": "Caché: activada"
 "status.proxy.on": "Proxy: activado"
@@ -481,6 +855,8 @@
 "node.category.group": "Grupo"
 "project.context.new": "Nuevo"
 "project.new_sequence": "Nueva secuencia"
+"project.add_text_footage": "Add Text Footage"
+"timeline.context.add_adjustment_layer": "Add Adjustment Layer"
 "project.context.reveal_in_finder": "Mostrar en Finder"
 "project.context.replace_footage": "Reemplazar material"
 "project.context.rename": "Cambiar nombre"
diff --git a/assets/i18n/fr.yaml b/assets/i18n/fr.yaml
index f89f70ed3..6c7355a81 100644
--- a/assets/i18n/fr.yaml
+++ b/assets/i18n/fr.yaml
@@ -171,6 +171,380 @@
 "multicam.switch_9": "Passer à la caméra 9"
 "effect_library.hint": "Double-cliquer pour ajouter au clip sélectionné"
 "effect_library.group.builtin": "Intégré"
+"effect_library.group.color": "Color"
+"effect_library.group.filter": "Filter"
+"effect_library.group.distort": "Distort"
+"effect_library.group.keying": "Keying"
+"effect_library.group.generator": "Generator"
+"effect_library.group.math": "Math"
+"effect_library.group.time": "Time"
+"effect_library.group.channel": "Channel"
+"effect_library.group.transition": "Transition"
+"effect_library.group.general": "General"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "Polygon"
+"node.polygon.input.base_in": "Base"
+"node.polygon.input.points_in": "Points"
+"node.polygon.input.color_in": "Color"
+"node.ortho.name": "Orthographic Matrix"
+"node.ortho.input.pos_in": "Position"
+"node.ortho.input.rot_in": "Rotation"
+"node.ortho.input.scale_in": "Scale"
+"node.ortho.input.uniform_scale_in": "Uniform Scale"
+"node.ortho.input.anchor_in": "Anchor Point"
+"node.transform.name": "Transform"
+"node.transform.input.tex_in": "Texture"
+"node.transform.input.parent_in": "Parent"
+"node.transform.input.autoscale_in": "Auto-Scale"
+"node.transform.input.interpolation_in": "Interpolation"
+"node.transform.input.pos_in": "Position"
+"node.transform.input.rot_in": "Rotation"
+"node.transform.input.scale_in": "Scale"
+"node.transform.input.uniform_scale_in": "Uniform Scale"
+"node.transform.input.anchor_in": "Anchor Point"
+"node.volume.name": "Volume"
+"node.volume.input.samples_in": "Samples"
+"node.volume.input.volume_in": "Volume"
+"node.pan.name": "Pan"
+"node.pan.input.samples_in": "Samples"
+"node.pan.input.panning_in": "Pan"
+"node.math.name": "Math"
+"node.math.input.method_in": "Method"
+"node.math.input.param_a_in": "Value"
+"node.math.input.param_b_in": "Value"
+"node.time.name": "Time"
+"node.time.input.enabled_in": "Enabled"
+"node.trigonometry.name": "Trigonometry"
+"node.trigonometry.input.method_in": "Method"
+"node.trigonometry.input.x_in": "Value"
+"node.blur.name": "Blur"
+"node.blur.input.tex_in": "Input"
+"node.blur.input.method_in": "Method"
+"node.blur.input.radius_in": "Radius"
+"node.blur.input.horiz_in": "Horizontal"
+"node.blur.input.vert_in": "Vertical"
+"node.blur.input.directional_degrees_in": "Direction"
+"node.blur.input.radial_center_in": "Center"
+"node.blur.input.repeat_edge_pixels_in": "Repeat Edge Pixels"
+"node.solidgenerator.name": "Solid"
+"node.solidgenerator.input.color_in": "Color"
+"node.merge.name": "Merge"
+"node.merge.input.base_in": "Base"
+"node.merge.input.blend_in": "Blend"
+"node.stroke.name": "Stroke"
+"node.stroke.input.tex_in": "Input"
+"node.stroke.input.color_in": "Color"
+"node.stroke.input.radius_in": "Radius"
+"node.stroke.input.opacity_in": "Opacity"
+"node.stroke.input.inner_in": "Inner"
+"node.textgenerator.name": "Text (Legacy)"
+"node.textgenerator.input.text_in": "Text"
+"node.textgenerator.input.html_in": "Enable HTML"
+"node.textgenerator.input.color_in": "Color"
+"node.textgenerator.input.valign_in": "Vertical Align"
+"node.textgenerator.input.font_in": "Font"
+"node.textgenerator.input.font_size_in": "Font Size"
+"node.text2.name": "Text (Legacy)"
+"node.text2.input.base_in": "Base"
+"node.text2.input.pos_in": "Position"
+"node.text2.input.size_in": "Size"
+"node.text2.input.color_in": "Color"
+"node.text2.input.text_in": "Text"
+"node.text2.input.html_in": "Enable HTML"
+"node.text2.input.valign_in": "Vertical Align"
+"node.text2.input.font_in": "Font"
+"node.text2.input.font_size_in": "Font Size"
+"node.text3.name": "Text"
+"node.text3.input.base_in": "Base"
+"node.text3.input.pos_in": "Position"
+"node.text3.input.size_in": "Size"
+"node.text3.input.plain_text_in": "Text"
+"node.text3.input.text_in": "Text"
+"node.text3.input.font_family_in": "Font Family"
+"node.text3.input.font_size_in": "Font Size"
+"node.text3.input.outline_enabled_in": "Outline"
+"node.text3.input.outline_color_in": "Outline Color"
+"node.text3.input.outline_width_in": "Outline Width"
+"node.text3.input.glow_enabled_in": "Glow"
+"node.text3.input.glow_color_in": "Glow Color"
+"node.text3.input.glow_radius_in": "Glow Radius"
+"node.text3.input.color_in": "Color"
+"node.text3.input.valign_in": "Vertical Alignment"
+"node.text3.input.args_in": "Arguments"
+"node.mosaicfilter.name": "Mosaic"
+"node.mosaicfilter.input.tex_in": "Texture"
+"node.mosaicfilter.input.horiz_in": "Horizontal"
+"node.mosaicfilter.input.vert_in": "Vertical"
+"node.crop.name": "Crop"
+"node.crop.input.tex_in": "Texture"
+"node.crop.input.left_in": "Left"
+"node.crop.input.top_in": "Top"
+"node.crop.input.right_in": "Right"
+"node.crop.input.bottom_in": "Bottom"
+"node.crop.input.feather_in": "Feather"
+"node.value.name": "Value"
+"node.value.input.type_in": "Type"
+"node.value.input.value_in": "Value"
+"node.timeremap.name": "Time Remap"
+"node.timeremap.input.time_in": "Time"
+"node.timeremap.input.input_in": "Input"
+"node.shape.name": "Shape"
+"node.shape.input.type_in": "Type"
+"node.shape.input.base_in": "Base"
+"node.shape.input.pos_in": "Position"
+"node.shape.input.size_in": "Size"
+"node.shape.input.color_in": "Color"
+"node.shape.input.radius_in": "Radius"
+"node.colordifferencekey.name": "Color Difference Key"
+"node.colordifferencekey.input.tex_in": "Input"
+"node.colordifferencekey.input.garbage_in": "Garbage Matte"
+"node.colordifferencekey.input.core_in": "Core Matte"
+"node.colordifferencekey.input.color_in": "Key Color"
+"node.colordifferencekey.input.highlights_in": "Highlights"
+"node.colordifferencekey.input.shadows_in": "Shadows"
+"node.colordifferencekey.input.mask_only_in": "Show Mask Only"
+"node.despill.name": "Despill"
+"node.despill.input.tex_in": "Input"
+"node.despill.input.color_in": "Key Color"
+"node.despill.input.method_in": "Method"
+"node.despill.input.preserve_luminance_input": "Preserve Luminance"
+"node.group.name": "Group"
+"node.opacity.name": "Opacity"
+"node.opacity.input.tex_in": "Texture"
+"node.opacity.input.opacity_in": "Opacity"
+"node.flip.name": "Flip"
+"node.flip.input.tex_in": "Input"
+"node.flip.input.horiz_in": "Horizontal"
+"node.flip.input.vert_in": "Vertical"
+"node.noise.name": "Noise"
+"node.noise.input.base_in": "Base"
+"node.noise.input.strength_in": "Strength"
+"node.noise.input.color_in": "Color"
+"node.timeoffset.name": "Time Offset"
+"node.timeoffset.input.time_in": "Time"
+"node.timeoffset.input.input_in": "Input"
+"node.transition.name": "Transition"
+"node.transition.input.tex_in": "From"
+"node.transition.input.blend_in": "To"
+"node.transition.input.progress_in": "Progress"
+"node.transition.input.type_in": "Type"
+"node.transitionfx.name": "Transition FX"
+"node.transitionfx.input.tex_in": "From"
+"node.transitionfx.input.blend_in": "To"
+"node.transitionfx.input.progress_in": "Progress"
+"node.transitionfx.input.color_in": "Color"
+"node.transitionfx.input.direction_in": "Direction"
+"node.transitionfx.input.type_in": "Type"
+"node.cornerpin.name": "Corner Pin"
+"node.cornerpin.input.tex_in": "Texture"
+"node.cornerpin.input.perspective_in": "Perspective"
+"node.cornerpin.input.top_left_in": "Top Left"
+"node.cornerpin.input.top_right_in": "Top Right"
+"node.cornerpin.input.bottom_right_in": "Bottom Right"
+"node.cornerpin.input.bottom_left_in": "Bottom Left"
+"node.displaytransform.name": "Display Transform"
+"node.displaytransform.input.tex_in": "Input"
+"node.displaytransform.input.display_in": "Display"
+"node.displaytransform.input.view_in": "View"
+"node.displaytransform.input.dir_in": "Direction"
+"node.ociogradingtransformlinear.name": "OCIO Color Grading (Linear)"
+"node.ociogradingtransformlinear.input.tex_in": "Input"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "Contrast"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "Offset"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "Exposure"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "Saturation"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "Pivot"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "Black Clamp"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "Enable White Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO Color Grading (Log)"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "Input"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "Lift"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "Gain"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "Gamma"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "Saturation"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "Pivot"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "Enable White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "White Clamp"
+"node.whitebalance.name": "White Balance"
+"node.whitebalance.input.tex_in": "Input"
+"node.whitebalance.input.temperature_in": "Temperature (K)"
+"node.whitebalance.input.tint_in": "Tint"
+"node.ociolut.name": "OCIO LUT"
+"node.ociolut.input.tex_in": "Input"
+"node.ociolut.input.lut_file_in": "LUT File"
+"node.ociolut.input.lut_dir_in": "Direction"
+"node.threewaycolor.name": "Three-Way Color"
+"node.threewaycolor.input.tex_in": "Input"
+"node.threewaycolor.input.shadows_color_in": "Shadows"
+"node.threewaycolor.input.midtones_color_in": "Midtones"
+"node.threewaycolor.input.highlights_color_in": "Highlights"
+"node.threewaycolor.input.shadows_amount_in": "Shadows Amount"
+"node.threewaycolor.input.midtones_amount_in": "Midtones Amount"
+"node.threewaycolor.input.highlights_amount_in": "Highlights Amount"
+"node.chromakey.name": "Chroma Key"
+"node.chromakey.input.tex_in": "Input"
+"node.chromakey.input.color_key": "Key Color"
+"node.chromakey.input.lower_tolerance_in": "Lower Tolerance"
+"node.chromakey.input.upper_tolerance_in": "Upper Tolerance"
+"node.chromakey.input.garbage_in": "Garbage Matte"
+"node.chromakey.input.core_in": "Core Matte"
+"node.chromakey.input.highlights_in": "Highlights"
+"node.chromakey.input.shadows_in": "Shadows"
+"node.chromakey.input.invert_in": "Invert Mask"
+"node.chromakey.input.mask_only_in": "Show Mask Only"
+"node.mask.name": "Mask"
+"node.mask.input.base_in": "Texture"
+"node.mask.input.points_in": "Points"
+"node.mask.input.color_in": "Color"
+"node.mask.input.invert_in": "Invert"
+"node.mask.input.feather_in": "Feather"
+"node.dropshadow.name": "Drop Shadow"
+"node.dropshadow.input.tex_in": "Texture"
+"node.dropshadow.input.color_in": "Color"
+"node.dropshadow.input.distance_in": "Distance"
+"node.dropshadow.input.angle_in": "Angle"
+"node.dropshadow.input.radius_in": "Softness"
+"node.dropshadow.input.opacity_in": "Opacity"
+"node.dropshadow.input.fast_in": "Faster (Lower Quality)"
+"node.timeformat.name": "Time Format"
+"node.timeformat.input.time_in": "Time"
+"node.timeformat.input.format_in": "Format"
+"node.timeformat.input.localtime_in": "Interpret time as local time"
+"node.wave.name": "Wave"
+"node.wave.input.tex_in": "Input"
+"node.wave.input.frequency_in": "Frequency"
+"node.wave.input.intensity_in": "Intensity"
+"node.wave.input.evolution_in": "Evolution"
+"node.wave.input.vertical_in": "Direction"
+"node.tile.name": "Tile"
+"node.tile.input.tex_in": "Input"
+"node.tile.input.scale_in": "Scale"
+"node.tile.input.position_in": "Position"
+"node.tile.input.anchor_in": "Anchor"
+"node.tile.input.mirrorx_in": "Mirror Horizontally"
+"node.tile.input.mirrory_in": "Mirror Vertically"
+"node.swirl.name": "Swirl"
+"node.swirl.input.tex_in": "Input"
+"node.swirl.input.radius_in": "Radius"
+"node.swirl.input.angle_in": "Angle"
+"node.swirl.input.pos_in": "Position"
+"node.ripple.name": "Ripple"
+"node.ripple.input.tex_in": "Input"
+"node.ripple.input.evolution_in": "Evolution"
+"node.ripple.input.intensity_in": "Intensity"
+"node.ripple.input.frequency_in": "Frequency"
+"node.ripple.input.position_in": "Position"
+"node.ripple.input.stretch_in": "Stretch"
+"node.colormatrix.name": "Color Matrix"
+"node.colormatrix.input.tex_in": "Input"
+"node.colormatrix.input.m0": "R <- R"
+"node.colormatrix.input.m1": "R <- G"
+"node.colormatrix.input.m2": "R <- B"
+"node.colormatrix.input.m3": "R <- A"
+"node.colormatrix.input.m4": "G <- R"
+"node.colormatrix.input.m5": "G <- G"
+"node.colormatrix.input.m6": "G <- B"
+"node.colormatrix.input.m7": "G <- A"
+"node.colormatrix.input.m8": "B <- R"
+"node.colormatrix.input.m9": "B <- G"
+"node.colormatrix.input.m10": "B <- B"
+"node.colormatrix.input.m11": "B <- A"
+"node.colormatrix.input.m12": "A <- R"
+"node.colormatrix.input.m13": "A <- G"
+"node.colormatrix.input.m14": "A <- B"
+"node.colormatrix.input.m15": "A <- A"
+"node.dilate.name": "Dilate"
+"node.dilate.input.tex_in": "Input"
+"node.dilate.input.radius_in": "Radius"
+"node.edgedetect.name": "Edge Detect"
+"node.edgedetect.input.tex_in": "Input"
+"node.edgedetect.input.threshold_in": "Threshold"
+"node.erode.name": "Erode"
+"node.erode.input.tex_in": "Input"
+"node.erode.input.radius_in": "Radius"
+"node.colorcorrect.name": "Color Correct"
+"node.colorcorrect.input.tex_in": "Input"
+"node.colorcorrect.input.saturation_in": "Saturation"
+"node.colorcorrect.input.contrast_in": "Contrast"
+"node.colorcorrect.input.gamma_in": "Gamma"
+"node.colorcorrect.input.gain_in": "Gain"
+"node.colorcorrect.input.offset_in": "Offset"
+"node.gamma.name": "Gamma"
+"node.gamma.input.tex_in": "Input"
+"node.gamma.input.gamma_in": "Gamma"
+"node.saturation.name": "Saturation"
+"node.saturation.input.tex_in": "Input"
+"node.saturation.input.saturation_in": "Saturation"
+"node.invert.name": "Invert"
+"node.invert.input.tex_in": "Input"
+"node.invert.input.invert_r_in": "Invert Red"
+"node.invert.input.invert_g_in": "Invert Green"
+"node.invert.input.invert_b_in": "Invert Blue"
+"node.invert.input.invert_a_in": "Invert Alpha"
+"node.clamp.name": "Clamp"
+"node.clamp.input.tex_in": "Input"
+"node.clamp.input.min_in": "Min"
+"node.clamp.input.max_in": "Max"
+"node.grade.name": "Grade"
+"node.grade.input.tex_in": "Input"
+"node.grade.input.blackpoint_in": "Black Point"
+"node.grade.input.whitepoint_in": "White Point"
+"node.grade.input.black_in": "Black Output"
+"node.grade.input.white_in": "White Output"
+"node.grade.input.gamma_in": "Gamma"
+"node.dirblur.name": "Directional Blur"
+"node.dirblur.input.tex_in": "Input"
+"node.dirblur.input.amount_in": "Amount"
+"node.dirblur.input.angle_in": "Angle"
+"node.sharpen.name": "Sharpen"
+"node.sharpen.input.tex_in": "Input"
+"node.sharpen.input.amount_in": "Amount"
+"node.dissolve.name": "Dissolve"
+"node.dissolve.input.tex_in": "Input"
+"node.dissolve.input.blend_in": "Blend"
+"node.dissolve.input.mix_in": "Mix"
+"node.keymix.name": "KeyMix"
+"node.keymix.input.tex_in": "Input"
+"node.keymix.input.blend_in": "Blend"
+"node.keymix.input.mask_in": "Mask"
+"node.premult.name": "Premultiply"
+"node.premult.input.tex_in": "Input"
+"node.premult.input.premult_channel_in": "Channel"
+"node.unpremult.name": "Unpremultiply"
+"node.unpremult.input.tex_in": "Input"
+"node.unpremult.input.unpremult_channel_in": "Channel"
+"node.position.name": "Position"
+"node.position.input.tex_in": "Input"
+"node.position.input.offset_in": "Translate"
+"node.mirror.name": "Mirror"
+"node.mirror.input.tex_in": "Input"
+"node.mirror.input.horizontal_in": "Horizontal (flop)"
+"node.mirror.input.vertical_in": "Vertical (flip)"
+"node.checkerboard.name": "Checkerboard"
+"node.checkerboard.input.base_in": "Base"
+"node.checkerboard.input.size_in": "Box Size"
+"node.checkerboard.input.color1_in": "Color 1"
+"node.checkerboard.input.color2_in": "Color 2"
+"node.colorbars.name": "Color Bars"
+"node.colorbars.input.base_in": "Base"
+"node.colorbars.input.standard_in": "Standard"
+"node.ramp.name": "Ramp"
+"node.ramp.input.base_in": "Base"
+"node.ramp.input.point0_in": "Point 0"
+"node.ramp.input.point1_in": "Point 1"
+"node.ramp.input.color0_in": "Color 0"
+"node.ramp.input.color1_in": "Color 1"
+"node.multicam.name": "Multi-Cam"
+"node.multicam.input.current_in": "Current"
+"node.multicam.input.sources_in": "Sources"
+"node.multicam.input.sequence_in": "Sequence"
+"node.multicam.input.sequence_type_in": "Sequence Type"
 "status.ready": "Prêt"
 "status.cache": "Cache : activé"
 "status.proxy.on": "Proxy : activé"
@@ -481,6 +855,8 @@
 "node.category.group": "Groupe"
 "project.context.new": "Nouveau"
 "project.new_sequence": "Nouvelle séquence"
+"project.add_text_footage": "Add Text Footage"
+"timeline.context.add_adjustment_layer": "Add Adjustment Layer"
 "project.context.reveal_in_finder": "Révéler dans le Finder"
 "project.context.replace_footage": "Remplacer les médias"
 "project.context.rename": "Renommer"
diff --git a/assets/i18n/ja.yaml b/assets/i18n/ja.yaml
index 5567c6fc4..fd8fa5f7b 100644
--- a/assets/i18n/ja.yaml
+++ b/assets/i18n/ja.yaml
@@ -171,6 +171,380 @@
 "multicam.switch_9": "カメラ9に切り替え"
 "effect_library.hint": "ダブルクリックで選択中のクリップに追加します"
 "effect_library.group.builtin": "内蔵"
+"effect_library.group.color": "Color"
+"effect_library.group.filter": "Filter"
+"effect_library.group.distort": "Distort"
+"effect_library.group.keying": "Keying"
+"effect_library.group.generator": "Generator"
+"effect_library.group.math": "Math"
+"effect_library.group.time": "Time"
+"effect_library.group.channel": "Channel"
+"effect_library.group.transition": "Transition"
+"effect_library.group.general": "General"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "Polygon"
+"node.polygon.input.base_in": "Base"
+"node.polygon.input.points_in": "Points"
+"node.polygon.input.color_in": "Color"
+"node.ortho.name": "Orthographic Matrix"
+"node.ortho.input.pos_in": "Position"
+"node.ortho.input.rot_in": "Rotation"
+"node.ortho.input.scale_in": "Scale"
+"node.ortho.input.uniform_scale_in": "Uniform Scale"
+"node.ortho.input.anchor_in": "Anchor Point"
+"node.transform.name": "Transform"
+"node.transform.input.tex_in": "Texture"
+"node.transform.input.parent_in": "Parent"
+"node.transform.input.autoscale_in": "Auto-Scale"
+"node.transform.input.interpolation_in": "Interpolation"
+"node.transform.input.pos_in": "Position"
+"node.transform.input.rot_in": "Rotation"
+"node.transform.input.scale_in": "Scale"
+"node.transform.input.uniform_scale_in": "Uniform Scale"
+"node.transform.input.anchor_in": "Anchor Point"
+"node.volume.name": "Volume"
+"node.volume.input.samples_in": "Samples"
+"node.volume.input.volume_in": "Volume"
+"node.pan.name": "Pan"
+"node.pan.input.samples_in": "Samples"
+"node.pan.input.panning_in": "Pan"
+"node.math.name": "Math"
+"node.math.input.method_in": "Method"
+"node.math.input.param_a_in": "Value"
+"node.math.input.param_b_in": "Value"
+"node.time.name": "Time"
+"node.time.input.enabled_in": "Enabled"
+"node.trigonometry.name": "Trigonometry"
+"node.trigonometry.input.method_in": "Method"
+"node.trigonometry.input.x_in": "Value"
+"node.blur.name": "Blur"
+"node.blur.input.tex_in": "Input"
+"node.blur.input.method_in": "Method"
+"node.blur.input.radius_in": "Radius"
+"node.blur.input.horiz_in": "Horizontal"
+"node.blur.input.vert_in": "Vertical"
+"node.blur.input.directional_degrees_in": "Direction"
+"node.blur.input.radial_center_in": "Center"
+"node.blur.input.repeat_edge_pixels_in": "Repeat Edge Pixels"
+"node.solidgenerator.name": "Solid"
+"node.solidgenerator.input.color_in": "Color"
+"node.merge.name": "Merge"
+"node.merge.input.base_in": "Base"
+"node.merge.input.blend_in": "Blend"
+"node.stroke.name": "Stroke"
+"node.stroke.input.tex_in": "Input"
+"node.stroke.input.color_in": "Color"
+"node.stroke.input.radius_in": "Radius"
+"node.stroke.input.opacity_in": "Opacity"
+"node.stroke.input.inner_in": "Inner"
+"node.textgenerator.name": "Text (Legacy)"
+"node.textgenerator.input.text_in": "Text"
+"node.textgenerator.input.html_in": "Enable HTML"
+"node.textgenerator.input.color_in": "Color"
+"node.textgenerator.input.valign_in": "Vertical Align"
+"node.textgenerator.input.font_in": "Font"
+"node.textgenerator.input.font_size_in": "Font Size"
+"node.text2.name": "Text (Legacy)"
+"node.text2.input.base_in": "Base"
+"node.text2.input.pos_in": "Position"
+"node.text2.input.size_in": "Size"
+"node.text2.input.color_in": "Color"
+"node.text2.input.text_in": "Text"
+"node.text2.input.html_in": "Enable HTML"
+"node.text2.input.valign_in": "Vertical Align"
+"node.text2.input.font_in": "Font"
+"node.text2.input.font_size_in": "Font Size"
+"node.text3.name": "Text"
+"node.text3.input.base_in": "Base"
+"node.text3.input.pos_in": "Position"
+"node.text3.input.size_in": "Size"
+"node.text3.input.plain_text_in": "Text"
+"node.text3.input.text_in": "Text"
+"node.text3.input.font_family_in": "Font Family"
+"node.text3.input.font_size_in": "Font Size"
+"node.text3.input.outline_enabled_in": "Outline"
+"node.text3.input.outline_color_in": "Outline Color"
+"node.text3.input.outline_width_in": "Outline Width"
+"node.text3.input.glow_enabled_in": "Glow"
+"node.text3.input.glow_color_in": "Glow Color"
+"node.text3.input.glow_radius_in": "Glow Radius"
+"node.text3.input.color_in": "Color"
+"node.text3.input.valign_in": "Vertical Alignment"
+"node.text3.input.args_in": "Arguments"
+"node.mosaicfilter.name": "Mosaic"
+"node.mosaicfilter.input.tex_in": "Texture"
+"node.mosaicfilter.input.horiz_in": "Horizontal"
+"node.mosaicfilter.input.vert_in": "Vertical"
+"node.crop.name": "Crop"
+"node.crop.input.tex_in": "Texture"
+"node.crop.input.left_in": "Left"
+"node.crop.input.top_in": "Top"
+"node.crop.input.right_in": "Right"
+"node.crop.input.bottom_in": "Bottom"
+"node.crop.input.feather_in": "Feather"
+"node.value.name": "Value"
+"node.value.input.type_in": "Type"
+"node.value.input.value_in": "Value"
+"node.timeremap.name": "Time Remap"
+"node.timeremap.input.time_in": "Time"
+"node.timeremap.input.input_in": "Input"
+"node.shape.name": "Shape"
+"node.shape.input.type_in": "Type"
+"node.shape.input.base_in": "Base"
+"node.shape.input.pos_in": "Position"
+"node.shape.input.size_in": "Size"
+"node.shape.input.color_in": "Color"
+"node.shape.input.radius_in": "Radius"
+"node.colordifferencekey.name": "Color Difference Key"
+"node.colordifferencekey.input.tex_in": "Input"
+"node.colordifferencekey.input.garbage_in": "Garbage Matte"
+"node.colordifferencekey.input.core_in": "Core Matte"
+"node.colordifferencekey.input.color_in": "Key Color"
+"node.colordifferencekey.input.highlights_in": "Highlights"
+"node.colordifferencekey.input.shadows_in": "Shadows"
+"node.colordifferencekey.input.mask_only_in": "Show Mask Only"
+"node.despill.name": "Despill"
+"node.despill.input.tex_in": "Input"
+"node.despill.input.color_in": "Key Color"
+"node.despill.input.method_in": "Method"
+"node.despill.input.preserve_luminance_input": "Preserve Luminance"
+"node.group.name": "Group"
+"node.opacity.name": "Opacity"
+"node.opacity.input.tex_in": "Texture"
+"node.opacity.input.opacity_in": "Opacity"
+"node.flip.name": "Flip"
+"node.flip.input.tex_in": "Input"
+"node.flip.input.horiz_in": "Horizontal"
+"node.flip.input.vert_in": "Vertical"
+"node.noise.name": "Noise"
+"node.noise.input.base_in": "Base"
+"node.noise.input.strength_in": "Strength"
+"node.noise.input.color_in": "Color"
+"node.timeoffset.name": "Time Offset"
+"node.timeoffset.input.time_in": "Time"
+"node.timeoffset.input.input_in": "Input"
+"node.transition.name": "Transition"
+"node.transition.input.tex_in": "From"
+"node.transition.input.blend_in": "To"
+"node.transition.input.progress_in": "Progress"
+"node.transition.input.type_in": "Type"
+"node.transitionfx.name": "Transition FX"
+"node.transitionfx.input.tex_in": "From"
+"node.transitionfx.input.blend_in": "To"
+"node.transitionfx.input.progress_in": "Progress"
+"node.transitionfx.input.color_in": "Color"
+"node.transitionfx.input.direction_in": "Direction"
+"node.transitionfx.input.type_in": "Type"
+"node.cornerpin.name": "Corner Pin"
+"node.cornerpin.input.tex_in": "Texture"
+"node.cornerpin.input.perspective_in": "Perspective"
+"node.cornerpin.input.top_left_in": "Top Left"
+"node.cornerpin.input.top_right_in": "Top Right"
+"node.cornerpin.input.bottom_right_in": "Bottom Right"
+"node.cornerpin.input.bottom_left_in": "Bottom Left"
+"node.displaytransform.name": "Display Transform"
+"node.displaytransform.input.tex_in": "Input"
+"node.displaytransform.input.display_in": "Display"
+"node.displaytransform.input.view_in": "View"
+"node.displaytransform.input.dir_in": "Direction"
+"node.ociogradingtransformlinear.name": "OCIO Color Grading (Linear)"
+"node.ociogradingtransformlinear.input.tex_in": "Input"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "Contrast"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "Offset"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "Exposure"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "Saturation"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "Pivot"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "Black Clamp"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "Enable White Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO Color Grading (Log)"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "Input"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "Lift"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "Gain"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "Gamma"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "Saturation"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "Pivot"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "Enable White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "White Clamp"
+"node.whitebalance.name": "White Balance"
+"node.whitebalance.input.tex_in": "Input"
+"node.whitebalance.input.temperature_in": "Temperature (K)"
+"node.whitebalance.input.tint_in": "Tint"
+"node.ociolut.name": "OCIO LUT"
+"node.ociolut.input.tex_in": "Input"
+"node.ociolut.input.lut_file_in": "LUT File"
+"node.ociolut.input.lut_dir_in": "Direction"
+"node.threewaycolor.name": "Three-Way Color"
+"node.threewaycolor.input.tex_in": "Input"
+"node.threewaycolor.input.shadows_color_in": "Shadows"
+"node.threewaycolor.input.midtones_color_in": "Midtones"
+"node.threewaycolor.input.highlights_color_in": "Highlights"
+"node.threewaycolor.input.shadows_amount_in": "Shadows Amount"
+"node.threewaycolor.input.midtones_amount_in": "Midtones Amount"
+"node.threewaycolor.input.highlights_amount_in": "Highlights Amount"
+"node.chromakey.name": "Chroma Key"
+"node.chromakey.input.tex_in": "Input"
+"node.chromakey.input.color_key": "Key Color"
+"node.chromakey.input.lower_tolerance_in": "Lower Tolerance"
+"node.chromakey.input.upper_tolerance_in": "Upper Tolerance"
+"node.chromakey.input.garbage_in": "Garbage Matte"
+"node.chromakey.input.core_in": "Core Matte"
+"node.chromakey.input.highlights_in": "Highlights"
+"node.chromakey.input.shadows_in": "Shadows"
+"node.chromakey.input.invert_in": "Invert Mask"
+"node.chromakey.input.mask_only_in": "Show Mask Only"
+"node.mask.name": "Mask"
+"node.mask.input.base_in": "Texture"
+"node.mask.input.points_in": "Points"
+"node.mask.input.color_in": "Color"
+"node.mask.input.invert_in": "Invert"
+"node.mask.input.feather_in": "Feather"
+"node.dropshadow.name": "Drop Shadow"
+"node.dropshadow.input.tex_in": "Texture"
+"node.dropshadow.input.color_in": "Color"
+"node.dropshadow.input.distance_in": "Distance"
+"node.dropshadow.input.angle_in": "Angle"
+"node.dropshadow.input.radius_in": "Softness"
+"node.dropshadow.input.opacity_in": "Opacity"
+"node.dropshadow.input.fast_in": "Faster (Lower Quality)"
+"node.timeformat.name": "Time Format"
+"node.timeformat.input.time_in": "Time"
+"node.timeformat.input.format_in": "Format"
+"node.timeformat.input.localtime_in": "Interpret time as local time"
+"node.wave.name": "Wave"
+"node.wave.input.tex_in": "Input"
+"node.wave.input.frequency_in": "Frequency"
+"node.wave.input.intensity_in": "Intensity"
+"node.wave.input.evolution_in": "Evolution"
+"node.wave.input.vertical_in": "Direction"
+"node.tile.name": "Tile"
+"node.tile.input.tex_in": "Input"
+"node.tile.input.scale_in": "Scale"
+"node.tile.input.position_in": "Position"
+"node.tile.input.anchor_in": "Anchor"
+"node.tile.input.mirrorx_in": "Mirror Horizontally"
+"node.tile.input.mirrory_in": "Mirror Vertically"
+"node.swirl.name": "Swirl"
+"node.swirl.input.tex_in": "Input"
+"node.swirl.input.radius_in": "Radius"
+"node.swirl.input.angle_in": "Angle"
+"node.swirl.input.pos_in": "Position"
+"node.ripple.name": "Ripple"
+"node.ripple.input.tex_in": "Input"
+"node.ripple.input.evolution_in": "Evolution"
+"node.ripple.input.intensity_in": "Intensity"
+"node.ripple.input.frequency_in": "Frequency"
+"node.ripple.input.position_in": "Position"
+"node.ripple.input.stretch_in": "Stretch"
+"node.colormatrix.name": "Color Matrix"
+"node.colormatrix.input.tex_in": "Input"
+"node.colormatrix.input.m0": "R <- R"
+"node.colormatrix.input.m1": "R <- G"
+"node.colormatrix.input.m2": "R <- B"
+"node.colormatrix.input.m3": "R <- A"
+"node.colormatrix.input.m4": "G <- R"
+"node.colormatrix.input.m5": "G <- G"
+"node.colormatrix.input.m6": "G <- B"
+"node.colormatrix.input.m7": "G <- A"
+"node.colormatrix.input.m8": "B <- R"
+"node.colormatrix.input.m9": "B <- G"
+"node.colormatrix.input.m10": "B <- B"
+"node.colormatrix.input.m11": "B <- A"
+"node.colormatrix.input.m12": "A <- R"
+"node.colormatrix.input.m13": "A <- G"
+"node.colormatrix.input.m14": "A <- B"
+"node.colormatrix.input.m15": "A <- A"
+"node.dilate.name": "Dilate"
+"node.dilate.input.tex_in": "Input"
+"node.dilate.input.radius_in": "Radius"
+"node.edgedetect.name": "Edge Detect"
+"node.edgedetect.input.tex_in": "Input"
+"node.edgedetect.input.threshold_in": "Threshold"
+"node.erode.name": "Erode"
+"node.erode.input.tex_in": "Input"
+"node.erode.input.radius_in": "Radius"
+"node.colorcorrect.name": "Color Correct"
+"node.colorcorrect.input.tex_in": "Input"
+"node.colorcorrect.input.saturation_in": "Saturation"
+"node.colorcorrect.input.contrast_in": "Contrast"
+"node.colorcorrect.input.gamma_in": "Gamma"
+"node.colorcorrect.input.gain_in": "Gain"
+"node.colorcorrect.input.offset_in": "Offset"
+"node.gamma.name": "Gamma"
+"node.gamma.input.tex_in": "Input"
+"node.gamma.input.gamma_in": "Gamma"
+"node.saturation.name": "Saturation"
+"node.saturation.input.tex_in": "Input"
+"node.saturation.input.saturation_in": "Saturation"
+"node.invert.name": "Invert"
+"node.invert.input.tex_in": "Input"
+"node.invert.input.invert_r_in": "Invert Red"
+"node.invert.input.invert_g_in": "Invert Green"
+"node.invert.input.invert_b_in": "Invert Blue"
+"node.invert.input.invert_a_in": "Invert Alpha"
+"node.clamp.name": "Clamp"
+"node.clamp.input.tex_in": "Input"
+"node.clamp.input.min_in": "Min"
+"node.clamp.input.max_in": "Max"
+"node.grade.name": "Grade"
+"node.grade.input.tex_in": "Input"
+"node.grade.input.blackpoint_in": "Black Point"
+"node.grade.input.whitepoint_in": "White Point"
+"node.grade.input.black_in": "Black Output"
+"node.grade.input.white_in": "White Output"
+"node.grade.input.gamma_in": "Gamma"
+"node.dirblur.name": "Directional Blur"
+"node.dirblur.input.tex_in": "Input"
+"node.dirblur.input.amount_in": "Amount"
+"node.dirblur.input.angle_in": "Angle"
+"node.sharpen.name": "Sharpen"
+"node.sharpen.input.tex_in": "Input"
+"node.sharpen.input.amount_in": "Amount"
+"node.dissolve.name": "Dissolve"
+"node.dissolve.input.tex_in": "Input"
+"node.dissolve.input.blend_in": "Blend"
+"node.dissolve.input.mix_in": "Mix"
+"node.keymix.name": "KeyMix"
+"node.keymix.input.tex_in": "Input"
+"node.keymix.input.blend_in": "Blend"
+"node.keymix.input.mask_in": "Mask"
+"node.premult.name": "Premultiply"
+"node.premult.input.tex_in": "Input"
+"node.premult.input.premult_channel_in": "Channel"
+"node.unpremult.name": "Unpremultiply"
+"node.unpremult.input.tex_in": "Input"
+"node.unpremult.input.unpremult_channel_in": "Channel"
+"node.position.name": "Position"
+"node.position.input.tex_in": "Input"
+"node.position.input.offset_in": "Translate"
+"node.mirror.name": "Mirror"
+"node.mirror.input.tex_in": "Input"
+"node.mirror.input.horizontal_in": "Horizontal (flop)"
+"node.mirror.input.vertical_in": "Vertical (flip)"
+"node.checkerboard.name": "Checkerboard"
+"node.checkerboard.input.base_in": "Base"
+"node.checkerboard.input.size_in": "Box Size"
+"node.checkerboard.input.color1_in": "Color 1"
+"node.checkerboard.input.color2_in": "Color 2"
+"node.colorbars.name": "Color Bars"
+"node.colorbars.input.base_in": "Base"
+"node.colorbars.input.standard_in": "Standard"
+"node.ramp.name": "Ramp"
+"node.ramp.input.base_in": "Base"
+"node.ramp.input.point0_in": "Point 0"
+"node.ramp.input.point1_in": "Point 1"
+"node.ramp.input.color0_in": "Color 0"
+"node.ramp.input.color1_in": "Color 1"
+"node.multicam.name": "Multi-Cam"
+"node.multicam.input.current_in": "Current"
+"node.multicam.input.sources_in": "Sources"
+"node.multicam.input.sequence_in": "Sequence"
+"node.multicam.input.sequence_type_in": "Sequence Type"
 "status.ready": "準備完了"
 "status.cache": "キャッシュ:有効"
 "status.proxy.on": "プロキシ:オン"
@@ -481,6 +855,8 @@
 "node.category.group": "グループ"
 "project.context.new": "新規"
 "project.new_sequence": "新規シーケンス"
+"project.add_text_footage": "Add Text Footage"
+"timeline.context.add_adjustment_layer": "Add Adjustment Layer"
 "project.context.reveal_in_finder": "Finderに表示"
 "project.context.replace_footage": "素材を置き換え"
 "project.context.rename": "名前を変更"
diff --git a/assets/i18n/pt.yaml b/assets/i18n/pt.yaml
index a7188b5ef..11faef4a5 100644
--- a/assets/i18n/pt.yaml
+++ b/assets/i18n/pt.yaml
@@ -171,6 +171,380 @@
 "multicam.switch_9": "Alternar para a câmera 9"
 "effect_library.hint": "Clique duas vezes para adicionar ao clipe selecionado"
 "effect_library.group.builtin": "Integrados"
+"effect_library.group.color": "Color"
+"effect_library.group.filter": "Filter"
+"effect_library.group.distort": "Distort"
+"effect_library.group.keying": "Keying"
+"effect_library.group.generator": "Generator"
+"effect_library.group.math": "Math"
+"effect_library.group.time": "Time"
+"effect_library.group.channel": "Channel"
+"effect_library.group.transition": "Transition"
+"effect_library.group.general": "General"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "Polygon"
+"node.polygon.input.base_in": "Base"
+"node.polygon.input.points_in": "Points"
+"node.polygon.input.color_in": "Color"
+"node.ortho.name": "Orthographic Matrix"
+"node.ortho.input.pos_in": "Position"
+"node.ortho.input.rot_in": "Rotation"
+"node.ortho.input.scale_in": "Scale"
+"node.ortho.input.uniform_scale_in": "Uniform Scale"
+"node.ortho.input.anchor_in": "Anchor Point"
+"node.transform.name": "Transform"
+"node.transform.input.tex_in": "Texture"
+"node.transform.input.parent_in": "Parent"
+"node.transform.input.autoscale_in": "Auto-Scale"
+"node.transform.input.interpolation_in": "Interpolation"
+"node.transform.input.pos_in": "Position"
+"node.transform.input.rot_in": "Rotation"
+"node.transform.input.scale_in": "Scale"
+"node.transform.input.uniform_scale_in": "Uniform Scale"
+"node.transform.input.anchor_in": "Anchor Point"
+"node.volume.name": "Volume"
+"node.volume.input.samples_in": "Samples"
+"node.volume.input.volume_in": "Volume"
+"node.pan.name": "Pan"
+"node.pan.input.samples_in": "Samples"
+"node.pan.input.panning_in": "Pan"
+"node.math.name": "Math"
+"node.math.input.method_in": "Method"
+"node.math.input.param_a_in": "Value"
+"node.math.input.param_b_in": "Value"
+"node.time.name": "Time"
+"node.time.input.enabled_in": "Enabled"
+"node.trigonometry.name": "Trigonometry"
+"node.trigonometry.input.method_in": "Method"
+"node.trigonometry.input.x_in": "Value"
+"node.blur.name": "Blur"
+"node.blur.input.tex_in": "Input"
+"node.blur.input.method_in": "Method"
+"node.blur.input.radius_in": "Radius"
+"node.blur.input.horiz_in": "Horizontal"
+"node.blur.input.vert_in": "Vertical"
+"node.blur.input.directional_degrees_in": "Direction"
+"node.blur.input.radial_center_in": "Center"
+"node.blur.input.repeat_edge_pixels_in": "Repeat Edge Pixels"
+"node.solidgenerator.name": "Solid"
+"node.solidgenerator.input.color_in": "Color"
+"node.merge.name": "Merge"
+"node.merge.input.base_in": "Base"
+"node.merge.input.blend_in": "Blend"
+"node.stroke.name": "Stroke"
+"node.stroke.input.tex_in": "Input"
+"node.stroke.input.color_in": "Color"
+"node.stroke.input.radius_in": "Radius"
+"node.stroke.input.opacity_in": "Opacity"
+"node.stroke.input.inner_in": "Inner"
+"node.textgenerator.name": "Text (Legacy)"
+"node.textgenerator.input.text_in": "Text"
+"node.textgenerator.input.html_in": "Enable HTML"
+"node.textgenerator.input.color_in": "Color"
+"node.textgenerator.input.valign_in": "Vertical Align"
+"node.textgenerator.input.font_in": "Font"
+"node.textgenerator.input.font_size_in": "Font Size"
+"node.text2.name": "Text (Legacy)"
+"node.text2.input.base_in": "Base"
+"node.text2.input.pos_in": "Position"
+"node.text2.input.size_in": "Size"
+"node.text2.input.color_in": "Color"
+"node.text2.input.text_in": "Text"
+"node.text2.input.html_in": "Enable HTML"
+"node.text2.input.valign_in": "Vertical Align"
+"node.text2.input.font_in": "Font"
+"node.text2.input.font_size_in": "Font Size"
+"node.text3.name": "Text"
+"node.text3.input.base_in": "Base"
+"node.text3.input.pos_in": "Position"
+"node.text3.input.size_in": "Size"
+"node.text3.input.plain_text_in": "Text"
+"node.text3.input.text_in": "Text"
+"node.text3.input.font_family_in": "Font Family"
+"node.text3.input.font_size_in": "Font Size"
+"node.text3.input.outline_enabled_in": "Outline"
+"node.text3.input.outline_color_in": "Outline Color"
+"node.text3.input.outline_width_in": "Outline Width"
+"node.text3.input.glow_enabled_in": "Glow"
+"node.text3.input.glow_color_in": "Glow Color"
+"node.text3.input.glow_radius_in": "Glow Radius"
+"node.text3.input.color_in": "Color"
+"node.text3.input.valign_in": "Vertical Alignment"
+"node.text3.input.args_in": "Arguments"
+"node.mosaicfilter.name": "Mosaic"
+"node.mosaicfilter.input.tex_in": "Texture"
+"node.mosaicfilter.input.horiz_in": "Horizontal"
+"node.mosaicfilter.input.vert_in": "Vertical"
+"node.crop.name": "Crop"
+"node.crop.input.tex_in": "Texture"
+"node.crop.input.left_in": "Left"
+"node.crop.input.top_in": "Top"
+"node.crop.input.right_in": "Right"
+"node.crop.input.bottom_in": "Bottom"
+"node.crop.input.feather_in": "Feather"
+"node.value.name": "Value"
+"node.value.input.type_in": "Type"
+"node.value.input.value_in": "Value"
+"node.timeremap.name": "Time Remap"
+"node.timeremap.input.time_in": "Time"
+"node.timeremap.input.input_in": "Input"
+"node.shape.name": "Shape"
+"node.shape.input.type_in": "Type"
+"node.shape.input.base_in": "Base"
+"node.shape.input.pos_in": "Position"
+"node.shape.input.size_in": "Size"
+"node.shape.input.color_in": "Color"
+"node.shape.input.radius_in": "Radius"
+"node.colordifferencekey.name": "Color Difference Key"
+"node.colordifferencekey.input.tex_in": "Input"
+"node.colordifferencekey.input.garbage_in": "Garbage Matte"
+"node.colordifferencekey.input.core_in": "Core Matte"
+"node.colordifferencekey.input.color_in": "Key Color"
+"node.colordifferencekey.input.highlights_in": "Highlights"
+"node.colordifferencekey.input.shadows_in": "Shadows"
+"node.colordifferencekey.input.mask_only_in": "Show Mask Only"
+"node.despill.name": "Despill"
+"node.despill.input.tex_in": "Input"
+"node.despill.input.color_in": "Key Color"
+"node.despill.input.method_in": "Method"
+"node.despill.input.preserve_luminance_input": "Preserve Luminance"
+"node.group.name": "Group"
+"node.opacity.name": "Opacity"
+"node.opacity.input.tex_in": "Texture"
+"node.opacity.input.opacity_in": "Opacity"
+"node.flip.name": "Flip"
+"node.flip.input.tex_in": "Input"
+"node.flip.input.horiz_in": "Horizontal"
+"node.flip.input.vert_in": "Vertical"
+"node.noise.name": "Noise"
+"node.noise.input.base_in": "Base"
+"node.noise.input.strength_in": "Strength"
+"node.noise.input.color_in": "Color"
+"node.timeoffset.name": "Time Offset"
+"node.timeoffset.input.time_in": "Time"
+"node.timeoffset.input.input_in": "Input"
+"node.transition.name": "Transition"
+"node.transition.input.tex_in": "From"
+"node.transition.input.blend_in": "To"
+"node.transition.input.progress_in": "Progress"
+"node.transition.input.type_in": "Type"
+"node.transitionfx.name": "Transition FX"
+"node.transitionfx.input.tex_in": "From"
+"node.transitionfx.input.blend_in": "To"
+"node.transitionfx.input.progress_in": "Progress"
+"node.transitionfx.input.color_in": "Color"
+"node.transitionfx.input.direction_in": "Direction"
+"node.transitionfx.input.type_in": "Type"
+"node.cornerpin.name": "Corner Pin"
+"node.cornerpin.input.tex_in": "Texture"
+"node.cornerpin.input.perspective_in": "Perspective"
+"node.cornerpin.input.top_left_in": "Top Left"
+"node.cornerpin.input.top_right_in": "Top Right"
+"node.cornerpin.input.bottom_right_in": "Bottom Right"
+"node.cornerpin.input.bottom_left_in": "Bottom Left"
+"node.displaytransform.name": "Display Transform"
+"node.displaytransform.input.tex_in": "Input"
+"node.displaytransform.input.display_in": "Display"
+"node.displaytransform.input.view_in": "View"
+"node.displaytransform.input.dir_in": "Direction"
+"node.ociogradingtransformlinear.name": "OCIO Color Grading (Linear)"
+"node.ociogradingtransformlinear.input.tex_in": "Input"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "Contrast"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "Offset"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "Exposure"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "Saturation"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "Pivot"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "Black Clamp"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "Enable White Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO Color Grading (Log)"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "Input"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "Lift"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "Gain"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "Gamma"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "Saturation"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "Pivot"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "Enable White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "White Clamp"
+"node.whitebalance.name": "White Balance"
+"node.whitebalance.input.tex_in": "Input"
+"node.whitebalance.input.temperature_in": "Temperature (K)"
+"node.whitebalance.input.tint_in": "Tint"
+"node.ociolut.name": "OCIO LUT"
+"node.ociolut.input.tex_in": "Input"
+"node.ociolut.input.lut_file_in": "LUT File"
+"node.ociolut.input.lut_dir_in": "Direction"
+"node.threewaycolor.name": "Three-Way Color"
+"node.threewaycolor.input.tex_in": "Input"
+"node.threewaycolor.input.shadows_color_in": "Shadows"
+"node.threewaycolor.input.midtones_color_in": "Midtones"
+"node.threewaycolor.input.highlights_color_in": "Highlights"
+"node.threewaycolor.input.shadows_amount_in": "Shadows Amount"
+"node.threewaycolor.input.midtones_amount_in": "Midtones Amount"
+"node.threewaycolor.input.highlights_amount_in": "Highlights Amount"
+"node.chromakey.name": "Chroma Key"
+"node.chromakey.input.tex_in": "Input"
+"node.chromakey.input.color_key": "Key Color"
+"node.chromakey.input.lower_tolerance_in": "Lower Tolerance"
+"node.chromakey.input.upper_tolerance_in": "Upper Tolerance"
+"node.chromakey.input.garbage_in": "Garbage Matte"
+"node.chromakey.input.core_in": "Core Matte"
+"node.chromakey.input.highlights_in": "Highlights"
+"node.chromakey.input.shadows_in": "Shadows"
+"node.chromakey.input.invert_in": "Invert Mask"
+"node.chromakey.input.mask_only_in": "Show Mask Only"
+"node.mask.name": "Mask"
+"node.mask.input.base_in": "Texture"
+"node.mask.input.points_in": "Points"
+"node.mask.input.color_in": "Color"
+"node.mask.input.invert_in": "Invert"
+"node.mask.input.feather_in": "Feather"
+"node.dropshadow.name": "Drop Shadow"
+"node.dropshadow.input.tex_in": "Texture"
+"node.dropshadow.input.color_in": "Color"
+"node.dropshadow.input.distance_in": "Distance"
+"node.dropshadow.input.angle_in": "Angle"
+"node.dropshadow.input.radius_in": "Softness"
+"node.dropshadow.input.opacity_in": "Opacity"
+"node.dropshadow.input.fast_in": "Faster (Lower Quality)"
+"node.timeformat.name": "Time Format"
+"node.timeformat.input.time_in": "Time"
+"node.timeformat.input.format_in": "Format"
+"node.timeformat.input.localtime_in": "Interpret time as local time"
+"node.wave.name": "Wave"
+"node.wave.input.tex_in": "Input"
+"node.wave.input.frequency_in": "Frequency"
+"node.wave.input.intensity_in": "Intensity"
+"node.wave.input.evolution_in": "Evolution"
+"node.wave.input.vertical_in": "Direction"
+"node.tile.name": "Tile"
+"node.tile.input.tex_in": "Input"
+"node.tile.input.scale_in": "Scale"
+"node.tile.input.position_in": "Position"
+"node.tile.input.anchor_in": "Anchor"
+"node.tile.input.mirrorx_in": "Mirror Horizontally"
+"node.tile.input.mirrory_in": "Mirror Vertically"
+"node.swirl.name": "Swirl"
+"node.swirl.input.tex_in": "Input"
+"node.swirl.input.radius_in": "Radius"
+"node.swirl.input.angle_in": "Angle"
+"node.swirl.input.pos_in": "Position"
+"node.ripple.name": "Ripple"
+"node.ripple.input.tex_in": "Input"
+"node.ripple.input.evolution_in": "Evolution"
+"node.ripple.input.intensity_in": "Intensity"
+"node.ripple.input.frequency_in": "Frequency"
+"node.ripple.input.position_in": "Position"
+"node.ripple.input.stretch_in": "Stretch"
+"node.colormatrix.name": "Color Matrix"
+"node.colormatrix.input.tex_in": "Input"
+"node.colormatrix.input.m0": "R <- R"
+"node.colormatrix.input.m1": "R <- G"
+"node.colormatrix.input.m2": "R <- B"
+"node.colormatrix.input.m3": "R <- A"
+"node.colormatrix.input.m4": "G <- R"
+"node.colormatrix.input.m5": "G <- G"
+"node.colormatrix.input.m6": "G <- B"
+"node.colormatrix.input.m7": "G <- A"
+"node.colormatrix.input.m8": "B <- R"
+"node.colormatrix.input.m9": "B <- G"
+"node.colormatrix.input.m10": "B <- B"
+"node.colormatrix.input.m11": "B <- A"
+"node.colormatrix.input.m12": "A <- R"
+"node.colormatrix.input.m13": "A <- G"
+"node.colormatrix.input.m14": "A <- B"
+"node.colormatrix.input.m15": "A <- A"
+"node.dilate.name": "Dilate"
+"node.dilate.input.tex_in": "Input"
+"node.dilate.input.radius_in": "Radius"
+"node.edgedetect.name": "Edge Detect"
+"node.edgedetect.input.tex_in": "Input"
+"node.edgedetect.input.threshold_in": "Threshold"
+"node.erode.name": "Erode"
+"node.erode.input.tex_in": "Input"
+"node.erode.input.radius_in": "Radius"
+"node.colorcorrect.name": "Color Correct"
+"node.colorcorrect.input.tex_in": "Input"
+"node.colorcorrect.input.saturation_in": "Saturation"
+"node.colorcorrect.input.contrast_in": "Contrast"
+"node.colorcorrect.input.gamma_in": "Gamma"
+"node.colorcorrect.input.gain_in": "Gain"
+"node.colorcorrect.input.offset_in": "Offset"
+"node.gamma.name": "Gamma"
+"node.gamma.input.tex_in": "Input"
+"node.gamma.input.gamma_in": "Gamma"
+"node.saturation.name": "Saturation"
+"node.saturation.input.tex_in": "Input"
+"node.saturation.input.saturation_in": "Saturation"
+"node.invert.name": "Invert"
+"node.invert.input.tex_in": "Input"
+"node.invert.input.invert_r_in": "Invert Red"
+"node.invert.input.invert_g_in": "Invert Green"
+"node.invert.input.invert_b_in": "Invert Blue"
+"node.invert.input.invert_a_in": "Invert Alpha"
+"node.clamp.name": "Clamp"
+"node.clamp.input.tex_in": "Input"
+"node.clamp.input.min_in": "Min"
+"node.clamp.input.max_in": "Max"
+"node.grade.name": "Grade"
+"node.grade.input.tex_in": "Input"
+"node.grade.input.blackpoint_in": "Black Point"
+"node.grade.input.whitepoint_in": "White Point"
+"node.grade.input.black_in": "Black Output"
+"node.grade.input.white_in": "White Output"
+"node.grade.input.gamma_in": "Gamma"
+"node.dirblur.name": "Directional Blur"
+"node.dirblur.input.tex_in": "Input"
+"node.dirblur.input.amount_in": "Amount"
+"node.dirblur.input.angle_in": "Angle"
+"node.sharpen.name": "Sharpen"
+"node.sharpen.input.tex_in": "Input"
+"node.sharpen.input.amount_in": "Amount"
+"node.dissolve.name": "Dissolve"
+"node.dissolve.input.tex_in": "Input"
+"node.dissolve.input.blend_in": "Blend"
+"node.dissolve.input.mix_in": "Mix"
+"node.keymix.name": "KeyMix"
+"node.keymix.input.tex_in": "Input"
+"node.keymix.input.blend_in": "Blend"
+"node.keymix.input.mask_in": "Mask"
+"node.premult.name": "Premultiply"
+"node.premult.input.tex_in": "Input"
+"node.premult.input.premult_channel_in": "Channel"
+"node.unpremult.name": "Unpremultiply"
+"node.unpremult.input.tex_in": "Input"
+"node.unpremult.input.unpremult_channel_in": "Channel"
+"node.position.name": "Position"
+"node.position.input.tex_in": "Input"
+"node.position.input.offset_in": "Translate"
+"node.mirror.name": "Mirror"
+"node.mirror.input.tex_in": "Input"
+"node.mirror.input.horizontal_in": "Horizontal (flop)"
+"node.mirror.input.vertical_in": "Vertical (flip)"
+"node.checkerboard.name": "Checkerboard"
+"node.checkerboard.input.base_in": "Base"
+"node.checkerboard.input.size_in": "Box Size"
+"node.checkerboard.input.color1_in": "Color 1"
+"node.checkerboard.input.color2_in": "Color 2"
+"node.colorbars.name": "Color Bars"
+"node.colorbars.input.base_in": "Base"
+"node.colorbars.input.standard_in": "Standard"
+"node.ramp.name": "Ramp"
+"node.ramp.input.base_in": "Base"
+"node.ramp.input.point0_in": "Point 0"
+"node.ramp.input.point1_in": "Point 1"
+"node.ramp.input.color0_in": "Color 0"
+"node.ramp.input.color1_in": "Color 1"
+"node.multicam.name": "Multi-Cam"
+"node.multicam.input.current_in": "Current"
+"node.multicam.input.sources_in": "Sources"
+"node.multicam.input.sequence_in": "Sequence"
+"node.multicam.input.sequence_type_in": "Sequence Type"
 "status.ready": "Pronto"
 "status.cache": "Cache: ativado"
 "status.proxy.on": "Proxy: ativado"
@@ -481,6 +855,8 @@
 "node.category.group": "Grupo"
 "project.context.new": "Novo"
 "project.new_sequence": "Nova sequência"
+"project.add_text_footage": "Add Text Footage"
+"timeline.context.add_adjustment_layer": "Add Adjustment Layer"
 "project.context.reveal_in_finder": "Revelar no Finder"
 "project.context.replace_footage": "Substituir mídia"
 "project.context.rename": "Renomear"
diff --git a/assets/i18n/ru.yaml b/assets/i18n/ru.yaml
index 9d6a58119..fb092621e 100644
--- a/assets/i18n/ru.yaml
+++ b/assets/i18n/ru.yaml
@@ -171,6 +171,380 @@
 "multicam.switch_9": "Переключиться на камеру 9"
 "effect_library.hint": "Двойной щелчок добавляет эффект к выбранному клипу"
 "effect_library.group.builtin": "Встроенные"
+"effect_library.group.color": "Color"
+"effect_library.group.filter": "Filter"
+"effect_library.group.distort": "Distort"
+"effect_library.group.keying": "Keying"
+"effect_library.group.generator": "Generator"
+"effect_library.group.math": "Math"
+"effect_library.group.time": "Time"
+"effect_library.group.channel": "Channel"
+"effect_library.group.transition": "Transition"
+"effect_library.group.general": "General"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "Polygon"
+"node.polygon.input.base_in": "Base"
+"node.polygon.input.points_in": "Points"
+"node.polygon.input.color_in": "Color"
+"node.ortho.name": "Orthographic Matrix"
+"node.ortho.input.pos_in": "Position"
+"node.ortho.input.rot_in": "Rotation"
+"node.ortho.input.scale_in": "Scale"
+"node.ortho.input.uniform_scale_in": "Uniform Scale"
+"node.ortho.input.anchor_in": "Anchor Point"
+"node.transform.name": "Transform"
+"node.transform.input.tex_in": "Texture"
+"node.transform.input.parent_in": "Parent"
+"node.transform.input.autoscale_in": "Auto-Scale"
+"node.transform.input.interpolation_in": "Interpolation"
+"node.transform.input.pos_in": "Position"
+"node.transform.input.rot_in": "Rotation"
+"node.transform.input.scale_in": "Scale"
+"node.transform.input.uniform_scale_in": "Uniform Scale"
+"node.transform.input.anchor_in": "Anchor Point"
+"node.volume.name": "Volume"
+"node.volume.input.samples_in": "Samples"
+"node.volume.input.volume_in": "Volume"
+"node.pan.name": "Pan"
+"node.pan.input.samples_in": "Samples"
+"node.pan.input.panning_in": "Pan"
+"node.math.name": "Math"
+"node.math.input.method_in": "Method"
+"node.math.input.param_a_in": "Value"
+"node.math.input.param_b_in": "Value"
+"node.time.name": "Time"
+"node.time.input.enabled_in": "Enabled"
+"node.trigonometry.name": "Trigonometry"
+"node.trigonometry.input.method_in": "Method"
+"node.trigonometry.input.x_in": "Value"
+"node.blur.name": "Blur"
+"node.blur.input.tex_in": "Input"
+"node.blur.input.method_in": "Method"
+"node.blur.input.radius_in": "Radius"
+"node.blur.input.horiz_in": "Horizontal"
+"node.blur.input.vert_in": "Vertical"
+"node.blur.input.directional_degrees_in": "Direction"
+"node.blur.input.radial_center_in": "Center"
+"node.blur.input.repeat_edge_pixels_in": "Repeat Edge Pixels"
+"node.solidgenerator.name": "Solid"
+"node.solidgenerator.input.color_in": "Color"
+"node.merge.name": "Merge"
+"node.merge.input.base_in": "Base"
+"node.merge.input.blend_in": "Blend"
+"node.stroke.name": "Stroke"
+"node.stroke.input.tex_in": "Input"
+"node.stroke.input.color_in": "Color"
+"node.stroke.input.radius_in": "Radius"
+"node.stroke.input.opacity_in": "Opacity"
+"node.stroke.input.inner_in": "Inner"
+"node.textgenerator.name": "Text (Legacy)"
+"node.textgenerator.input.text_in": "Text"
+"node.textgenerator.input.html_in": "Enable HTML"
+"node.textgenerator.input.color_in": "Color"
+"node.textgenerator.input.valign_in": "Vertical Align"
+"node.textgenerator.input.font_in": "Font"
+"node.textgenerator.input.font_size_in": "Font Size"
+"node.text2.name": "Text (Legacy)"
+"node.text2.input.base_in": "Base"
+"node.text2.input.pos_in": "Position"
+"node.text2.input.size_in": "Size"
+"node.text2.input.color_in": "Color"
+"node.text2.input.text_in": "Text"
+"node.text2.input.html_in": "Enable HTML"
+"node.text2.input.valign_in": "Vertical Align"
+"node.text2.input.font_in": "Font"
+"node.text2.input.font_size_in": "Font Size"
+"node.text3.name": "Text"
+"node.text3.input.base_in": "Base"
+"node.text3.input.pos_in": "Position"
+"node.text3.input.size_in": "Size"
+"node.text3.input.plain_text_in": "Text"
+"node.text3.input.text_in": "Text"
+"node.text3.input.font_family_in": "Font Family"
+"node.text3.input.font_size_in": "Font Size"
+"node.text3.input.outline_enabled_in": "Outline"
+"node.text3.input.outline_color_in": "Outline Color"
+"node.text3.input.outline_width_in": "Outline Width"
+"node.text3.input.glow_enabled_in": "Glow"
+"node.text3.input.glow_color_in": "Glow Color"
+"node.text3.input.glow_radius_in": "Glow Radius"
+"node.text3.input.color_in": "Color"
+"node.text3.input.valign_in": "Vertical Alignment"
+"node.text3.input.args_in": "Arguments"
+"node.mosaicfilter.name": "Mosaic"
+"node.mosaicfilter.input.tex_in": "Texture"
+"node.mosaicfilter.input.horiz_in": "Horizontal"
+"node.mosaicfilter.input.vert_in": "Vertical"
+"node.crop.name": "Crop"
+"node.crop.input.tex_in": "Texture"
+"node.crop.input.left_in": "Left"
+"node.crop.input.top_in": "Top"
+"node.crop.input.right_in": "Right"
+"node.crop.input.bottom_in": "Bottom"
+"node.crop.input.feather_in": "Feather"
+"node.value.name": "Value"
+"node.value.input.type_in": "Type"
+"node.value.input.value_in": "Value"
+"node.timeremap.name": "Time Remap"
+"node.timeremap.input.time_in": "Time"
+"node.timeremap.input.input_in": "Input"
+"node.shape.name": "Shape"
+"node.shape.input.type_in": "Type"
+"node.shape.input.base_in": "Base"
+"node.shape.input.pos_in": "Position"
+"node.shape.input.size_in": "Size"
+"node.shape.input.color_in": "Color"
+"node.shape.input.radius_in": "Radius"
+"node.colordifferencekey.name": "Color Difference Key"
+"node.colordifferencekey.input.tex_in": "Input"
+"node.colordifferencekey.input.garbage_in": "Garbage Matte"
+"node.colordifferencekey.input.core_in": "Core Matte"
+"node.colordifferencekey.input.color_in": "Key Color"
+"node.colordifferencekey.input.highlights_in": "Highlights"
+"node.colordifferencekey.input.shadows_in": "Shadows"
+"node.colordifferencekey.input.mask_only_in": "Show Mask Only"
+"node.despill.name": "Despill"
+"node.despill.input.tex_in": "Input"
+"node.despill.input.color_in": "Key Color"
+"node.despill.input.method_in": "Method"
+"node.despill.input.preserve_luminance_input": "Preserve Luminance"
+"node.group.name": "Group"
+"node.opacity.name": "Opacity"
+"node.opacity.input.tex_in": "Texture"
+"node.opacity.input.opacity_in": "Opacity"
+"node.flip.name": "Flip"
+"node.flip.input.tex_in": "Input"
+"node.flip.input.horiz_in": "Horizontal"
+"node.flip.input.vert_in": "Vertical"
+"node.noise.name": "Noise"
+"node.noise.input.base_in": "Base"
+"node.noise.input.strength_in": "Strength"
+"node.noise.input.color_in": "Color"
+"node.timeoffset.name": "Time Offset"
+"node.timeoffset.input.time_in": "Time"
+"node.timeoffset.input.input_in": "Input"
+"node.transition.name": "Transition"
+"node.transition.input.tex_in": "From"
+"node.transition.input.blend_in": "To"
+"node.transition.input.progress_in": "Progress"
+"node.transition.input.type_in": "Type"
+"node.transitionfx.name": "Transition FX"
+"node.transitionfx.input.tex_in": "From"
+"node.transitionfx.input.blend_in": "To"
+"node.transitionfx.input.progress_in": "Progress"
+"node.transitionfx.input.color_in": "Color"
+"node.transitionfx.input.direction_in": "Direction"
+"node.transitionfx.input.type_in": "Type"
+"node.cornerpin.name": "Corner Pin"
+"node.cornerpin.input.tex_in": "Texture"
+"node.cornerpin.input.perspective_in": "Perspective"
+"node.cornerpin.input.top_left_in": "Top Left"
+"node.cornerpin.input.top_right_in": "Top Right"
+"node.cornerpin.input.bottom_right_in": "Bottom Right"
+"node.cornerpin.input.bottom_left_in": "Bottom Left"
+"node.displaytransform.name": "Display Transform"
+"node.displaytransform.input.tex_in": "Input"
+"node.displaytransform.input.display_in": "Display"
+"node.displaytransform.input.view_in": "View"
+"node.displaytransform.input.dir_in": "Direction"
+"node.ociogradingtransformlinear.name": "OCIO Color Grading (Linear)"
+"node.ociogradingtransformlinear.input.tex_in": "Input"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "Contrast"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "Offset"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "Exposure"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "Saturation"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "Pivot"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "Black Clamp"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "Enable White Clamp"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO Color Grading (Log)"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "Input"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "Lift"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "Gain"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "Gamma"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "Saturation"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "Pivot"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "Enable Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "Black Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "Enable White Clamp"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "White Clamp"
+"node.whitebalance.name": "White Balance"
+"node.whitebalance.input.tex_in": "Input"
+"node.whitebalance.input.temperature_in": "Temperature (K)"
+"node.whitebalance.input.tint_in": "Tint"
+"node.ociolut.name": "OCIO LUT"
+"node.ociolut.input.tex_in": "Input"
+"node.ociolut.input.lut_file_in": "LUT File"
+"node.ociolut.input.lut_dir_in": "Direction"
+"node.threewaycolor.name": "Three-Way Color"
+"node.threewaycolor.input.tex_in": "Input"
+"node.threewaycolor.input.shadows_color_in": "Shadows"
+"node.threewaycolor.input.midtones_color_in": "Midtones"
+"node.threewaycolor.input.highlights_color_in": "Highlights"
+"node.threewaycolor.input.shadows_amount_in": "Shadows Amount"
+"node.threewaycolor.input.midtones_amount_in": "Midtones Amount"
+"node.threewaycolor.input.highlights_amount_in": "Highlights Amount"
+"node.chromakey.name": "Chroma Key"
+"node.chromakey.input.tex_in": "Input"
+"node.chromakey.input.color_key": "Key Color"
+"node.chromakey.input.lower_tolerance_in": "Lower Tolerance"
+"node.chromakey.input.upper_tolerance_in": "Upper Tolerance"
+"node.chromakey.input.garbage_in": "Garbage Matte"
+"node.chromakey.input.core_in": "Core Matte"
+"node.chromakey.input.highlights_in": "Highlights"
+"node.chromakey.input.shadows_in": "Shadows"
+"node.chromakey.input.invert_in": "Invert Mask"
+"node.chromakey.input.mask_only_in": "Show Mask Only"
+"node.mask.name": "Mask"
+"node.mask.input.base_in": "Texture"
+"node.mask.input.points_in": "Points"
+"node.mask.input.color_in": "Color"
+"node.mask.input.invert_in": "Invert"
+"node.mask.input.feather_in": "Feather"
+"node.dropshadow.name": "Drop Shadow"
+"node.dropshadow.input.tex_in": "Texture"
+"node.dropshadow.input.color_in": "Color"
+"node.dropshadow.input.distance_in": "Distance"
+"node.dropshadow.input.angle_in": "Angle"
+"node.dropshadow.input.radius_in": "Softness"
+"node.dropshadow.input.opacity_in": "Opacity"
+"node.dropshadow.input.fast_in": "Faster (Lower Quality)"
+"node.timeformat.name": "Time Format"
+"node.timeformat.input.time_in": "Time"
+"node.timeformat.input.format_in": "Format"
+"node.timeformat.input.localtime_in": "Interpret time as local time"
+"node.wave.name": "Wave"
+"node.wave.input.tex_in": "Input"
+"node.wave.input.frequency_in": "Frequency"
+"node.wave.input.intensity_in": "Intensity"
+"node.wave.input.evolution_in": "Evolution"
+"node.wave.input.vertical_in": "Direction"
+"node.tile.name": "Tile"
+"node.tile.input.tex_in": "Input"
+"node.tile.input.scale_in": "Scale"
+"node.tile.input.position_in": "Position"
+"node.tile.input.anchor_in": "Anchor"
+"node.tile.input.mirrorx_in": "Mirror Horizontally"
+"node.tile.input.mirrory_in": "Mirror Vertically"
+"node.swirl.name": "Swirl"
+"node.swirl.input.tex_in": "Input"
+"node.swirl.input.radius_in": "Radius"
+"node.swirl.input.angle_in": "Angle"
+"node.swirl.input.pos_in": "Position"
+"node.ripple.name": "Ripple"
+"node.ripple.input.tex_in": "Input"
+"node.ripple.input.evolution_in": "Evolution"
+"node.ripple.input.intensity_in": "Intensity"
+"node.ripple.input.frequency_in": "Frequency"
+"node.ripple.input.position_in": "Position"
+"node.ripple.input.stretch_in": "Stretch"
+"node.colormatrix.name": "Color Matrix"
+"node.colormatrix.input.tex_in": "Input"
+"node.colormatrix.input.m0": "R <- R"
+"node.colormatrix.input.m1": "R <- G"
+"node.colormatrix.input.m2": "R <- B"
+"node.colormatrix.input.m3": "R <- A"
+"node.colormatrix.input.m4": "G <- R"
+"node.colormatrix.input.m5": "G <- G"
+"node.colormatrix.input.m6": "G <- B"
+"node.colormatrix.input.m7": "G <- A"
+"node.colormatrix.input.m8": "B <- R"
+"node.colormatrix.input.m9": "B <- G"
+"node.colormatrix.input.m10": "B <- B"
+"node.colormatrix.input.m11": "B <- A"
+"node.colormatrix.input.m12": "A <- R"
+"node.colormatrix.input.m13": "A <- G"
+"node.colormatrix.input.m14": "A <- B"
+"node.colormatrix.input.m15": "A <- A"
+"node.dilate.name": "Dilate"
+"node.dilate.input.tex_in": "Input"
+"node.dilate.input.radius_in": "Radius"
+"node.edgedetect.name": "Edge Detect"
+"node.edgedetect.input.tex_in": "Input"
+"node.edgedetect.input.threshold_in": "Threshold"
+"node.erode.name": "Erode"
+"node.erode.input.tex_in": "Input"
+"node.erode.input.radius_in": "Radius"
+"node.colorcorrect.name": "Color Correct"
+"node.colorcorrect.input.tex_in": "Input"
+"node.colorcorrect.input.saturation_in": "Saturation"
+"node.colorcorrect.input.contrast_in": "Contrast"
+"node.colorcorrect.input.gamma_in": "Gamma"
+"node.colorcorrect.input.gain_in": "Gain"
+"node.colorcorrect.input.offset_in": "Offset"
+"node.gamma.name": "Gamma"
+"node.gamma.input.tex_in": "Input"
+"node.gamma.input.gamma_in": "Gamma"
+"node.saturation.name": "Saturation"
+"node.saturation.input.tex_in": "Input"
+"node.saturation.input.saturation_in": "Saturation"
+"node.invert.name": "Invert"
+"node.invert.input.tex_in": "Input"
+"node.invert.input.invert_r_in": "Invert Red"
+"node.invert.input.invert_g_in": "Invert Green"
+"node.invert.input.invert_b_in": "Invert Blue"
+"node.invert.input.invert_a_in": "Invert Alpha"
+"node.clamp.name": "Clamp"
+"node.clamp.input.tex_in": "Input"
+"node.clamp.input.min_in": "Min"
+"node.clamp.input.max_in": "Max"
+"node.grade.name": "Grade"
+"node.grade.input.tex_in": "Input"
+"node.grade.input.blackpoint_in": "Black Point"
+"node.grade.input.whitepoint_in": "White Point"
+"node.grade.input.black_in": "Black Output"
+"node.grade.input.white_in": "White Output"
+"node.grade.input.gamma_in": "Gamma"
+"node.dirblur.name": "Directional Blur"
+"node.dirblur.input.tex_in": "Input"
+"node.dirblur.input.amount_in": "Amount"
+"node.dirblur.input.angle_in": "Angle"
+"node.sharpen.name": "Sharpen"
+"node.sharpen.input.tex_in": "Input"
+"node.sharpen.input.amount_in": "Amount"
+"node.dissolve.name": "Dissolve"
+"node.dissolve.input.tex_in": "Input"
+"node.dissolve.input.blend_in": "Blend"
+"node.dissolve.input.mix_in": "Mix"
+"node.keymix.name": "KeyMix"
+"node.keymix.input.tex_in": "Input"
+"node.keymix.input.blend_in": "Blend"
+"node.keymix.input.mask_in": "Mask"
+"node.premult.name": "Premultiply"
+"node.premult.input.tex_in": "Input"
+"node.premult.input.premult_channel_in": "Channel"
+"node.unpremult.name": "Unpremultiply"
+"node.unpremult.input.tex_in": "Input"
+"node.unpremult.input.unpremult_channel_in": "Channel"
+"node.position.name": "Position"
+"node.position.input.tex_in": "Input"
+"node.position.input.offset_in": "Translate"
+"node.mirror.name": "Mirror"
+"node.mirror.input.tex_in": "Input"
+"node.mirror.input.horizontal_in": "Horizontal (flop)"
+"node.mirror.input.vertical_in": "Vertical (flip)"
+"node.checkerboard.name": "Checkerboard"
+"node.checkerboard.input.base_in": "Base"
+"node.checkerboard.input.size_in": "Box Size"
+"node.checkerboard.input.color1_in": "Color 1"
+"node.checkerboard.input.color2_in": "Color 2"
+"node.colorbars.name": "Color Bars"
+"node.colorbars.input.base_in": "Base"
+"node.colorbars.input.standard_in": "Standard"
+"node.ramp.name": "Ramp"
+"node.ramp.input.base_in": "Base"
+"node.ramp.input.point0_in": "Point 0"
+"node.ramp.input.point1_in": "Point 1"
+"node.ramp.input.color0_in": "Color 0"
+"node.ramp.input.color1_in": "Color 1"
+"node.multicam.name": "Multi-Cam"
+"node.multicam.input.current_in": "Current"
+"node.multicam.input.sources_in": "Sources"
+"node.multicam.input.sequence_in": "Sequence"
+"node.multicam.input.sequence_type_in": "Sequence Type"
 "status.ready": "Готово"
 "status.cache": "Кэш: включён"
 "status.proxy.on": "Прокси: вкл"
@@ -481,6 +855,8 @@
 "node.category.group": "Группа"
 "project.context.new": "Создать"
 "project.new_sequence": "Новая последовательность"
+"project.add_text_footage": "Add Text Footage"
+"timeline.context.add_adjustment_layer": "Add Adjustment Layer"
 "project.context.reveal_in_finder": "Показать в Finder"
 "project.context.replace_footage": "Заменить медиафайлы"
 "project.context.rename": "Переименовать"
diff --git a/assets/i18n/zh-CN.yaml b/assets/i18n/zh-CN.yaml
index bcd82e187..18438d02a 100644
--- a/assets/i18n/zh-CN.yaml
+++ b/assets/i18n/zh-CN.yaml
@@ -169,6 +169,380 @@
 "multicam.switch_9": "切换到机位 9"
 "effect_library.hint": "双击添加到选中片段"
 "effect_library.group.builtin": "内置"
+"effect_library.group.color": "调色"
+"effect_library.group.filter": "滤镜"
+"effect_library.group.distort": "扭曲"
+"effect_library.group.keying": "键控"
+"effect_library.group.generator": "生成器"
+"effect_library.group.math": "数学"
+"effect_library.group.time": "时间"
+"effect_library.group.channel": "通道"
+"effect_library.group.transition": "转场"
+"effect_library.group.general": "通用"
+# Built-in effect (node) and effect-parameter (input) names. The effect
+# library and the inspector build these keys at runtime as
+# `node.<type-id suffix>.name` and `node.<type-id suffix>.input.<input id>`.
+"node.polygon.name": "多边形"
+"node.polygon.input.base_in": "基础"
+"node.polygon.input.points_in": "顶点"
+"node.polygon.input.color_in": "颜色"
+"node.ortho.name": "正交矩阵"
+"node.ortho.input.pos_in": "位置"
+"node.ortho.input.rot_in": "旋转"
+"node.ortho.input.scale_in": "缩放"
+"node.ortho.input.uniform_scale_in": "等比缩放"
+"node.ortho.input.anchor_in": "锚点"
+"node.transform.name": "变换"
+"node.transform.input.tex_in": "纹理"
+"node.transform.input.parent_in": "父级"
+"node.transform.input.autoscale_in": "自动缩放"
+"node.transform.input.interpolation_in": "插值"
+"node.transform.input.pos_in": "位置"
+"node.transform.input.rot_in": "旋转"
+"node.transform.input.scale_in": "缩放"
+"node.transform.input.uniform_scale_in": "等比缩放"
+"node.transform.input.anchor_in": "锚点"
+"node.volume.name": "音量"
+"node.volume.input.samples_in": "采样数"
+"node.volume.input.volume_in": "音量"
+"node.pan.name": "声像"
+"node.pan.input.samples_in": "采样数"
+"node.pan.input.panning_in": "声像"
+"node.math.name": "数学"
+"node.math.input.method_in": "方法"
+"node.math.input.param_a_in": "数值"
+"node.math.input.param_b_in": "数值"
+"node.time.name": "时间"
+"node.time.input.enabled_in": "启用"
+"node.trigonometry.name": "三角函数"
+"node.trigonometry.input.method_in": "方法"
+"node.trigonometry.input.x_in": "数值"
+"node.blur.name": "模糊"
+"node.blur.input.tex_in": "输入"
+"node.blur.input.method_in": "方法"
+"node.blur.input.radius_in": "半径"
+"node.blur.input.horiz_in": "水平"
+"node.blur.input.vert_in": "垂直"
+"node.blur.input.directional_degrees_in": "方向"
+"node.blur.input.radial_center_in": "中心"
+"node.blur.input.repeat_edge_pixels_in": "重复边缘像素"
+"node.solidgenerator.name": "纯色"
+"node.solidgenerator.input.color_in": "颜色"
+"node.merge.name": "合并"
+"node.merge.input.base_in": "基础"
+"node.merge.input.blend_in": "混合层"
+"node.stroke.name": "描边"
+"node.stroke.input.tex_in": "输入"
+"node.stroke.input.color_in": "颜色"
+"node.stroke.input.radius_in": "半径"
+"node.stroke.input.opacity_in": "不透明度"
+"node.stroke.input.inner_in": "内部"
+"node.textgenerator.name": "文本（旧版）"
+"node.textgenerator.input.text_in": "文本"
+"node.textgenerator.input.html_in": "启用 HTML"
+"node.textgenerator.input.color_in": "颜色"
+"node.textgenerator.input.valign_in": "垂直对齐"
+"node.textgenerator.input.font_in": "字体"
+"node.textgenerator.input.font_size_in": "字号"
+"node.text2.name": "文本（旧版）"
+"node.text2.input.base_in": "基础"
+"node.text2.input.pos_in": "位置"
+"node.text2.input.size_in": "尺寸"
+"node.text2.input.color_in": "颜色"
+"node.text2.input.text_in": "文本"
+"node.text2.input.html_in": "启用 HTML"
+"node.text2.input.valign_in": "垂直对齐"
+"node.text2.input.font_in": "字体"
+"node.text2.input.font_size_in": "字号"
+"node.text3.name": "文本"
+"node.text3.input.base_in": "基础"
+"node.text3.input.pos_in": "位置"
+"node.text3.input.size_in": "尺寸"
+"node.text3.input.plain_text_in": "文本"
+"node.text3.input.text_in": "文本"
+"node.text3.input.font_family_in": "字体族"
+"node.text3.input.font_size_in": "字号"
+"node.text3.input.outline_enabled_in": "描边"
+"node.text3.input.outline_color_in": "描边颜色"
+"node.text3.input.outline_width_in": "描边宽度"
+"node.text3.input.glow_enabled_in": "辉光"
+"node.text3.input.glow_color_in": "辉光颜色"
+"node.text3.input.glow_radius_in": "辉光半径"
+"node.text3.input.color_in": "颜色"
+"node.text3.input.valign_in": "垂直对齐"
+"node.text3.input.args_in": "参数列表"
+"node.mosaicfilter.name": "马赛克"
+"node.mosaicfilter.input.tex_in": "纹理"
+"node.mosaicfilter.input.horiz_in": "水平"
+"node.mosaicfilter.input.vert_in": "垂直"
+"node.crop.name": "裁剪"
+"node.crop.input.tex_in": "纹理"
+"node.crop.input.left_in": "左"
+"node.crop.input.top_in": "上"
+"node.crop.input.right_in": "右"
+"node.crop.input.bottom_in": "下"
+"node.crop.input.feather_in": "羽化"
+"node.value.name": "数值"
+"node.value.input.type_in": "类型"
+"node.value.input.value_in": "数值"
+"node.timeremap.name": "时间重映射"
+"node.timeremap.input.time_in": "时间"
+"node.timeremap.input.input_in": "输入"
+"node.shape.name": "形状"
+"node.shape.input.type_in": "类型"
+"node.shape.input.base_in": "基础"
+"node.shape.input.pos_in": "位置"
+"node.shape.input.size_in": "尺寸"
+"node.shape.input.color_in": "颜色"
+"node.shape.input.radius_in": "半径"
+"node.colordifferencekey.name": "色差键控"
+"node.colordifferencekey.input.tex_in": "输入"
+"node.colordifferencekey.input.garbage_in": "垃圾遮罩"
+"node.colordifferencekey.input.core_in": "核心遮罩"
+"node.colordifferencekey.input.color_in": "键控色"
+"node.colordifferencekey.input.highlights_in": "高光"
+"node.colordifferencekey.input.shadows_in": "阴影"
+"node.colordifferencekey.input.mask_only_in": "仅显示遮罩"
+"node.despill.name": "去溢色"
+"node.despill.input.tex_in": "输入"
+"node.despill.input.color_in": "键控色"
+"node.despill.input.method_in": "方法"
+"node.despill.input.preserve_luminance_input": "保持亮度"
+"node.group.name": "组"
+"node.opacity.name": "不透明度"
+"node.opacity.input.tex_in": "纹理"
+"node.opacity.input.opacity_in": "不透明度"
+"node.flip.name": "翻转"
+"node.flip.input.tex_in": "输入"
+"node.flip.input.horiz_in": "水平"
+"node.flip.input.vert_in": "垂直"
+"node.noise.name": "噪波"
+"node.noise.input.base_in": "基础"
+"node.noise.input.strength_in": "强度"
+"node.noise.input.color_in": "颜色"
+"node.timeoffset.name": "时间偏移"
+"node.timeoffset.input.time_in": "时间"
+"node.timeoffset.input.input_in": "输入"
+"node.transition.name": "转场"
+"node.transition.input.tex_in": "起点"
+"node.transition.input.blend_in": "终点"
+"node.transition.input.progress_in": "进度"
+"node.transition.input.type_in": "类型"
+"node.transitionfx.name": "转场特效"
+"node.transitionfx.input.tex_in": "起点"
+"node.transitionfx.input.blend_in": "终点"
+"node.transitionfx.input.progress_in": "进度"
+"node.transitionfx.input.color_in": "颜色"
+"node.transitionfx.input.direction_in": "方向"
+"node.transitionfx.input.type_in": "类型"
+"node.cornerpin.name": "角点定位"
+"node.cornerpin.input.tex_in": "纹理"
+"node.cornerpin.input.perspective_in": "透视"
+"node.cornerpin.input.top_left_in": "左上角"
+"node.cornerpin.input.top_right_in": "右上角"
+"node.cornerpin.input.bottom_right_in": "右下角"
+"node.cornerpin.input.bottom_left_in": "左下角"
+"node.displaytransform.name": "显示变换"
+"node.displaytransform.input.tex_in": "输入"
+"node.displaytransform.input.display_in": "显示"
+"node.displaytransform.input.view_in": "视图"
+"node.displaytransform.input.dir_in": "方向"
+"node.ociogradingtransformlinear.name": "OCIO 调色（线性）"
+"node.ociogradingtransformlinear.input.tex_in": "输入"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_contrast": "对比度"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_offset": "偏移"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_exposure": "曝光"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_saturation": "饱和度"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_pivot": "轴心"
+"node.ociogradingtransformlinear.input.clamp_black_enable_in": "启用黑场限幅"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampBlack": "黑场限幅"
+"node.ociogradingtransformlinear.input.clamp_white_enable_in": "启用白场限幅"
+"node.ociogradingtransformlinear.input.ocio_grading_primary_clampWhite": "白场限幅"
+"node.OCIO_NAMESPACEgradingtransformlog.name": "OCIO 调色（对数）"
+"node.OCIO_NAMESPACEgradingtransformlog.input.tex_in": "输入"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_brightness": "提升"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_contrast": "增益"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_gamma": "伽马"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_saturation": "饱和度"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_pivot": "轴心"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_black_enable_in": "启用黑场限幅"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampBlack": "黑场限幅"
+"node.OCIO_NAMESPACEgradingtransformlog.input.clamp_white_enable_in": "启用白场限幅"
+"node.OCIO_NAMESPACEgradingtransformlog.input.OCIO_NAMESPACE_grading_primary_clampWhite": "白场限幅"
+"node.whitebalance.name": "白平衡"
+"node.whitebalance.input.tex_in": "输入"
+"node.whitebalance.input.temperature_in": "色温 (K)"
+"node.whitebalance.input.tint_in": "色调"
+"node.ociolut.name": "OCIO 查找表"
+"node.ociolut.input.tex_in": "输入"
+"node.ociolut.input.lut_file_in": "LUT 文件"
+"node.ociolut.input.lut_dir_in": "方向"
+"node.threewaycolor.name": "三向调色"
+"node.threewaycolor.input.tex_in": "输入"
+"node.threewaycolor.input.shadows_color_in": "阴影"
+"node.threewaycolor.input.midtones_color_in": "中间调"
+"node.threewaycolor.input.highlights_color_in": "高光"
+"node.threewaycolor.input.shadows_amount_in": "阴影量"
+"node.threewaycolor.input.midtones_amount_in": "中间调量"
+"node.threewaycolor.input.highlights_amount_in": "高光量"
+"node.chromakey.name": "色度键"
+"node.chromakey.input.tex_in": "输入"
+"node.chromakey.input.color_key": "键控色"
+"node.chromakey.input.lower_tolerance_in": "下限容差"
+"node.chromakey.input.upper_tolerance_in": "上限容差"
+"node.chromakey.input.garbage_in": "垃圾遮罩"
+"node.chromakey.input.core_in": "核心遮罩"
+"node.chromakey.input.highlights_in": "高光"
+"node.chromakey.input.shadows_in": "阴影"
+"node.chromakey.input.invert_in": "反转遮罩"
+"node.chromakey.input.mask_only_in": "仅显示遮罩"
+"node.mask.name": "遮罩"
+"node.mask.input.base_in": "纹理"
+"node.mask.input.points_in": "顶点"
+"node.mask.input.color_in": "颜色"
+"node.mask.input.invert_in": "反转"
+"node.mask.input.feather_in": "羽化"
+"node.dropshadow.name": "投影"
+"node.dropshadow.input.tex_in": "纹理"
+"node.dropshadow.input.color_in": "颜色"
+"node.dropshadow.input.distance_in": "距离"
+"node.dropshadow.input.angle_in": "角度"
+"node.dropshadow.input.radius_in": "柔和度"
+"node.dropshadow.input.opacity_in": "不透明度"
+"node.dropshadow.input.fast_in": "更快（较低质量）"
+"node.timeformat.name": "时间格式"
+"node.timeformat.input.time_in": "时间"
+"node.timeformat.input.format_in": "格式"
+"node.timeformat.input.localtime_in": "按本地时间解释"
+"node.wave.name": "波纹"
+"node.wave.input.tex_in": "输入"
+"node.wave.input.frequency_in": "频率"
+"node.wave.input.intensity_in": "强度"
+"node.wave.input.evolution_in": "演化"
+"node.wave.input.vertical_in": "方向"
+"node.tile.name": "平铺"
+"node.tile.input.tex_in": "输入"
+"node.tile.input.scale_in": "缩放"
+"node.tile.input.position_in": "位置"
+"node.tile.input.anchor_in": "锚点"
+"node.tile.input.mirrorx_in": "水平镜像"
+"node.tile.input.mirrory_in": "垂直镜像"
+"node.swirl.name": "漩涡"
+"node.swirl.input.tex_in": "输入"
+"node.swirl.input.radius_in": "半径"
+"node.swirl.input.angle_in": "角度"
+"node.swirl.input.pos_in": "位置"
+"node.ripple.name": "涟漪"
+"node.ripple.input.tex_in": "输入"
+"node.ripple.input.evolution_in": "演化"
+"node.ripple.input.intensity_in": "强度"
+"node.ripple.input.frequency_in": "频率"
+"node.ripple.input.position_in": "位置"
+"node.ripple.input.stretch_in": "拉伸"
+"node.colormatrix.name": "颜色矩阵"
+"node.colormatrix.input.tex_in": "输入"
+"node.colormatrix.input.m0": "R ← R"
+"node.colormatrix.input.m1": "R ← G"
+"node.colormatrix.input.m2": "R ← B"
+"node.colormatrix.input.m3": "R ← A"
+"node.colormatrix.input.m4": "G ← R"
+"node.colormatrix.input.m5": "G ← G"
+"node.colormatrix.input.m6": "G ← B"
+"node.colormatrix.input.m7": "G ← A"
+"node.colormatrix.input.m8": "B ← R"
+"node.colormatrix.input.m9": "B ← G"
+"node.colormatrix.input.m10": "B ← B"
+"node.colormatrix.input.m11": "B ← A"
+"node.colormatrix.input.m12": "A ← R"
+"node.colormatrix.input.m13": "A ← G"
+"node.colormatrix.input.m14": "A ← B"
+"node.colormatrix.input.m15": "A ← A"
+"node.dilate.name": "膨胀"
+"node.dilate.input.tex_in": "输入"
+"node.dilate.input.radius_in": "半径"
+"node.edgedetect.name": "边缘检测"
+"node.edgedetect.input.tex_in": "输入"
+"node.edgedetect.input.threshold_in": "阈值"
+"node.erode.name": "腐蚀"
+"node.erode.input.tex_in": "输入"
+"node.erode.input.radius_in": "半径"
+"node.colorcorrect.name": "颜色校正"
+"node.colorcorrect.input.tex_in": "输入"
+"node.colorcorrect.input.saturation_in": "饱和度"
+"node.colorcorrect.input.contrast_in": "对比度"
+"node.colorcorrect.input.gamma_in": "伽马"
+"node.colorcorrect.input.gain_in": "增益"
+"node.colorcorrect.input.offset_in": "偏移"
+"node.gamma.name": "伽马"
+"node.gamma.input.tex_in": "输入"
+"node.gamma.input.gamma_in": "伽马"
+"node.saturation.name": "饱和度"
+"node.saturation.input.tex_in": "输入"
+"node.saturation.input.saturation_in": "饱和度"
+"node.invert.name": "反转"
+"node.invert.input.tex_in": "输入"
+"node.invert.input.invert_r_in": "反转红"
+"node.invert.input.invert_g_in": "反转绿"
+"node.invert.input.invert_b_in": "反转蓝"
+"node.invert.input.invert_a_in": "反转 Alpha"
+"node.clamp.name": "限幅"
+"node.clamp.input.tex_in": "输入"
+"node.clamp.input.min_in": "最小"
+"node.clamp.input.max_in": "最大"
+"node.grade.name": "调色"
+"node.grade.input.tex_in": "输入"
+"node.grade.input.blackpoint_in": "黑点"
+"node.grade.input.whitepoint_in": "白点"
+"node.grade.input.black_in": "黑场输出"
+"node.grade.input.white_in": "白场输出"
+"node.grade.input.gamma_in": "伽马"
+"node.dirblur.name": "方向模糊"
+"node.dirblur.input.tex_in": "输入"
+"node.dirblur.input.amount_in": "强度"
+"node.dirblur.input.angle_in": "角度"
+"node.sharpen.name": "锐化"
+"node.sharpen.input.tex_in": "输入"
+"node.sharpen.input.amount_in": "强度"
+"node.dissolve.name": "溶解"
+"node.dissolve.input.tex_in": "输入"
+"node.dissolve.input.blend_in": "混合层"
+"node.dissolve.input.mix_in": "混合"
+"node.keymix.name": "键控混合"
+"node.keymix.input.tex_in": "输入"
+"node.keymix.input.blend_in": "混合层"
+"node.keymix.input.mask_in": "遮罩"
+"node.premult.name": "预乘"
+"node.premult.input.tex_in": "输入"
+"node.premult.input.premult_channel_in": "通道"
+"node.unpremult.name": "反预乘"
+"node.unpremult.input.tex_in": "输入"
+"node.unpremult.input.unpremult_channel_in": "通道"
+"node.position.name": "位置"
+"node.position.input.tex_in": "输入"
+"node.position.input.offset_in": "平移"
+"node.mirror.name": "镜像"
+"node.mirror.input.tex_in": "输入"
+"node.mirror.input.horizontal_in": "水平（翻转）"
+"node.mirror.input.vertical_in": "垂直（翻转）"
+"node.checkerboard.name": "棋盘格"
+"node.checkerboard.input.base_in": "基础"
+"node.checkerboard.input.size_in": "盒子尺寸"
+"node.checkerboard.input.color1_in": "颜色 1"
+"node.checkerboard.input.color2_in": "颜色 2"
+"node.colorbars.name": "彩条"
+"node.colorbars.input.base_in": "基础"
+"node.colorbars.input.standard_in": "标准"
+"node.ramp.name": "渐变"
+"node.ramp.input.base_in": "基础"
+"node.ramp.input.point0_in": "点 0"
+"node.ramp.input.point1_in": "点 1"
+"node.ramp.input.color0_in": "颜色 0"
+"node.ramp.input.color1_in": "颜色 1"
+"node.multicam.name": "多机位"
+"node.multicam.input.current_in": "当前"
+"node.multicam.input.sources_in": "源"
+"node.multicam.input.sequence_in": "序列"
+"node.multicam.input.sequence_type_in": "序列类型"
 "status.ready": "就绪"
 "status.cache": "缓存:已启用"
 "status.proxy.on": "代理：开"
@@ -479,6 +853,8 @@
 "node.category.group": "组"
 "project.context.new": "新建"
 "project.new_sequence": "新建序列"
+"project.add_text_footage": "添加文本素材"
+"timeline.context.add_adjustment_layer": "添加调整图层"
 "project.context.reveal_in_finder": "在 Finder 中显示"
 "project.context.replace_footage": "替换素材"
 "project.context.rename": "重命名"
diff --git a/crates/oak-app/src/app.rs b/crates/oak-app/src/app.rs
index dd5092189..a86c40d45 100644
--- a/crates/oak-app/src/app.rs
+++ b/crates/oak-app/src/app.rs
@@ -655,6 +655,23 @@ impl<E: AppEngine> OakApp<E> {
 		)
 		.detach();
 
+		// The project explorer's 添加文本素材 button creates a text
+		// generator bin entry in the open project.
+		cx.subscribe(
+			&panels.project,
+			|this,
+			 _panel,
+			 _event: &crate::panels::project_explorer::NewTextFootageRequested,
+			 cx| {
+				this.engine.update(cx, |engine, cx| {
+					if let Err(e) = engine.create_text_footage(cx) {
+						println!("[shell] add text footage: {e}");
+					}
+				});
+			},
+		)
+		.detach();
+
 		// A footage drop onto a timeline with no open sequence goes through
 		// the probe-vs-manual choice before any sequence is created.
 		cx.subscribe(
@@ -5448,6 +5465,46 @@ mod tests {
 		assert_eq!(drops[0].time, gpui::timeline::Frame(50));
 	}
 
+	/// The project explorer's 添加文本素材 button creates a text generator
+	/// bin entry: the click emits `NewTextFootageRequested`, the shell's
+	/// subscription calls `AppEngine::create_text_footage`, and the bin
+	/// lists the new entry (labelled 文本, the shared node label) beside
+	/// the demo rows.
+	#[gpui::test]
+	async fn clicking_add_text_footage_creates_a_text_entry(cx: &mut TestAppContext) {
+		let _guard = crate::i18n::lang_test_lock()
+			.lock()
+			.unwrap_or_else(|e| e.into_inner());
+		let (window, root) = mock_shell(cx);
+		let mut vcx = VisualTestContext::from_window(window.into(), cx);
+
+		let button = vcx
+			.debug_bounds("project-add-text-footage")
+			.expect("the explorer's 添加文本素材 button is rendered");
+		vcx.simulate_click(button.center(), gpui::Modifiers::none());
+		drop(vcx);
+		cx.run_until_parked();
+
+		let created = cx.read(|app| root.read(app).engine.read(app).text_footages().to_vec());
+		assert_eq!(
+			created,
+			vec![100],
+			"the click reached the engine exactly once"
+		);
+		let listed = cx.read(|app| {
+			use gpui_widgets::project_explorer::ProjectDataSource as _;
+			root.read(app)
+				.engine
+				.read(app)
+				.roots()
+				.iter()
+				.any(|e| {
+					e.id == 100 && e.name == crate::oakui::graphops::TEXT_FOOTAGE_LABEL
+				})
+		});
+		assert!(listed, "the created text entry is listed in the bin");
+	}
+
 	/// The project explorer lists root-level imported footage in BOTH views:
 	/// the tree shows the entry as a row, and after switching to the icon
 	/// grid the same entry (plus folder children) appears as icons.
diff --git a/crates/oak-app/src/i18n.rs b/crates/oak-app/src/i18n.rs
index 8a80a3d68..20da1be37 100644
--- a/crates/oak-app/src/i18n.rs
+++ b/crates/oak-app/src/i18n.rs
@@ -141,6 +141,27 @@ pub fn tr(key: &'static str) -> &'static str {
 	key
 }
 
+/// Translates a key built at runtime — the built-in effect and parameter
+/// names, whose keys are `node.<type-id suffix>.name` and
+/// `node.<type-id suffix>.input.<input id>` (see
+/// [`oakui::effectchain::addable_effects`](crate::oakui::effectchain::addable_effects)).
+///
+/// Same lookup as [`tr`] — the active language, then `en-US` — with the
+/// node layer's own English name as the last resort: `tr` takes
+/// `&'static str` keys and returns the key itself on a miss, which for a
+/// formatted key would put a raw `node.…` identifier in the UI.
+pub fn tr_or(key: &str, fallback: &str) -> String {
+	let tables = tables().read().unwrap_or_else(|e| e.into_inner());
+	let current = language_code();
+	if let Some(value) = tables.get(&current).and_then(|t| t.get(key)) {
+		return value.to_string();
+	}
+	if let Some(value) = tables.get(FALLBACK_CODE).and_then(|t| t.get(key)) {
+		return value.to_string();
+	}
+	fallback.to_string()
+}
+
 /// The keys the gpui widget crates localize through
 /// `gpui_widgets::i18n` (viewer transport labels, effect-stack empty state).
 pub const WIDGET_KEYS: &[&str] = &[
@@ -358,6 +379,41 @@ mod tests {
 		}
 	}
 
+	/// Every built-in node type the factory registers carries a
+	/// `node.<type-id suffix>.name` entry in the reference pack *and* in
+	/// the translation: the effect library and the inspector build that key
+	/// at runtime, so a node without an entry would show its English name
+	/// in every language — and one whose zh value repeats the English name
+	/// was never translated at all. The matching parameter keys are held by
+	/// `effectchain`'s own test against the live parameter list.
+	#[test]
+	fn every_builtin_node_name_is_translated() {
+		let tables = tables().read().unwrap_or_else(|e| e.into_inner());
+		let en = tables.get("en-US").expect("en-US pack");
+		let zh = tables.get("zh-CN").expect("zh-CN pack");
+		let mut nodes = 0usize;
+		for meta in oak_node::factory::Factory::global().entries() {
+			let suffix = meta.type_id.rsplit('.').next().unwrap_or(meta.type_id);
+			let key = format!("node.{suffix}.name");
+			let en_value = en
+				.get(&key)
+				.unwrap_or_else(|| panic!("{key} missing from the en-US pack"));
+			assert_eq!(
+				*en_value, meta.name,
+				"{key} does not carry the node's English name"
+			);
+			let zh_value = zh
+				.get(&key)
+				.unwrap_or_else(|| panic!("{key} missing from the zh-CN pack"));
+			assert_ne!(zh_value, en_value, "{key} is not translated ({en_value:?})");
+			nodes += 1;
+		}
+		assert!(
+			nodes >= 60,
+			"the factory registers the built-in node types ({nodes})"
+		);
+	}
+
 	/// `tr` falls back from the active language to English and then to
 	/// the key itself.
 	#[test]
diff --git a/crates/oak-app/src/oakui/effectchain.rs b/crates/oak-app/src/oakui/effectchain.rs
index d617b57e0..fb6e4b824 100644
--- a/crates/oak-app/src/oakui/effectchain.rs
+++ b/crates/oak-app/src/oakui/effectchain.rs
@@ -93,11 +93,96 @@ pub fn is_enabled(g: &Graph, node: NodeId) -> bool {
 		.unwrap_or(false)
 }
 
+/// The built-in effect group keys, in presentation order.
+///
+/// The effect library and the inspector's add menu head the built-in
+/// effects with these groups, in this order; `general` collects every
+/// category without a group of its own and sorts last. The keys are
+/// **language-pack keys**, never display strings: a header's label is
+/// [`group_label`]'s `effect_library.group.<key>` lookup.
+pub const GROUP_KEYS: &[&str] = &[
+	"color",
+	"filter",
+	"distort",
+	"keying",
+	"generator",
+	"math",
+	"time",
+	"channel",
+	"transition",
+	"general",
+];
+
+/// The group key of a node [`Category`](oak_node::node::Category).
+///
+/// The category enum mirrors the C++ `Node::CategoryID`, which is finer
+/// grained than the create menu: outputs, plain effects, inputs, OpenFX
+/// plugins, timeline and group nodes have no group of their own and land
+/// in `general`. `time` / `channel` / `transition` are part of the key
+/// set (the language packs carry their labels) but no category maps to
+/// them yet.
+pub fn category_group_key(category: oak_node::node::Category) -> &'static str {
+	use oak_node::node::Category;
+	match category {
+		Category::Color => "color",
+		Category::Filter => "filter",
+		Category::Distort => "distort",
+		Category::Keying => "keying",
+		Category::Generator => "generator",
+		Category::Math => "math",
+		_ => "general",
+	}
+}
+
+/// The display label of an effect-group key: the localized
+/// `effect_library.group.<key>` string for the built-in keys, and the key
+/// itself for OpenFX sub-categories (their labels come from the plugin
+/// descriptor, not from the language packs).
+pub fn group_label(group_key: &str) -> String {
+	match group_key {
+		"color" => crate::i18n::tr("effect_library.group.color"),
+		"filter" => crate::i18n::tr("effect_library.group.filter"),
+		"distort" => crate::i18n::tr("effect_library.group.distort"),
+		"keying" => crate::i18n::tr("effect_library.group.keying"),
+		"generator" => crate::i18n::tr("effect_library.group.generator"),
+		"math" => crate::i18n::tr("effect_library.group.math"),
+		"time" => crate::i18n::tr("effect_library.group.time"),
+		"channel" => crate::i18n::tr("effect_library.group.channel"),
+		"transition" => crate::i18n::tr("effect_library.group.transition"),
+		"general" => crate::i18n::tr("effect_library.group.general"),
+		// OpenFX sub-categories (and any key a future pack ships without a
+		// label): shown verbatim rather than as a raw pack key.
+		other => other,
+	}
+	.to_string()
+}
+
+/// The language-pack key of a built-in node's display name, built at
+/// runtime: `node.<type-id suffix>.name`, the suffix being the type id's
+/// last dotted segment (`org.olivevideoeditor.Olive.colorcorrect` →
+/// `colorcorrect`).
+fn node_name_key(type_id: &str) -> String {
+	format!("node.{}.name", type_suffix(type_id))
+}
+
+/// The language-pack key of a built-in node input's display name:
+/// `node.<type-id suffix>.input.<input id>`.
+fn input_name_key(type_id: &str, input_id: &str) -> String {
+	format!("node.{}.input.{input_id}", type_suffix(type_id))
+}
+
+/// The last dotted segment of a node type id — the language-pack key's
+/// `<type-id suffix>`.
+fn type_suffix(type_id: &str) -> &str {
+	type_id.rsplit('.').next().unwrap_or(type_id)
+}
+
 /// The effect types the user can add to a clip's chain — the built-in
 /// factory entries flagged `video_effect` and not hidden from the create
-/// menu (no group), plus every runtime-registered OpenFX plugin entry
-/// (grouped by its sub-category: Filter / Generator / Transition /
-/// General — the C++ `factorymenu.cpp` OpenFX branch).
+/// menu (grouped by their first category, see [`category_group_key`]),
+/// plus every runtime-registered OpenFX plugin entry (grouped by its
+/// sub-category: Filter / Generator / Transition / General — the C++
+/// `factorymenu.cpp` OpenFX branch).
 pub fn addable_effects() -> Vec<super::engine::EffectEntry> {
 	use super::engine::EffectEntry;
 	use oak_node::node::flags as node_flags;
@@ -115,10 +200,18 @@ pub fn addable_effects() -> Vec<super::engine::EffectEntry> {
 			} else {
 				meta.name.to_string()
 			};
+			// The library shows the pack's `node.<type-id suffix>.name`; a
+			// type the packs don't know keeps the node's own English name.
+			let name = crate::i18n::tr_or(&node_name_key(meta.type_id), &name);
+			let group = meta
+				.categories
+				.first()
+				.map(|category| category_group_key(*category))
+				.unwrap_or("general");
 			out.push(EffectEntry {
 				type_id: meta.type_id.to_string(),
 				name,
-				group: None,
+				group: Some(group.to_string()),
 			});
 		}
 	}
@@ -136,18 +229,30 @@ pub fn addable_effects() -> Vec<super::engine::EffectEntry> {
 			group: Some(meta.sub_category),
 		});
 	}
-	// Default presentation order: built-ins first, then the OpenFX
-	// sub-category groups alphabetically; names alphabetical (folded)
-	// within each group.
+	// Default presentation order: built-in groups first, in GROUP_KEYS
+	// order (`general` last), then the OpenFX sub-category groups
+	// alphabetically; names alphabetical (folded) within each group.
 	out.sort_by(|a, b| {
 		let ga = a.group.as_deref().unwrap_or("");
 		let gb = b.group.as_deref().unwrap_or("");
-		ga.cmp(gb)
-			.then_with(|| a.name.to_lowercase().cmp(&b.name.to_lowercase()))
+		match (group_rank(ga), group_rank(gb)) {
+			(Some(ra), Some(rb)) => ra.cmp(&rb),
+			(Some(_), None) => std::cmp::Ordering::Less,
+			(None, Some(_)) => std::cmp::Ordering::Greater,
+			(None, None) => ga.cmp(gb),
+		}
+		.then_with(|| a.name.to_lowercase().cmp(&b.name.to_lowercase()))
 	});
 	out
 }
 
+/// The presentation rank of a built-in group key, or `None` for a key
+/// outside [`GROUP_KEYS`] (an OpenFX sub-category, which the sort orders
+/// after every built-in group).
+fn group_rank(group: &str) -> Option<usize> {
+	GROUP_KEYS.iter().position(|key| *key == group)
+}
+
 // ---------------------------------------------------------------------------
 // OFX parameter data model (stage 6b)
 // ---------------------------------------------------------------------------
@@ -176,6 +281,13 @@ pub fn effect_params(
 	use oak_node::value::ValueType;
 
 	let entry = g.get(node)?;
+	// The pack keys describe the built-in nodes only: an OpenFX plugin's
+	// labels come from the plugin (and the OFX translation pass), and a
+	// plugin type id happens to end in a built-in type's suffix, the
+	// built-in name would otherwise leak into it.
+	let builtin = oak_node::factory::Factory::global()
+		.find(entry.behavior.type_id())
+		.is_some();
 	let mut out = Vec::new();
 	for input in &entry.core.inputs {
 		// Clip/texture/sample/matrix inputs are graph connections or
@@ -202,6 +314,18 @@ pub fn effect_params(
 		} else {
 			input.display_name.clone()
 		};
+		// Built-in nodes label their inputs from the pack instead
+		// (`node.<type-id suffix>.input.<input id>`), so an input the
+		// behavior names in English — or leaves at its raw input id —
+		// follows the active language. The behavior's name is the fallback.
+		let display_name = if builtin {
+			crate::i18n::tr_or(
+				&input_name_key(entry.behavior.type_id(), &input.id),
+				&display_name,
+			)
+		} else {
+			display_name
+		};
 		// Combo options: built-in nodes carry them on the behavior (C++
 		// `set_combo_box_strings`); OFX plugin params already carry the
 		// ("combo_option", _) properties from the translation pass.
@@ -215,6 +339,35 @@ pub fn effect_params(
 				));
 			}
 		}
+		// The v3 text node's `font_family_in` is a str-combo whose options
+		// the node layer cannot supply: oak-node deliberately links no font
+		// or shaping crate, so the app's text backend enumerates the system
+		// families and the facade injects them here (the same
+		// `("combo_option", Text)` shape the OFX translation pass uses).
+		if input.id == "font_family_in" && !properties.iter().any(|(k, _)| k == "combo_option") {
+			for name in super::textengine::font_families() {
+				properties.push((
+					"combo_option".to_string(),
+					oak_node::value::NodeValue::Text(name),
+				));
+			}
+		}
+		// The v3 text node's editable text inputs hold more than one line
+		// (the node shapes each `\n`-separated line of the string), and the
+		// inspector renders them as multi-line fields so the user can type
+		// line breaks. The node layer cannot say so — an input carries no
+		// presentation metadata beyond its properties — so the facade marks
+		// the multi-line text ids here, the same way it injects the font
+		// families above.
+		if input.value_type == ValueType::Text
+			&& matches!(input.id.as_str(), "plain_text_in" | "text_in")
+			&& !properties.iter().any(|(k, _)| k == "multiline")
+		{
+			properties.push((
+				"multiline".to_string(),
+				oak_node::value::NodeValue::Boolean(true),
+			));
+		}
 		out.push(super::engine::EffectParam {
 			input_id: input.id.clone(),
 			display_name,
@@ -780,21 +933,146 @@ mod tests {
 		}
 	}
 
-	/// The effect-library grouping: every addable effect is either an
-	/// ungrouped built-in or an OpenFX entry with one of the four
-	/// sub-categories.
+	/// Every built-in entry carries its category's group key — one of
+	/// [`GROUP_KEYS`], never a display string — and the shipped nodes use
+	/// more than one group (a single "general" for everything would mean
+	/// the category mapping is dead).
+	#[test]
+	fn addable_effects_group_builtins_by_category() {
+		let factory = oak_node::factory::Factory::global();
+		let mut groups: Vec<&'static str> = Vec::new();
+		let mut builtins = 0usize;
+		for entry in addable_effects() {
+			if factory.find(&entry.type_id).is_none() {
+				continue; // an OpenFX plugin entry
+			}
+			builtins += 1;
+			let group = entry
+				.group
+				.as_deref()
+				.unwrap_or_else(|| panic!("built-in {} has no group", entry.type_id));
+			let key = GROUP_KEYS.iter().copied().find(|key| *key == group).unwrap_or_else(|| {
+				panic!("built-in {} has the unknown group {group:?}", entry.type_id)
+			});
+			if !groups.contains(&key) {
+				groups.push(key);
+			}
+		}
+		assert!(builtins > 0, "the factory ships built-in video effects");
+		assert!(
+			groups.len() > 1,
+			"built-ins spread over several groups: {groups:?}"
+		);
+	}
+
+	/// OpenFX plugin entries keep their plugin-supplied sub-category
+	/// (Filter / Generator / Transition / General), which never collides
+	/// with a built-in group key.
 	#[test]
 	fn addable_effects_group_openfx_by_sub_category() {
+		let factory = oak_node::factory::Factory::global();
 		for entry in addable_effects() {
-			if let Some(group) = entry.group {
+			if factory.find_dynamic(&entry.type_id).is_none() {
+				continue; // a built-in
+			}
+			let group = entry
+				.group
+				.as_deref()
+				.expect("an OpenFX entry carries a sub-category");
+			assert!(
+				["Filter", "Generator", "Transition", "General"].contains(&group),
+				"unexpected OpenFX sub-category {group:?}"
+			);
+			assert!(
+				!GROUP_KEYS.contains(&group),
+				"OpenFX sub-category {group:?} collides with a built-in group key"
+			);
+		}
+	}
+
+	/// The sort the library and the inspector render: built-in groups in
+	/// [`GROUP_KEYS`] order, then the OpenFX sub-categories, names folded
+	/// alphabetically inside a group.
+	#[test]
+	fn addable_effects_sort_builtins_first() {
+		let entries = addable_effects();
+		let rank = |entry: &crate::oakui::engine::EffectEntry| {
+			entry
+				.group
+				.as_deref()
+				.and_then(group_rank)
+				.unwrap_or(GROUP_KEYS.len())
+		};
+		for pair in entries.windows(2) {
+			let (a, b) = (&pair[0], &pair[1]);
+			let (ra, rb) = (rank(a), rank(b));
+			if ra != rb {
 				assert!(
-					["Filter", "Generator", "Transition", "General"].contains(&group.as_str()),
-					"unexpected OpenFX sub-category {group:?}"
+					ra < rb,
+					"group order: {a:?} ({ra}) before {b:?} ({rb})"
+				);
+			} else if a.group == b.group {
+				assert!(
+					a.name.to_lowercase() <= b.name.to_lowercase(),
+					"name order inside {:?}: {a:?} before {b:?}",
+					a.group
+				);
+			} else {
+				// Two OpenFX sub-categories (rank only separates built-ins).
+				assert!(
+					a.group.as_deref().unwrap_or("") <= b.group.as_deref().unwrap_or(""),
+					"OpenFX sub-category order: {a:?} before {b:?}"
 				);
 			}
 		}
 	}
 
+	/// The category → group-key mapping covers every `Category` variant
+	/// and falls back to `general`.
+	#[test]
+	fn category_group_key_maps_every_category() {
+		use oak_node::node::Category;
+		let all = [
+			Category::Output,
+			Category::Effect,
+			Category::Generator,
+			Category::Input,
+			Category::Math,
+			Category::Color,
+			Category::Distort,
+			Category::Filter,
+			Category::Keying,
+			Category::OpenFx,
+			Category::Timeline,
+			Category::Group,
+		];
+		for category in all {
+			assert!(
+				GROUP_KEYS.contains(&category_group_key(category)),
+				"{category:?} maps outside GROUP_KEYS"
+			);
+		}
+		assert_eq!(category_group_key(Category::Color), "color");
+		assert_eq!(category_group_key(Category::Keying), "keying");
+		assert_eq!(category_group_key(Category::Effect), "general");
+		assert_eq!(category_group_key(Category::OpenFx), "general");
+	}
+
+	/// A group label is the language pack's string for the built-in keys —
+	/// never the raw key — and passes OpenFX sub-categories through.
+	#[test]
+	fn group_label_localizes_keys_and_passes_sub_categories_through() {
+		for key in GROUP_KEYS {
+			let label = group_label(key);
+			assert!(
+				!label.starts_with("effect_library.group."),
+				"language pack entry missing for the {key} group (got {label:?})"
+			);
+		}
+		assert_eq!(group_label("Filter"), "Filter");
+		assert_eq!(group_label("Sapphire"), "Sapphire");
+	}
+
 	/// `set_input_value` writes the standard value undoably and rejects
 	/// unknown input ids.
 	#[test]
@@ -1072,4 +1350,180 @@ mod tests {
 		};
 		assert_eq!(ui_section_of(&grouped), Some(("Basic".into(), "Main".into())));
 	}
+
+	/// W6: the v3 text node is a footage entry now (the project panel's
+	/// "add text footage" button and the timeline drop), so the effect
+	/// library / add menus no longer offer it. A text3 node that is in a
+	/// project anyway — a legacy chain, or the one the footage path
+	/// creates — must keep chaining and keep exposing its params.
+	#[test]
+	fn text3_is_not_addable_but_a_chain_node_still_reports_its_params() {
+		let _g = stack_lock();
+		oak_undo::global::clear().unwrap();
+
+		const TEXT3: &str = "org.olivevideoeditor.Olive.text3";
+		assert!(
+			!addable_effects().iter().any(|e| e.type_id == TEXT3),
+			"text3 left the effect add menus"
+		);
+
+		// `insert` resolves through the factory (which still registers the
+		// type) rather than the addable table: legacy chains and the
+		// footage path keep working.
+		let (project, host) = project_with_clip();
+		let node = insert(&project, host, 0, TEXT3).expect("chain a legacy text3 effect");
+		assert_eq!(chain(&lock(&project).graph, host), vec![node]);
+
+		let params = effect_params(&lock(&project).graph, node).expect("the node exposes params");
+		let plain_text = params
+			.iter()
+			.find(|p| p.input_id == "plain_text_in")
+			.expect("the structured text parameter survives");
+		assert_eq!(plain_text.value_type, oak_node::value::ValueType::Text);
+		// The facade marks the multi-line text ids (the inspector renders a
+		// multi-line field off this property); the node itself carries no
+		// presentation metadata.
+		assert!(
+			plain_text
+				.properties
+				.iter()
+				.any(|(k, v)| k == "multiline"
+					&& matches!(v, oak_node::value::NodeValue::Boolean(true))),
+			"plain_text_in is marked as multi-line"
+		);
+		assert!(params.iter().any(|p| p.input_id == "font_size_in"));
+		assert!(params.iter().any(|p| p.input_id == "outline_enabled_in"));
+		// The hidden legacy HTML input stays a param *input* in the node,
+		// never in the inspector.
+		assert!(!params.iter().any(|p| p.input_id == "text_in"));
+
+		oak_undo::global::clear().unwrap();
+	}
+
+	/// The effect library and the inspector name built-in effects and their
+	/// parameters from runtime-built pack keys (`node.<type-id suffix>.name`
+	/// / `node.<type-id suffix>.input.<input id>`): switching the active
+	/// language renames the whole library and parameter list, while a node
+	/// the packs don't know keeps the name it ships with.
+	#[test]
+	fn effect_and_parameter_names_follow_the_language_pack() {
+		let _lang = crate::i18n::lang_test_lock()
+			.lock()
+			.unwrap_or_else(|e| e.into_inner());
+		let _g = stack_lock();
+		oak_undo::global::clear().unwrap();
+		crate::i18n::set_language_code("zh-CN");
+
+		// Every addable built-in shows its pack name, never the factory's
+		// English one.
+		let factory = oak_node::factory::Factory::global();
+		let mut builtins = 0usize;
+		for entry in addable_effects() {
+			let Some(meta) = factory.find(&entry.type_id) else {
+				continue; // an OpenFX plugin entry: the plugin names it
+			};
+			assert_ne!(
+				entry.name, meta.name,
+				"{} is still English in zh-CN",
+				entry.type_id
+			);
+			builtins += 1;
+		}
+		assert!(builtins > 0, "the factory ships addable built-ins");
+		let name_of = |type_id: &str| {
+			addable_effects()
+				.into_iter()
+				.find(|e| e.type_id == type_id)
+				.unwrap_or_else(|| panic!("{type_id} is addable"))
+				.name
+		};
+		assert_eq!(
+			name_of("org.olivevideoeditor.Olive.colorcorrect"),
+			"颜色校正"
+		);
+		assert_eq!(name_of("org.olivevideoeditor.Olive.position"), "位置");
+
+		// The inspector's parameter labels follow the same keys.
+		let (project, host) = project_with_clip();
+		let node = insert(&project, host, 0, "org.olivevideoeditor.Olive.colorcorrect")
+			.expect("chain Color Correct");
+		let params = effect_params(&lock(&project).graph, node).expect("the node exposes params");
+		let label = |input_id: &str| {
+			params
+				.iter()
+				.find(|p| p.input_id == input_id)
+				.unwrap_or_else(|| panic!("{input_id} is a parameter"))
+				.display_name
+				.clone()
+		};
+		assert_eq!(label("saturation_in"), "饱和度");
+		assert_eq!(label("gamma_in"), "伽马");
+		assert_eq!(label("gain_in"), "增益");
+
+		crate::i18n::set_language_code("en-US");
+		oak_undo::global::clear().unwrap();
+	}
+
+	/// The pack coverage behind the lookup above: every name a built-in
+	/// node gives one of its inputs has a `node.<type-id suffix>.input.<id>`
+	/// entry, so no parameter label falls back to English. Inputs the
+	/// behavior does not name (`enabled_in`, `use_args_in`) carry no entry —
+	/// their label is the raw input id, and the enabled input never renders.
+	#[test]
+	fn every_builtin_parameter_name_has_a_pack_entry() {
+		let _lang = crate::i18n::lang_test_lock()
+			.lock()
+			.unwrap_or_else(|e| e.into_inner());
+		crate::i18n::set_language_code("zh-CN");
+		let mut missing = Vec::new();
+		let mut named = 0usize;
+		for meta in oak_node::factory::Factory::global().entries() {
+			let (core, behavior) = (meta.create)();
+			for input in &core.inputs {
+				let english = behavior.input_name(&input.id);
+				if english == input.id {
+					continue;
+				}
+				if crate::i18n::tr_or(&input_name_key(meta.type_id, &input.id), english) == english {
+					missing.push(format!("node.{}.input.{}", type_suffix(meta.type_id), input.id));
+				}
+				named += 1;
+			}
+		}
+		assert!(missing.is_empty(), "no pack entry for {missing:?}");
+		assert!(named >= 250, "the built-in nodes name their inputs ({named})");
+		crate::i18n::set_language_code("en-US");
+	}
+
+	/// The v3 text node's `font_family_in` is a str-combo the node layer
+	/// cannot populate (oak-node links no font crate): `effect_params`
+	/// injects the app's font backend families as its combo options, sorted
+	/// and de-duplicated.
+	#[test]
+	fn text_font_family_combo_is_injected_from_the_font_backend() {
+		let _g = stack_lock();
+		let (project, _clip) = project_with_clip();
+		let node = {
+			let mut g = lock(&project);
+			let (core, behavior) = oak_node::factory::Factory::global()
+				.create_any("org.olivevideoeditor.Olive.text3")
+				.expect("the factory registers the v3 text node");
+			g.graph.add_node(core, behavior)
+		};
+
+		let params = effect_params(&lock(&project).graph, node).expect("the node exposes params");
+		let param = params
+			.iter()
+			.find(|p| p.input_id == "font_family_in")
+			.expect("the font family combo is a parameter");
+		assert_eq!(param.value_type, oak_node::value::ValueType::StrCombo);
+
+		let options = combo_options(param);
+		assert!(!options.is_empty(), "the font backend found no families");
+		assert_eq!(options, crate::oakui::textengine::font_families());
+		let mut sorted = options.clone();
+		sorted.sort();
+		sorted.dedup();
+		assert_eq!(options, sorted);
+	}
 }
diff --git a/crates/oak-app/src/oakui/engine.rs b/crates/oak-app/src/oakui/engine.rs
index f69e575ba..e7ad5a600 100644
--- a/crates/oak-app/src/oakui/engine.rs
+++ b/crates/oak-app/src/oakui/engine.rs
@@ -339,6 +339,20 @@ pub trait AppEngine:
 	/// it).
 	fn remove_track(&mut self, index: usize, cx: &mut Context<Self>);
 
+	/// Adds an adjustment layer to the display track at `index`, starting
+	/// at `frame` and five seconds long (undoable where the backend
+	/// supports it). Returns a user-facing error message on failure.
+	/// Default: unsupported.
+	fn add_adjustment_layer(
+		&mut self,
+		index: usize,
+		frame: Frame,
+		cx: &mut Context<Self>,
+	) -> Result<(), String> {
+		let _ = (index, frame, cx);
+		Err("add adjustment layer not supported".into())
+	}
+
 	/// Sets the row height of every timeline track (timeline toolbar).
 	fn set_track_height(&mut self, height: Pixels, cx: &mut Context<Self>);
 
@@ -1071,6 +1085,15 @@ pub trait AppEngine:
 		Err("no project open".to_string())
 	}
 
+	/// Creates a text generator in the open project's root folder (the
+	/// project explorer's "添加文本素材" action) and returns its entry id.
+	/// The bin entry is NOT a sequence: a double-click selects it like
+	/// footage, and the timeline drop places a generator clip.
+	fn create_text_footage(&mut self, cx: &mut Context<Self>) -> Result<u64, String> {
+		let _ = cx;
+		Err("no project open".to_string())
+	}
+
 	/// Applies new name/format/interlaced parameters to the sequence `id`.
 	fn update_sequence_parameters(
 		&mut self,
@@ -1213,6 +1236,18 @@ pub trait AppEngine:
 		let _ = (clips, cx);
 	}
 
+	/// Adds a transition of the configured default length (the
+	/// `DefaultTransitionLength` preference, seconds) at every contiguous
+	/// seam around `clips` — the clip menu's Default Transition item and the
+	/// Ctrl+Shift+D action (the C++ `timeline::add_default_transition`).
+	/// One undoable "Add Transition" entry; returns how many transitions
+	/// were created, or `Err` when no selected seam accepts one.
+	fn add_default_transition(
+		&mut self,
+		clips: Vec<ClipId>,
+		cx: &mut Context<Self>,
+	) -> Result<usize, String>;
+
 	// -------------------------------------------------------------------
 	// Multi-camera (the C++ MulticamWidget / timeline Multi-Cam menu):
 	// detection state for the panel, angle-frame rendering, the timeline
diff --git a/crates/oak-app/src/oakui/graphops.rs b/crates/oak-app/src/oakui/graphops.rs
index a5a74666c..b9d1f59c4 100644
--- a/crates/oak-app/src/oakui/graphops.rs
+++ b/crates/oak-app/src/oakui/graphops.rs
@@ -33,10 +33,10 @@ use std::path::Path;
 use std::sync::{Arc, Mutex, MutexGuard};
 
 use oak_core::{Rational, TimeRange};
-use oak_node::block::ClipBlockBehavior;
+use oak_node::block::{AdjustmentBlockBehavior, BlockCore, ClipBlockBehavior};
 use oak_node::folder::FolderBehavior;
 use oak_node::footage::FootageBehavior;
-use oak_node::graph::Graph;
+use oak_node::graph::{Graph, NodeEntry};
 use oak_node::id::NodeId;
 use oak_node::project::Project;
 use oak_node::sequence::SequenceBehavior;
@@ -296,6 +296,96 @@ pub fn create_folder(project: &ProjectRef, name: &str) -> Result<NodeId, String>
 	Ok(id)
 }
 
+/// The stable type id of the text generator node (the redesigned `textv3`,
+/// "org.olivevideoeditor.Olive.text3"): the bin entry the project panel's
+/// "添加文本素材" action creates, and the type the timeline drop routes to
+/// the generator-clip path instead of the footage path.
+pub const TEXT_FOOTAGE_TYPE_ID: &str = "org.olivevideoeditor.Olive.text3";
+
+/// The bin label a created text generator starts with (the plan's "文本"
+/// entry name). Fixed, not localized at creation time: it is the node's
+/// label, which the user renames like any other bin entry.
+pub const TEXT_FOOTAGE_LABEL: &str = "文本";
+
+/// Create a text generator in the project's root folder (the project
+/// panel's "添加文本素材" action). ONE undo row covers the node AND its
+/// mount: the undo detaches the node from the root folder and removes it
+/// from the graph (its entry stays inside the command), so no orphan node
+/// outlives the undo — unlike [`create_folder`], which leaves its node in
+/// the graph and only undoes the mount. The redo re-inserts the entry,
+/// which keeps its stable id while its slot is free (the
+/// [`Graph::add_entry`] contract).
+pub fn create_text_footage_node(project: &ProjectRef) -> Result<NodeId, String> {
+	let root = {
+		let guard = lock(project);
+		if !guard.root.valid() {
+			return Err("the project has no root folder".to_string());
+		}
+		guard.root
+	};
+	let Some((mut core, behavior)) =
+		oak_node::factory::Factory::global().create_any(TEXT_FOOTAGE_TYPE_ID)
+	else {
+		return Err(format!(
+			"the text node \"{TEXT_FOOTAGE_TYPE_ID}\" is not registered"
+		));
+	};
+	core.label = TEXT_FOOTAGE_LABEL.to_string();
+	// The eager insert gives the caller an id for the bin entry before the
+	// command is built; the undo parks the entry here between runs.
+	let id = {
+		let mut guard = lock(project);
+		guard.graph.add_node(core, behavior)
+	};
+	let detached: Arc<Mutex<Option<NodeEntry>>> = Arc::new(Mutex::new(None));
+	let (redo_detached, undo_detached) = (detached.clone(), detached.clone());
+	let (redo_project, undo_project) = (project.clone(), project.clone());
+	let redo = move || {
+		let mut guard = lock(&redo_project);
+		if !guard.graph.is_valid(id) {
+			let entry = redo_detached
+				.lock()
+				.unwrap_or_else(|e| e.into_inner())
+				.take();
+			if let Some(entry) = entry {
+				guard.graph.add_entry(entry, id);
+			}
+		}
+		if let Some(entry) = guard.graph.get_mut(id) {
+			entry.core.bin_folder = Some(root);
+		}
+		if let Some(entry) = guard.graph.get_mut(root) {
+			if let Some(folder) = entry
+				.behavior
+				.as_any_mut()
+				.and_then(|a| a.downcast_mut::<FolderBehavior>())
+			{
+				folder.add_child(id);
+			}
+		}
+	};
+	let undo = move || {
+		let mut guard = lock(&undo_project);
+		if let Some(entry) = guard.graph.get_mut(root) {
+			if let Some(folder) = entry
+				.behavior
+				.as_any_mut()
+				.and_then(|a| a.downcast_mut::<FolderBehavior>())
+			{
+				folder.remove_child(id);
+			}
+		}
+		if let Some(entry) = guard.graph.get_mut(id) {
+			entry.core.bin_folder = None;
+		}
+		let taken = guard.graph.take_node(id);
+		*undo_detached.lock().unwrap_or_else(|e| e.into_inner()) = taken;
+	};
+	let cmd = oak_undo::undocommand::UndoCommand::from_closures(redo, undo);
+	push_command(cmd, "Add Text Footage")?;
+	Ok(id)
+}
+
 /// Every folder node in the graph, in arena order.
 pub fn folder_ids(p: &Project) -> Vec<NodeId> {
 	p.graph
@@ -455,7 +545,7 @@ pub fn sequence_length(g: &Graph, seq: NodeId) -> Rational {
 }
 
 /// Borrow a block's core (any block kind).
-fn block_core(entry: &oak_node::graph::NodeEntry) -> Option<&oak_node::block::BlockCore> {
+fn block_core(entry: &oak_node::graph::NodeEntry) -> Option<&BlockCore> {
 	let any = entry.behavior.as_any()?;
 	if let Some(c) = any.downcast_ref::<ClipBlockBehavior>() {
 		return Some(&c.core);
@@ -463,8 +553,77 @@ fn block_core(entry: &oak_node::graph::NodeEntry) -> Option<&oak_node::block::Bl
 	if let Some(g) = any.downcast_ref::<oak_node::block::GapBlockBehavior>() {
 		return Some(&g.core);
 	}
-	any.downcast_ref::<oak_node::block::TransitionBlockBehavior>()
-		.map(|t| &t.core)
+	if let Some(t) = any.downcast_ref::<oak_node::block::TransitionBlockBehavior>() {
+		return Some(&t.core);
+	}
+	any.downcast_ref::<AdjustmentBlockBehavior>()
+		.map(|a| &a.core)
+}
+
+/// Borrow a block's core mutably (any block kind).
+fn block_core_mut(entry: &mut oak_node::graph::NodeEntry) -> Option<&mut BlockCore> {
+	let any = entry.behavior.as_any_mut()?;
+	if any.is::<ClipBlockBehavior>() {
+		Some(
+			&mut any
+				.downcast_mut::<ClipBlockBehavior>()
+				.expect("checked above")
+				.core,
+		)
+	} else if any.is::<oak_node::block::GapBlockBehavior>() {
+		Some(
+			&mut any
+				.downcast_mut::<oak_node::block::GapBlockBehavior>()
+				.expect("checked above")
+				.core,
+		)
+	} else if any.is::<oak_node::block::TransitionBlockBehavior>() {
+		Some(
+			&mut any
+				.downcast_mut::<oak_node::block::TransitionBlockBehavior>()
+				.expect("checked above")
+				.core,
+		)
+	} else if any.is::<AdjustmentBlockBehavior>() {
+		Some(
+			&mut any
+				.downcast_mut::<AdjustmentBlockBehavior>()
+				.expect("checked above")
+				.core,
+		)
+	} else {
+		None
+	}
+}
+
+/// Borrow the block core at `node` (any block kind).
+pub fn block_core_of(g: &Graph, node: NodeId) -> Option<&BlockCore> {
+	block_core(g.get(node)?)
+}
+
+/// Mutate the block core at `node` (any block kind). Returns `false` and
+/// leaves the graph untouched when `node` is not a block.
+fn with_block_core_mut(g: &mut Graph, node: NodeId, f: impl FnOnce(&mut BlockCore)) -> bool {
+	let Some(entry) = g.get_mut(node) else {
+		return false;
+	};
+	let Some(core) = block_core_mut(entry) else {
+		return false;
+	};
+	f(core);
+	true
+}
+
+/// Whether `node` is a clip-class block: a clip or an adjustment layer.
+/// Gaps and transitions are blocks too, but not clips.
+fn is_timeline_clip(g: &Graph, node: NodeId) -> bool {
+	let Some(entry) = g.get(node) else {
+		return false;
+	};
+	let Some(any) = entry.behavior.as_any() else {
+		return false;
+	};
+	any.is::<ClipBlockBehavior>() || any.is::<AdjustmentBlockBehavior>()
 }
 
 /// The sequence playhead (rational seconds).
@@ -592,28 +751,29 @@ pub fn track_ids(g: &Graph, seq: NodeId, kind: TrackType) -> Vec<NodeId> {
 		.unwrap_or_default()
 }
 
-/// The clip blocks of a track (gaps skipped), in timeline order.
+/// The clip blocks of a track, in timeline order. Gaps and transitions are
+/// not clips and are skipped; adjustment layers are.
 pub fn clip_ids(g: &Graph, track: NodeId) -> Vec<NodeId> {
 	track_behavior(g, track)
 		.map(|t| {
 			t.blocks
 				.iter()
 				.copied()
-				.filter(|&b| clip_behavior(g, b).is_some())
+				.filter(|&b| is_timeline_clip(g, b))
 				.collect()
 		})
 		.unwrap_or_default()
 }
 
-/// A clip's timeline range and media in-point (rational seconds).
+/// A block's timeline range and media in-point (rational seconds).
 pub fn clip_range(g: &Graph, clip: NodeId) -> Option<(Rational, Rational, Rational)> {
-	let c = clip_behavior(g, clip)?;
-	Some((c.core.in_(), c.core.out(), c.core.media_in))
+	let c = block_core_of(g, clip)?;
+	Some((c.in_(), c.out(), c.media_in))
 }
 
-/// The clip's owning track.
+/// The block's owning track.
 pub fn clip_track(g: &Graph, clip: NodeId) -> Option<NodeId> {
-	clip_behavior(g, clip)?.core.track
+	block_core_of(g, clip)?.track
 }
 
 /// The first footage node feeding `id` (upstream BFS over input edges),
@@ -1505,6 +1665,401 @@ pub fn place_footage_clip(
 	Ok(clip)
 }
 
+/// Add an adjustment layer to `track` spanning `[in_ts, out_ts)` (frame
+/// timestamps in the sequence's timebase): an [`AdjustmentBlockBehavior`]
+/// block with no footage, so it renders its effect chain over the clips
+/// underneath it. Only video tracks take adjustment layers.
+///
+/// Undoable "Add Adjustment Layer" (the block is created in the project
+/// and placed by the module's `TrackPlaceBlockCommand`); on undo the
+/// block is detached from the track but left as an orphan node in the
+/// project graph (the module command's documented behavior, same as
+/// [`place_footage_clip`]).
+pub fn create_adjustment_layer(
+	p: &ProjectRef,
+	seq: NodeId,
+	track: NodeId,
+	in_ts: i64,
+	out_ts: i64,
+) -> Result<NodeId, String> {
+	if in_ts < 0 || out_ts <= in_ts {
+		return Err("invalid adjustment layer range (need 0 <= in < out)".to_string());
+	}
+	let (list, track_index, in_r, out_r) = {
+		let g = lock(p);
+		let behavior = track_behavior(&g.graph, track)
+			.ok_or_else(|| "the track is not in the project".to_string())?;
+		if behavior.kind != TrackType::Video {
+			return Err("adjustment layers are only supported on video tracks".to_string());
+		}
+		let tb = sequence_time_base(&g.graph, seq)
+			.ok_or_else(|| "sequence has no valid frame rate".to_string())?;
+		let list = behavior
+			.track_list
+			.ok_or_else(|| "the track is not in a track list".to_string())?;
+		let index = track_list_behavior(&g.graph, list)
+			.and_then(|l| l.tracks.iter().position(|&t| t == track))
+			.ok_or_else(|| "the track is not in its list".to_string())?;
+		(list, index, ts_to_rational(in_ts, tb), ts_to_rational(out_ts, tb))
+	};
+
+	// The block itself: no label, no color override, no media wiring —
+	// just its timeline span. Its placement below is the undoable part.
+	let block = {
+		let mut g = lock(p);
+		let (core, behavior) = oak_node::block::adjustment_create();
+		let block = g.graph.add_node(core, behavior);
+		with_block_core_mut(&mut g.graph, block, |core| {
+			core.range = TimeRange::new(in_r, out_r);
+		});
+		block
+	};
+
+	push(
+		oak_timeline::undopointer::TrackPlaceBlockCommand::new(
+			node_ref(p, list),
+			track_index as i32,
+			node_ref(p, block),
+			in_r,
+		)
+		.to_command(),
+		"Add Adjustment Layer",
+	)?;
+	Ok(block)
+}
+
+/// The transition block joining `prev` to `next` on `track`, when one is
+/// already wired between them.
+fn transition_between(g: &Graph, track: NodeId, prev: NodeId, next: NodeId) -> Option<NodeId> {
+	track_behavior(g, track)
+		.map(|t| t.blocks.clone())
+		.unwrap_or_default()
+		.into_iter()
+		.find(|&b| {
+			g.get(b)
+				.and_then(|e| e.behavior.as_any())
+				.map(|a| a.is::<oak_node::block::TransitionBlockBehavior>())
+				.unwrap_or(false)
+				&& g.connected_output(b, oak_node::block::transition_input::OUT_BLOCK, -1)
+					== Some(prev)
+				&& g.connected_output(b, oak_node::block::transition_input::IN_BLOCK, -1)
+					== Some(next)
+		})
+}
+
+/// Build (without pushing) the commands that create a transition block at
+/// the seam between `prev` and `next` and wire both sides into it.
+///
+/// The block covers `[out(prev) - half, out(prev) + half]`: `in_offset`
+/// reaches into `prev` and `out_offset` into `next` (the module's
+/// transition geometry), both `half` for a symmetric default transition.
+/// `prev` and `next` must be clip blocks listed back to back on one track
+/// and must touch exactly. Everything is validated before the block node
+/// is created, so a rejected seam leaves the graph untouched.
+///
+/// The block enters the track through the module's non-destructive
+/// `TrackInsertBlockAfterCommand`; `TrackPlaceBlockCommand` is unusable
+/// here because its redo ripple-removes the span the new block covers,
+/// which would trim away both clips the transition straddles.
+fn transition_commands(
+	p: &ProjectRef,
+	prev: NodeId,
+	next: NodeId,
+	half: Rational,
+) -> Result<(NodeId, Vec<oak_undo::undocommand::UndoCommand>), String> {
+	if half <= Rational::new(0, 1) {
+		return Err("transition: the length must be positive".to_string());
+	}
+	let (seam, owner) = {
+		let g = lock(p);
+		if clip_behavior(&g.graph, prev).is_none() {
+			return Err("transition: the outgoing block is not a clip".to_string());
+		}
+		if clip_behavior(&g.graph, next).is_none() {
+			return Err("transition: the incoming block is not a clip".to_string());
+		}
+		let Some(prev_core) = block_core_of(&g.graph, prev) else {
+			return Err("transition: the outgoing clip has no block core".to_string());
+		};
+		let Some(next_core) = block_core_of(&g.graph, next) else {
+			return Err("transition: the incoming clip has no block core".to_string());
+		};
+		if next_core.in_() != prev_core.out() {
+			return Err("transition: the clips are not contiguous at the seam".to_string());
+		}
+		let Some(track) = prev_core.track else {
+			return Err("transition: the outgoing clip is not on a track".to_string());
+		};
+		if next_core.track != Some(track) {
+			return Err("transition: the clips are not on the same track".to_string());
+		}
+		let Some(blocks) = track_behavior(&g.graph, track).map(|t| t.blocks.clone()) else {
+			return Err("transition: the track is not in the project".to_string());
+		};
+		let adjacent = blocks
+			.iter()
+			.position(|&b| b == prev)
+			.and_then(|i| blocks.get(i + 1).copied())
+			== Some(next);
+		if !adjacent {
+			return Err("transition: the clips are not adjacent on the track".to_string());
+		}
+		if transition_between(&g.graph, track, prev, next).is_some() {
+			return Err("transition: the seam already carries a transition".to_string());
+		}
+		(prev_core.out(), track)
+	};
+
+	let block = {
+		let mut g = lock(p);
+		let (core, behavior) = oak_node::block::transition_create();
+		let id = g.graph.add_node(core, behavior);
+		let Some(t) = g
+			.graph
+			.get_mut(id)
+			.and_then(|e| e.behavior.as_any_mut())
+			.and_then(|a| a.downcast_mut::<oak_node::block::TransitionBlockBehavior>())
+		else {
+			return Err("transition: could not create the block".to_string());
+		};
+		t.core.range = TimeRange::new(seam - half, seam + half);
+		t.in_offset = half;
+		t.out_offset = half;
+		id
+	};
+	let insert = oak_timeline::undotrack::TrackInsertBlockAfterCommand::new(
+		node_ref(p, owner),
+		node_ref(p, block),
+		Some(node_ref(p, prev)),
+	)
+	.to_command();
+	let from = connect_command(p, prev, block, oak_node::block::transition_input::OUT_BLOCK)?;
+	let to = connect_command(p, next, block, oak_node::block::transition_input::IN_BLOCK)?;
+	Ok((block, vec![insert, from, to]))
+}
+
+/// Create a transition at the seam between two contiguous clips and push
+/// it as one "Add Transition" undo row (the timeline's wedge drag path).
+pub fn add_transition_at_seam(
+	p: &ProjectRef,
+	prev: NodeId,
+	next: NodeId,
+	half: Rational,
+) -> Result<NodeId, String> {
+	let (block, commands) = transition_commands(p, prev, next, half)?;
+	push_multi(commands, "Add Transition")?;
+	Ok(block)
+}
+
+/// Add a default transition at every contiguous seam around the selected
+/// clips (the Ctrl+Shift+D action / the timeline clip-menu item).
+///
+/// `half` is half of the default transition length: the transition spans
+/// `half` on either side of its seam. Every selected clip contributes the
+/// seam before it and the seam after it; a neighbor only counts when it is
+/// a clip block (not a gap, a transition or an adjustment layer) that
+/// touches the clip exactly, and a seam shared by two selected clips is
+/// built once. Seams that cannot take a transition (an existing
+/// transition, a non-contiguous neighbor) are skipped. The whole batch is
+/// one "Add Transition" undo row; a batch that builds nothing is an error,
+/// so the action never leaves an empty row. Returns the number of
+/// transitions created.
+///
+/// `seq` scopes the action: a selected block whose track list belongs to
+/// another sequence is ignored.
+pub fn add_default_transition(
+	p: &ProjectRef,
+	seq: NodeId,
+	clip_blocks: &[NodeId],
+	half: Rational,
+) -> Result<usize, String> {
+	let lists = {
+		let g = lock(p);
+		let Some(s) = sequence_behavior(&g.graph, seq) else {
+			return Err("transition: the sequence is not in the project".to_string());
+		};
+		s.track_lists.clone()
+	};
+
+	let mut seams: Vec<(NodeId, NodeId)> = Vec::new();
+	for &clip in clip_blocks {
+		let candidates: Vec<(NodeId, NodeId)> = {
+			let g = lock(p);
+			if clip_behavior(&g.graph, clip).is_none() {
+				continue;
+			}
+			let Some(track) = clip_track(&g.graph, clip) else {
+				continue;
+			};
+			let Some(list) = track_behavior(&g.graph, track).and_then(|t| t.track_list) else {
+				continue;
+			};
+			if !lists.contains(&list) {
+				continue;
+			}
+			let Some(blocks) = track_behavior(&g.graph, track).map(|t| t.blocks.clone()) else {
+				continue;
+			};
+			let Some(index) = blocks.iter().position(|&b| b == clip) else {
+				continue;
+			};
+			let range = block_core_of(&g.graph, clip).map(|c| (c.in_(), c.out()));
+			let mut sides = Vec::new();
+			if let Some(&prev) = index.checked_sub(1).and_then(|i| blocks.get(i)) {
+				if clip_behavior(&g.graph, prev).is_some()
+					&& block_core_of(&g.graph, prev).map(|c| c.out()) == range.map(|r| r.0)
+				{
+					sides.push((prev, clip));
+				}
+			}
+			if let Some(&next) = blocks.get(index + 1) {
+				if clip_behavior(&g.graph, next).is_some()
+					&& block_core_of(&g.graph, next).map(|c| c.in_()) == range.map(|r| r.1)
+				{
+					sides.push((clip, next));
+				}
+			}
+			sides
+		};
+		for seam in candidates {
+			if !seams.contains(&seam) {
+				seams.push(seam);
+			}
+		}
+	}
+	if seams.is_empty() {
+		return Err("transition: no selected clip borders another clip".to_string());
+	}
+
+	let mut commands = Vec::new();
+	let mut created = 0usize;
+	for (prev, next) in seams {
+		if let Ok((_, children)) = transition_commands(p, prev, next, half) {
+			commands.extend(children);
+			created += 1;
+		}
+	}
+	if created == 0 {
+		return Err("transition: no contiguous seam accepts a transition".to_string());
+	}
+	push_multi(commands, "Add Transition")?;
+	Ok(created)
+}
+
+/// The transition block drawn on one edge of `clip`, when its neighbour on
+/// that side is a transition wired into the clip.
+///
+/// The timeline widget addresses a wedge through the clip it is drawn on:
+/// the start wedge belongs to the transition *before* the clip (the clip
+/// feeds its `in_block_in`) and the end wedge to the one *after* it
+/// (`out_block_in`).
+pub fn transition_of_clip(g: &Graph, clip: NodeId, start_edge: bool) -> Option<NodeId> {
+	let track = clip_track(g, clip)?;
+	let blocks = track_behavior(g, track)?.blocks.clone();
+	let index = blocks.iter().position(|&b| b == clip)?;
+	let candidate = if start_edge {
+		blocks.get(index.checked_sub(1)?).copied()?
+	} else {
+		blocks.get(index + 1).copied()?
+	};
+	let input = if start_edge {
+		oak_node::block::transition_input::IN_BLOCK
+	} else {
+		oak_node::block::transition_input::OUT_BLOCK
+	};
+	if g.connected_output(candidate, input, -1) == Some(clip) {
+		Some(candidate)
+	} else {
+		None
+	}
+}
+
+/// Resize one wedge of the transition attached to `clip` (the timeline
+/// widget's wedge drag, its `TransitionChanged` event).
+///
+/// `start_edge` names the wedge by the clip it is drawn on: the start wedge
+/// is the incoming transition's `out_offset`, the end wedge the outgoing
+/// one's `in_offset`. The width is clamped to `frame..=clip length` — one
+/// frame at the smallest, and never wider than the clip it eats into — and
+/// written through [`oak_timeline::undogeneral::TransitionSetOffsetsCommand`],
+/// so one "Transition Length" undo row covers the drag. The seam does not
+/// move. Returns the width actually written.
+pub fn set_transition_length(
+	p: &ProjectRef,
+	clip: NodeId,
+	start_edge: bool,
+	frame: Rational,
+	new_length: Rational,
+) -> Result<Rational, String> {
+	let (transition, clip_length) = {
+		let g = lock(p);
+		let Some(transition) = transition_of_clip(&g.graph, clip, start_edge) else {
+			return Err("transition: the clip has no transition on that edge".to_string());
+		};
+		let Some(core) = block_core_of(&g.graph, clip) else {
+			return Err("transition: the clip has no block core".to_string());
+		};
+		(transition, core.length())
+	};
+	let clamped = new_length.max(frame).min(clip_length);
+	push_command(
+		oak_timeline::undogeneral::TransitionSetOffsetsCommand::new(
+			node_ref(p, transition),
+			if start_edge { None } else { Some(clamped) },
+			if start_edge { Some(clamped) } else { None },
+		)
+		.to_command(),
+		"Transition Length",
+	)?;
+	Ok(clamped)
+}
+
+/// The transition wedges of every clip on `track`, keyed by clip:
+/// `(start, end)` are the widths of the transition before the clip (drawn
+/// on its head) and after it (drawn on its tail); `None` when that side has
+/// no transition. Gaps and transitions themselves are not keyed.
+pub fn clip_transition_widths(
+	g: &Graph,
+	track: NodeId,
+) -> std::collections::HashMap<NodeId, (Option<Rational>, Option<Rational>)> {
+	let mut widths: std::collections::HashMap<NodeId, (Option<Rational>, Option<Rational>)> =
+		std::collections::HashMap::new();
+	let Some(blocks) = track_behavior(g, track).map(|t| t.blocks.clone()) else {
+		return widths;
+	};
+	for &block in &blocks {
+		if is_timeline_clip(g, block) {
+			widths.insert(block, (None, None));
+		}
+	}
+	for &block in &blocks {
+		let Some(transition) = g
+			.get(block)
+			.and_then(|e| e.behavior.as_any())
+			.and_then(|a| a.downcast_ref::<oak_node::block::TransitionBlockBehavior>())
+		else {
+			continue;
+		};
+		// A transition overlaps its neighbours: it reaches `in_offset` into
+		// the block before it and `out_offset` into the block after it.
+		if let Some(prev) =
+			g.connected_output(block, oak_node::block::transition_input::OUT_BLOCK, -1)
+		{
+			if let Some(pair) = widths.get_mut(&prev) {
+				pair.1 = Some(transition.in_offset);
+			}
+		}
+		if let Some(next) =
+			g.connected_output(block, oak_node::block::transition_input::IN_BLOCK, -1)
+		{
+			if let Some(pair) = widths.get_mut(&next) {
+				pair.0 = Some(transition.out_offset);
+			}
+		}
+	}
+	widths
+}
+
 /// Places ONE video clip fed by a sequence node — a nested-sequence clip
 /// (drag a sequence entry from the project explorer onto the timeline).
 /// The clip reads the sequence's output during playback; the sequence's
@@ -1585,6 +2140,85 @@ pub fn place_nested_sequence_clip(
 	Ok(clip)
 }
 
+/// Places ONE video clip fed by the text generator `text` (drag a text
+/// entry from the project explorer onto the timeline) — a generator clip,
+/// not a decode of media: the clip reads the generator's output through
+/// `tex_in`, so the text re-renders at whatever time the clip spans.
+/// Undoable as one "Add Clip". `in_ts`/`out_ts` are frame timestamps in
+/// the host sequence's frame-rate timebase; a clip dragged shorter or
+/// longer than the drop's default simply re-spans the same generator.
+pub fn place_text_clip(
+	p: &ProjectRef,
+	host_seq: NodeId,
+	text: NodeId,
+	track_index: usize,
+	in_ts: i64,
+	out_ts: i64,
+) -> Result<NodeId, String> {
+	if in_ts < 0 || out_ts <= in_ts {
+		return Err("invalid clip range (need 0 <= in < out)".to_string());
+	}
+	let (tb, list, label) = {
+		let g = lock(p);
+		if node_type_id(&g.graph, text) != TEXT_FOOTAGE_TYPE_ID {
+			return Err("the text entry is not in the project".to_string());
+		}
+		let tb = sequence_time_base(&g.graph, host_seq)
+			.ok_or_else(|| "host sequence has no valid frame rate".to_string())?;
+		let list = track_list_of(&g.graph, host_seq, TrackType::Video)
+			.ok_or_else(|| "host sequence has no video track list".to_string())?;
+		let label = node_label(&g.graph, text);
+		(tb, list, label)
+	};
+	let track_count = {
+		let g = lock(p);
+		track_list_behavior(&g.graph, list)
+			.map(|l| l.tracks.len())
+			.unwrap_or(0)
+	};
+	if track_index >= track_count {
+		return Err(format!(
+			"track index {track_index} out of range ({track_count} tracks)"
+		));
+	}
+
+	let in_r = ts_to_rational(in_ts, tb);
+	let out_r = ts_to_rational(out_ts, tb);
+	let length = out_r - in_r;
+
+	// The clip block, labelled after its generator and stamped with its
+	// creation-time color (the footage-clip convention). A generator has no
+	// media, so only the length is written; `TrackPlaceBlockCommand` sets
+	// the in-point from the drop.
+	let clip = oak_timeline::util::block_clip_create(p).id;
+	{
+		let mut g = lock(p);
+		let Some(entry) = g.graph.get_mut(clip) else {
+			return Err("clip node is not in the project".to_string());
+		};
+		entry.core.override_color = clip_color_index(clip);
+		entry.core.label = label;
+		if let Some(c) = entry
+			.behavior
+			.as_any_mut()
+			.and_then(|a| a.downcast_mut::<ClipBlockBehavior>())
+		{
+			c.core.set_length_and_media_in(length);
+		}
+	}
+
+	let place = oak_timeline::undopointer::TrackPlaceBlockCommand::new(
+		node_ref(p, list),
+		track_index as i32,
+		node_ref(p, clip),
+		in_r,
+	)
+	.to_command();
+	let edge = connect_command(p, text, clip, oak_node::block::clip_input::TEXTURE_INPUT)?;
+	push_multi(vec![place, edge], "Add Clip")?;
+	Ok(clip)
+}
+
 /// Place linked clips of one footage on several tracks in ONE undoable
 /// "Add Clip" entry (the NLE A/V-drop: a video-with-audio file lands as
 /// a video clip plus a linked audio clip). `placements` lists the
@@ -1762,38 +2396,27 @@ pub(crate) fn trim_command(
 	old: Rational,
 	new: Rational,
 ) -> oak_undo::undocommand::UndoCommand {
-	use oak_node::block::BlockCore;
 	let (p1, p2) = (p.clone(), p.clone());
 	oak_undo::undocommand::UndoCommand::from_closures(
 		move || {
-			let mut g = lock(&p1);
-			if let Some(c) = g
-				.graph
-				.get_mut(clip)
-				.and_then(|e| e.behavior.as_any_mut())
-				.and_then(|a| a.downcast_mut::<ClipBlockBehavior>())
-			{
+			let mut guard = lock(&p1);
+			with_block_core_mut(&mut guard.graph, clip, |core| {
 				if out_anchored {
-					BlockCore::set_length_and_media_out(&mut c.core, new);
+					core.set_length_and_media_out(new);
 				} else {
-					BlockCore::set_length_and_media_in(&mut c.core, new);
+					core.set_length_and_media_in(new);
 				}
-			}
+			});
 		},
 		move || {
-			let mut g = lock(&p2);
-			if let Some(c) = g
-				.graph
-				.get_mut(clip)
-				.and_then(|e| e.behavior.as_any_mut())
-				.and_then(|a| a.downcast_mut::<ClipBlockBehavior>())
-			{
+			let mut guard = lock(&p2);
+			with_block_core_mut(&mut guard.graph, clip, |core| {
 				if out_anchored {
-					BlockCore::set_length_and_media_out(&mut c.core, old);
+					core.set_length_and_media_out(old);
 				} else {
-					BlockCore::set_length_and_media_in(&mut c.core, old);
+					core.set_length_and_media_in(old);
 				}
-			}
+			});
 		},
 	)
 }
@@ -2164,26 +2787,12 @@ fn move_clip_to_track_commands(
 	let (p1, p2) = (p.clone(), p.clone());
 	let rehome = oak_undo::undocommand::UndoCommand::from_closures(
 		move || {
-			let mut g = lock(&p1);
-			if let Some(c) = g
-				.graph
-				.get_mut(clip)
-				.and_then(|e| e.behavior.as_any_mut())
-				.and_then(|a| a.downcast_mut::<ClipBlockBehavior>())
-			{
-				c.core.set_in(in_r);
-			}
+			let mut guard = lock(&p1);
+			with_block_core_mut(&mut guard.graph, clip, |core| core.set_in(in_r));
 		},
 		move || {
-			let mut g = lock(&p2);
-			if let Some(c) = g
-				.graph
-				.get_mut(clip)
-				.and_then(|e| e.behavior.as_any_mut())
-				.and_then(|a| a.downcast_mut::<ClipBlockBehavior>())
-			{
-				c.core.set_in(old_in);
-			}
+			let mut guard = lock(&p2);
+			with_block_core_mut(&mut guard.graph, clip, |core| core.set_in(old_in));
 		},
 	);
 	let place = oak_timeline::undopointer::TrackPlaceBlockCommand::new(
@@ -3390,4 +3999,616 @@ mod undo_cycle_ops_tests {
 		assert!(interlaced, "the interlaced flag updates too");
 		oak_undo::global::clear().unwrap();
 	}
+
+	// ---- adjustment layers ------------------------------------------------
+
+	/// Drop a fresh adjustment-layer block onto `track` spanning
+	/// `[in_ts, out_ts)` (timeline frames) and return its node id.
+	fn add_adjustment_layer(
+		p: &ProjectRef,
+		seq: NodeId,
+		track: NodeId,
+		in_ts: i64,
+		out_ts: i64,
+	) -> NodeId {
+		let tb = {
+			let g = lock(p);
+			sequence_time_base(&g.graph, seq).expect("the sequence has a timebase")
+		};
+		let in_r = ts_to_rational(in_ts, tb);
+		let block = {
+			let mut g = lock(p);
+			let (core, behavior) = oak_node::block::adjustment_create();
+			let id = g.graph.add_node(core, behavior);
+			let entry = g.graph.get_mut(id).expect("the new block");
+			let a = entry
+				.behavior
+				.as_any_mut()
+				.and_then(|a| a.downcast_mut::<AdjustmentBlockBehavior>())
+				.expect("an adjustment behavior");
+			a.core.range = TimeRange::new(in_r, ts_to_rational(out_ts, tb));
+			id
+		};
+		let (list, index) = {
+			let g = lock(p);
+			let list = track_behavior(&g.graph, track)
+				.and_then(|t| t.track_list)
+				.expect("the track has a list");
+			let index = track_list_behavior(&g.graph, list)
+				.and_then(|l| l.tracks.iter().position(|&t| t == track))
+				.expect("the track is in its list") as i32;
+			(list, index)
+		};
+		push(
+			oak_timeline::undopointer::TrackPlaceBlockCommand::new(
+				node_ref(p, list),
+				index,
+				node_ref(p, block),
+				in_r,
+			)
+			.to_command(),
+			"Add Adjustment Layer",
+		)
+		.expect("place the adjustment layer");
+		block
+	}
+
+	/// An adjustment layer is a timeline block: the generic clip queries
+	/// see it, its length counts toward the track, and the generic trim /
+	/// delete operations apply to it with converging undo-redo cycles.
+	#[test]
+	fn adjustment_layer_trim_and_delete_cycles_converge() {
+		let _g = test_lock();
+		oak_undo::global::clear().unwrap();
+		let project = create_project();
+		let seq = create_sequence(&project, "Adjustment Ops");
+		add_track(&project, seq, TrackType::Video).expect("add video track");
+		let track = {
+			let g = lock(&project);
+			track_ids(&g.graph, seq, TrackType::Video)[0]
+		};
+		let tb = {
+			let g = lock(&project);
+			sequence_time_base(&g.graph, seq).expect("timebase")
+		};
+		let adj = add_adjustment_layer(&project, seq, track, 50, 125);
+
+		{
+			let g = lock(&project);
+			assert!(
+				clip_ids(&g.graph, track).contains(&adj),
+				"an adjustment layer counts as a timeline block"
+			);
+			assert_eq!(
+				clip_range(&g.graph, adj),
+				Some((
+					ts_to_rational(50, tb),
+					ts_to_rational(125, tb),
+					Rational::new(0, 1)
+				)),
+				"the block reports its timeline range and a zero media in"
+			);
+			assert_eq!(
+				clip_track(&g.graph, adj),
+				Some(track),
+				"the block knows its track"
+			);
+		}
+		assert_eq!(
+			oak_timeline::util::track_length(&node_ref(&project, track)),
+			ts_to_rational(125, tb),
+			"the track length counts the leading gap plus the adjustment layer"
+		);
+
+		// Out-trim: the in point stays, the out point moves.
+		trim_clip(&project, adj, 50, 100).expect("trim the adjustment layer");
+		{
+			let g = lock(&project);
+			let (in_r, out_r, _) = clip_range(&g.graph, adj).expect("range");
+			assert_eq!(in_r, ts_to_rational(50, tb));
+			assert_eq!(
+				out_r,
+				ts_to_rational(100, tb),
+				"the out-trim lands on the adjustment layer"
+			);
+		}
+		let post = snapshot(&project, seq);
+		cycle_assert(&project, seq, &post, "adjustment trim");
+		oak_undo::global::clear().unwrap();
+
+		// Delete: the block leaves the track (and the clip list), and the
+		// replaced span is a gap, not a block.
+		delete_clip(&project, adj).expect("delete the adjustment layer");
+		{
+			let g = lock(&project);
+			assert!(
+				!clip_ids(&g.graph, track).contains(&adj),
+				"the deleted block leaves the clip list"
+			);
+			assert_eq!(
+				clip_track(&g.graph, adj),
+				None,
+				"the removed block no longer names a track"
+			);
+			let blocks = track_behavior(&g.graph, track)
+				.map(|t| t.blocks.clone())
+				.unwrap_or_default();
+			assert_eq!(
+				blocks.len(),
+				0,
+				"the trailing block and the gap that preceded it are gone: {blocks:?}"
+			);
+		}
+		let post = snapshot(&project, seq);
+		cycle_assert(&project, seq, &post, "adjustment delete");
+		oak_undo::global::clear().unwrap();
+	}
+
+	/// The production entry point ([`create_adjustment_layer`]) lands the
+	/// block on the video track with the requested span (nothing wired to
+	/// it), the undo detaches it without removing the node and the redo
+	/// re-places the same block.
+	#[test]
+	fn create_adjustment_layer_places_on_video_track_and_undoes() {
+		let _g = test_lock();
+		oak_undo::global::clear().unwrap();
+		let project = create_project();
+		let seq = create_sequence(&project, "Adjustment UI");
+		add_track(&project, seq, TrackType::Video).expect("add video track");
+		let track = {
+			let g = lock(&project);
+			track_ids(&g.graph, seq, TrackType::Video)[0]
+		};
+		let tb = {
+			let g = lock(&project);
+			sequence_time_base(&g.graph, seq).expect("timebase")
+		};
+
+		let adj = create_adjustment_layer(&project, seq, track, 50, 125).expect("create");
+		{
+			let g = lock(&project);
+			assert_eq!(
+				clip_ids(&g.graph, track),
+				vec![adj],
+				"the adjustment layer is the track's only block"
+			);
+			assert_eq!(
+				clip_range(&g.graph, adj),
+				Some((
+					ts_to_rational(50, tb),
+					ts_to_rational(125, tb),
+					Rational::new(0, 1)
+				)),
+				"the block spans the requested range with a zero media in"
+			);
+			assert_eq!(
+				clip_track(&g.graph, adj),
+				Some(track),
+				"the block knows its track"
+			);
+			assert!(
+				g.graph
+					.get(adj)
+					.and_then(|e| e.behavior.as_any())
+					.map(|a| a.is::<AdjustmentBlockBehavior>())
+					.unwrap_or(false),
+				"the new block is an adjustment layer, not a clip"
+			);
+		}
+		assert_eq!(
+			oak_timeline::util::track_length(&node_ref(&project, track)),
+			ts_to_rational(125, tb),
+			"the track length counts the leading gap plus the layer"
+		);
+		let post = snapshot(&project, seq);
+		cycle_assert(&project, seq, &post, "create adjustment");
+
+		oak_undo::global::undo().unwrap();
+		{
+			let g = lock(&project);
+			assert_eq!(
+				clip_track(&g.graph, adj),
+				None,
+				"the undo detaches the block from its track"
+			);
+			assert!(
+				g.graph.is_valid(adj),
+				"the detached block stays in the graph as an orphan"
+			);
+		}
+		oak_undo::global::redo().unwrap();
+		{
+			let g = lock(&project);
+			assert_eq!(
+				clip_track(&g.graph, adj),
+				Some(track),
+				"the redo re-places the same block"
+			);
+		}
+		oak_undo::global::clear().unwrap();
+	}
+
+	/// Only video tracks take adjustment layers, and the span must be a
+	/// forward range of non-negative frames; a rejected call creates no
+	/// node and leaves the undo stack untouched.
+	#[test]
+	fn create_adjustment_layer_rejects_audio_tracks_and_bad_ranges() {
+		let _g = test_lock();
+		oak_undo::global::clear().unwrap();
+		let project = create_project();
+		let seq = create_sequence(&project, "Adjustment Rejects");
+		add_track(&project, seq, TrackType::Video).expect("add video track");
+		add_track(&project, seq, TrackType::Audio).expect("add audio track");
+		let (video, audio) = {
+			let g = lock(&project);
+			(
+				track_ids(&g.graph, seq, TrackType::Video)[0],
+				track_ids(&g.graph, seq, TrackType::Audio)[0],
+			)
+		};
+		let before = oak_undo::global::count().unwrap();
+
+		assert!(
+			create_adjustment_layer(&project, seq, audio, 0, 25).is_err(),
+			"an audio track takes no adjustment layer"
+		);
+		assert!(
+			create_adjustment_layer(&project, seq, video, 25, 25).is_err(),
+			"an empty range is rejected"
+		);
+		assert!(
+			create_adjustment_layer(&project, seq, video, 50, 25).is_err(),
+			"a reversed range is rejected"
+		);
+		assert!(
+			create_adjustment_layer(&project, seq, video, -1, 25).is_err(),
+			"a negative in point is rejected"
+		);
+		assert_eq!(
+			oak_undo::global::count().unwrap(),
+			before,
+			"a rejected call leaves the undo stack untouched"
+		);
+		{
+			let g = lock(&project);
+			assert!(
+				clip_ids(&g.graph, video).is_empty(),
+				"no block was created for the rejected calls"
+			);
+		}
+		oak_undo::global::clear().unwrap();
+	}
+
+	/// 添加文本素材: the helper creates a text generator mounted under the
+	/// root as ONE undoable row. The undo detaches it from the folder AND
+	/// removes the node from the graph (not a bare unmount), and the redo
+	/// restores it under the same id.
+	#[test]
+	fn create_text_footage_mounts_under_root_and_undoes() {
+		let _g = test_lock();
+		oak_undo::global::clear().unwrap();
+		let project = create_project();
+		let before = oak_undo::global::count().unwrap();
+		let text = create_text_footage_node(&project).expect("create text footage");
+		assert_eq!(
+			oak_undo::global::count().unwrap(),
+			before + 1,
+			"the whole creation is one undo row"
+		);
+		let root = {
+			let g = lock(&project);
+			assert_eq!(node_type_id(&g.graph, text), TEXT_FOOTAGE_TYPE_ID);
+			assert_eq!(node_label(&g.graph, text), TEXT_FOOTAGE_LABEL);
+			assert_eq!(
+				g.graph.get(text).map(|e| e.core.bin_folder),
+				Some(Some(g.root)),
+				"the text node names the root as its bin folder"
+			);
+			g.root
+		};
+		assert!(
+			root_children(&project).contains(&text),
+			"the text entry mounts under the root folder"
+		);
+
+		oak_undo::global::undo().unwrap();
+		assert!(
+			!root_children(&project).contains(&text),
+			"the undo unmounts the entry"
+		);
+		assert!(
+			!lock(&project).graph.is_valid(text),
+			"the undo removes the node itself — no orphan outlives it"
+		);
+
+		oak_undo::global::redo().unwrap();
+		assert!(
+			root_children(&project).contains(&text),
+			"the redo remounts the entry"
+		);
+		let g = lock(&project);
+		assert_eq!(
+			node_type_id(&g.graph, text),
+			TEXT_FOOTAGE_TYPE_ID,
+			"the restored node keeps its id and type"
+		);
+		assert_eq!(
+			g.graph.get(text).map(|e| e.core.bin_folder),
+			Some(Some(root)),
+			"the restored node names the root again"
+		);
+		drop(g);
+		oak_undo::global::clear().unwrap();
+	}
+
+	/// The text drop places a GENERATOR clip: the block spans the dropped
+	/// range on the target track and its `tex_in` reads the text node
+	/// instead of a decode. Non-text entries and out-of-range tracks are
+	/// rejected.
+	#[test]
+	fn place_text_clip_wires_the_generator_and_spans_the_range() {
+		let _g = test_lock();
+		oak_undo::global::clear().unwrap();
+		let project = create_project();
+		let seq = create_sequence(&project, "Text Drop");
+		let text = create_text_footage_node(&project).expect("create text footage");
+		let (tb, track) = {
+			let g = lock(&project);
+			let tb = sequence_time_base(&g.graph, seq).expect("time base");
+			let track = track_ids(&g.graph, seq, TrackType::Video)[0];
+			(tb, track)
+		};
+		let clip = place_text_clip(&project, seq, text, 0, 12, 137).expect("place the text clip");
+		{
+			let g = lock(&project);
+			let (in_r, out_r, _) = clip_range(&g.graph, clip).expect("clip range");
+			assert_eq!(in_r, ts_to_rational(12, tb));
+			assert_eq!(
+				out_r,
+				ts_to_rational(137, tb),
+				"the clip spans the dropped frames"
+			);
+			assert_eq!(
+				g.graph
+					.connected_output(clip, oak_node::block::clip_input::TEXTURE_INPUT, -1),
+				Some(text),
+				"the clip reads the generator through tex_in"
+			);
+			assert_eq!(
+				clip_track(&g.graph, clip),
+				Some(track),
+				"the clip lands on the target video track"
+			);
+		}
+		// Only the text generator takes the generator-clip path: a folder
+		// entry on the same call is rejected.
+		let folder = create_folder(&project, "Not Text").expect("create folder");
+		assert!(
+			place_text_clip(&project, seq, folder, 0, 0, 10).is_err(),
+			"a non-text node is not placeable as a text clip"
+		);
+		assert!(
+			place_text_clip(&project, seq, text, 9, 0, 10).is_err(),
+			"a track index past the video tracks is rejected"
+		);
+		oak_undo::global::clear().unwrap();
+	}
+
+	// ---- default transitions (Ctrl+Shift+D / clip menu) ---------------------
+
+	/// Two touching clips on the video track of a fresh project, plus the
+	/// track, the time base and the symmetric half length the transition
+	/// tests use.
+	fn two_touching_clips(
+		media: &std::path::Path,
+	) -> (ProjectRef, NodeId, NodeId, NodeId, NodeId, (i64, i64), Rational) {
+		let project = create_project();
+		let seq = create_sequence(&project, "Default Transition");
+		let footage = import_footage(&project, media).expect("import");
+		let a = place_footage_clip(&project, seq, footage, TrackType::Video, 0, 0, 10, 0)
+			.expect("place a");
+		let b = place_footage_clip(&project, seq, footage, TrackType::Video, 0, 10, 20, 0)
+			.expect("place b");
+		let (tb, track) = {
+			let g = lock(&project);
+			(
+				sequence_time_base(&g.graph, seq).expect("time base"),
+				track_ids(&g.graph, seq, TrackType::Video)[0],
+			)
+		};
+		let half = ts_to_rational(3, tb);
+		(project, seq, track, a, b, tb, half)
+	}
+
+	/// 编辑 → 设为默认转场: one transition per contiguous seam around the
+	/// selection — the single seam here, built once even though both clips
+	/// are selected — wired onto both clips as ONE undo row.
+	#[test]
+	fn default_transition_covers_the_selected_seams() {
+		let _g = test_lock();
+		oak_undo::global::clear().unwrap();
+		let media = std::env::temp_dir()
+			.join(format!("oak_default_transition_{}.mp4", std::process::id()));
+		oak_codec::testmedia::write_test_clip(&media, 64, 64, 10, 10).expect("generate");
+		let (project, seq, track, a, b, tb, half) = two_touching_clips(&media);
+
+		let before = oak_undo::global::count().unwrap();
+		let created = add_default_transition(&project, seq, &[a, b], half).expect("add transition");
+		assert_eq!(
+			created, 1,
+			"both clips name the same seam; it is built once"
+		);
+		assert_eq!(
+			oak_undo::global::count().unwrap(),
+			before + 1,
+			"the whole batch is one undo row"
+		);
+
+		let transition = {
+			let g = lock(&project);
+			let blocks = track_behavior(&g.graph, track)
+				.map(|t| t.blocks.clone())
+				.unwrap_or_default();
+			assert_eq!(blocks.len(), 3, "the transition joins the track: {blocks:?}");
+			let t = blocks[1];
+			let seam = ts_to_rational(10, tb);
+			let core = block_core_of(&g.graph, t).expect("transition core");
+			assert_eq!(
+				(core.in_(), core.out()),
+				(seam - half, seam + half),
+				"the transition straddles the seam"
+			);
+			let behavior = g
+				.graph
+				.get(t)
+				.and_then(|e| e.behavior.as_any())
+				.and_then(|a| a.downcast_ref::<oak_node::block::TransitionBlockBehavior>())
+				.expect("transition behavior");
+			assert_eq!(
+				(behavior.in_offset, behavior.out_offset),
+				(half, half),
+				"a default transition is symmetric"
+			);
+			assert_eq!(
+				g.graph.connected_output(t, oak_node::block::transition_input::OUT_BLOCK, -1),
+				Some(a),
+				"the previous clip feeds out_block_in"
+			);
+			assert_eq!(
+				g.graph.connected_output(t, oak_node::block::transition_input::IN_BLOCK, -1),
+				Some(b),
+				"the next clip feeds in_block_in"
+			);
+			// The wedges the widget paints: the earlier clip's tail is the
+			// transition's in_offset, the later clip's head its out_offset.
+			let widths = clip_transition_widths(&g.graph, track);
+			assert_eq!(
+				widths.get(&a),
+				Some(&(None, Some(half))),
+				"the head clip's tail wedge"
+			);
+			assert_eq!(
+				widths.get(&b),
+				Some(&(Some(half), None)),
+				"the tail clip's head wedge"
+			);
+			assert_eq!(transition_of_clip(&g.graph, b, true), Some(t));
+			assert_eq!(transition_of_clip(&g.graph, a, false), Some(t));
+			assert_eq!(
+				transition_of_clip(&g.graph, a, true),
+				None,
+				"the first clip has nothing before it"
+			);
+			assert_eq!(
+				transition_of_clip(&g.graph, b, false),
+				None,
+				"the last clip has nothing after it"
+			);
+			t
+		};
+
+		let post = snapshot(&project, seq);
+		cycle_assert(&project, seq, &post, "add default transition");
+
+		// The seam carries a transition now: nothing left to build, and the
+		// refused batch leaves no empty undo row behind.
+		let rows = oak_undo::global::count().unwrap();
+		assert!(
+			add_default_transition(&project, seq, &[a, b], half).is_err(),
+			"a seam that already carries a transition is skipped"
+		);
+		assert_eq!(oak_undo::global::count().unwrap(), rows);
+		{
+			let g = lock(&project);
+			assert!(g.graph.is_valid(transition), "the refused batch changed nothing");
+		}
+
+		oak_undo::global::clear().unwrap();
+		let _ = std::fs::remove_file(&media);
+	}
+
+	/// The wedge drag (`TransitionChanged`): the later clip's head wedge
+	/// resizes the transition's out_offset, the earlier clip's tail wedge its
+	/// in_offset, the seam stays put, and a too-long request is clamped to
+	/// the dragged clip's length.
+	#[test]
+	fn transition_length_drag_moves_one_wedge_and_keeps_the_seam() {
+		let _g = test_lock();
+		oak_undo::global::clear().unwrap();
+		let media =
+			std::env::temp_dir().join(format!("oak_transition_drag_{}.mp4", std::process::id()));
+		oak_codec::testmedia::write_test_clip(&media, 64, 64, 10, 10).expect("generate");
+		let (project, seq, track, a, b, tb, half) = two_touching_clips(&media);
+		add_default_transition(&project, seq, &[a, b], half).expect("add transition");
+
+		let frame = ts_to_rational(1, tb);
+		let seam = ts_to_rational(10, tb);
+		let offsets = |p: &ProjectRef, t: NodeId| {
+			let g = lock(p);
+			g.graph
+				.get(t)
+				.and_then(|e| e.behavior.as_any())
+				.and_then(|a| a.downcast_ref::<oak_node::block::TransitionBlockBehavior>())
+				.map(|t| (t.in_offset, t.out_offset))
+				.expect("transition behavior")
+		};
+		let transition = {
+			let g = lock(&project);
+			transition_of_clip(&g.graph, b, true).expect("the head transition of b")
+		};
+
+		// Growing the head wedge of the later clip (edge = Start) only
+		// touches the out offset.
+		let longer = ts_to_rational(5, tb);
+		let applied = set_transition_length(&project, b, true, frame, longer).expect("grow");
+		assert_eq!(applied, longer);
+		assert_eq!(offsets(&project, transition), (half, longer));
+		{
+			let g = lock(&project);
+			let core = block_core_of(&g.graph, transition).expect("transition core");
+			assert_eq!((core.in_(), core.out()), (seam - half, seam + longer));
+			assert_eq!(
+				clip_range(&g.graph, a).map(|r| r.1),
+				Some(seam),
+				"the seam does not move"
+			);
+		}
+		let post = snapshot(&project, seq);
+		cycle_assert(&project, seq, &post, "transition length");
+
+		// A request longer than the dragged clip is clamped to its length.
+		let applied =
+			set_transition_length(&project, b, true, frame, ts_to_rational(9_999, tb)).expect("clamp");
+		assert_eq!(
+			applied,
+			ts_to_rational(10, tb),
+			"clamped to the later clip's length"
+		);
+
+		// The other side: the earlier clip's tail wedge (edge = End) drives
+		// the in offset.
+		let applied = set_transition_length(&project, a, false, frame, ts_to_rational(6, tb))
+			.expect("grow in");
+		assert_eq!(applied, ts_to_rational(6, tb));
+		assert_eq!(
+			offsets(&project, transition),
+			(ts_to_rational(6, tb), ts_to_rational(10, tb))
+		);
+		let post = snapshot(&project, seq);
+		cycle_assert(&project, seq, &post, "transition length (in side)");
+
+		// A clip without a transition on that edge is rejected, not silently
+		// rewired.
+		assert!(
+			set_transition_length(&project, b, false, frame, half).is_err(),
+			"the last clip has no tail transition"
+		);
+		assert!(
+			set_transition_length(&project, a, true, frame, half).is_err(),
+			"the first clip has no head transition"
+		);
+
+		oak_undo::global::clear().unwrap();
+		let _ = std::fs::remove_file(&media);
+	}
 }
diff --git a/crates/oak-app/src/oakui/mock.rs b/crates/oak-app/src/oakui/mock.rs
index 441cadc54..7ac8e7286 100644
--- a/crates/oak-app/src/oakui/mock.rs
+++ b/crates/oak-app/src/oakui/mock.rs
@@ -516,6 +516,14 @@ pub struct MockEngine {
 	/// the applied (id, track kind, track index, start frame) — drives
 	/// app-level tests of the explorer→timeline drag).
 	footage_drops: Vec<MockFootageDrop>,
+	/// Text-generator bin entries created via
+	/// [`AppEngine::create_text_footage`] since creation, as entry ids (mock
+	/// state; the mock's [`ProjectDataSource::roots`] lists them beside the
+	/// demo folders and files — drives app-level tests of the "添加文本素材"
+	/// action).
+	text_footages: Vec<u64>,
+	/// Id allocator for the text entries above.
+	next_text_footage_id: u64,
 	/// The fake project library the project manager browses (M13 D4): an
 	/// in-memory row set the library trait methods operate on, so the app
 	/// flow (list / open / create / rename / duplicate / delete / import /	/// export) is testable without a database.
@@ -811,6 +819,8 @@ impl MockEngine {
 			cpu_frame_cache: Mutex::new(HashMap::new()),
 			imported_footage: Vec::new(),
 			footage_drops: Vec::new(),
+			text_footages: Vec::new(),
+			next_text_footage_id: 100,
 			library: demo_library(),
 			next_library_id: 100,
 			library_opened: Vec::new(),
@@ -1347,6 +1357,49 @@ impl AppEngine for MockEngine {
 		cx.notify();
 	}
 
+	fn add_adjustment_layer(
+		&mut self,
+		index: usize,
+		frame: Frame,
+		cx: &mut Context<Self>,
+	) -> Result<(), String> {
+		let Some(kind) = self.tracks.get(index).map(|t| t.kind) else {
+			return Err(format!("no track at index {index}"));
+		};
+		if kind != TrackKind::Video {
+			return Err("adjustment layers are only supported on video tracks".to_string());
+		}
+		// The demo's canned layer: five seconds from the clicked frame,
+		// colored apart from the demo clips so the panel tests can tell it
+		// apart at a glance.
+		let fps = self.frame_rate();
+		let length = (5.0 * fps.num as f64 / fps.den.max(1) as f64)
+			.round()
+			.max(1.0) as i64;
+		let start = Frame(frame.0.max(0));
+		let clip = MockClip {
+			id: ClipId(self.next_mock_clip_id()),
+			range: FrameRange::new(start, Frame(start.0 + length)),
+			media_in: Frame(0),
+			label: "调整图层".into(),
+			color: Hsla {
+				h: 0.55,
+				s: 0.55,
+				l: 0.5,
+				a: 1.0,
+			},
+		};
+		let track = &mut self.tracks[index];
+		let position = track
+			.clips
+			.iter()
+			.position(|c| c.range.start.0 > start.0)
+			.unwrap_or(track.clips.len());
+		track.clips.insert(position, clip);
+		cx.notify();
+		Ok(())
+	}
+
 	fn set_track_height(&mut self, height: Pixels, cx: &mut Context<Self>) {
 		self.set_track_height(height, cx);
 	}
@@ -1405,17 +1458,163 @@ impl AppEngine for MockEngine {
 	}
 
 	fn effect_params(&self, effect: EffectId) -> Option<Vec<crate::oakui::engine::EffectParam>> {
-		// A canned text parameter behind a sentinel id, for the params
-		// view's text-field tests.
+		// A canned parameter list behind a sentinel id, mirroring the shape
+		// of the v3 text node's inputs (ids, value types, combo options and
+		// min bounds included) so the params view's tests can cover every
+		// control the text inspector builds. Three entries carry the HIDDEN
+		// flag: a real node's hidden inputs are dropped by the facade's
+		// `effect_params` before they get here, and the view must skip them
+		// on its own too (plugin and mock engines can hand them over).
 		(effect.0 == 900).then(|| {
-			vec![crate::oakui::engine::EffectParam {
-				input_id: "text_in".to_string(),
-				display_name: "Text".to_string(),
-				value_type: oak_node::value::ValueType::Text,
-				value: oak_node::value::NodeValue::Text("<p>engine text</p>".to_string()),
-				flags: 0,
-				properties: Vec::new(),
-			}]
+			use oak_node::input::flags::{ARRAY, HIDDEN};
+			use oak_node::value::{NodeValue, ValueType};
+
+			fn param(
+				input_id: &str,
+				display_name: &str,
+				value_type: ValueType,
+				value: NodeValue,
+				flags: u32,
+				properties: Vec<(&str, NodeValue)>,
+			) -> crate::oakui::engine::EffectParam {
+				crate::oakui::engine::EffectParam {
+					input_id: input_id.to_string(),
+					display_name: display_name.to_string(),
+					value_type,
+					value,
+					flags,
+					properties: properties
+						.into_iter()
+						.map(|(k, v)| (k.to_string(), v))
+						.collect(),
+				}
+			}
+
+			vec![
+				param(
+					"pos_in",
+					"Position",
+					ValueType::Vec2,
+					NodeValue::Vec2([0.0, 0.0]),
+					0,
+					Vec::new(),
+				),
+				param(
+					"size_in",
+					"Size",
+					ValueType::Vec2,
+					NodeValue::Vec2([400.0, 300.0]),
+					0,
+					vec![("min", NodeValue::Vec2([0.0, 0.0]))],
+				),
+				param(
+					"plain_text_in",
+					"Text",
+					ValueType::Text,
+					NodeValue::Text("文本\nsecond line".to_string()),
+					0,
+					vec![("multiline", NodeValue::Boolean(true))],
+				),
+				param(
+					"text_in",
+					"Legacy text",
+					ValueType::Text,
+					NodeValue::Text("<p>engine text</p>".to_string()),
+					HIDDEN,
+					vec![("vieweronly", NodeValue::Boolean(true))],
+				),
+				param(
+					"font_family_in",
+					"Font family",
+					ValueType::StrCombo,
+					NodeValue::StrCombo(String::new()),
+					0,
+					vec![
+						("combo_option", NodeValue::Text("Sans".to_string())),
+						("combo_option", NodeValue::Text("Serif".to_string())),
+					],
+				),
+				param(
+					"font_size_in",
+					"Font size",
+					ValueType::Float,
+					NodeValue::Float(72.0),
+					0,
+					vec![("min", NodeValue::Float(1.0))],
+				),
+				param(
+					"outline_enabled_in",
+					"Outline",
+					ValueType::Boolean,
+					NodeValue::Boolean(false),
+					0,
+					Vec::new(),
+				),
+				param(
+					"outline_color_in",
+					"Outline color",
+					ValueType::Color,
+					NodeValue::Color([0.0, 0.0, 0.0, 1.0]),
+					0,
+					Vec::new(),
+				),
+				param(
+					"outline_width_in",
+					"Outline width",
+					ValueType::Float,
+					NodeValue::Float(2.0),
+					0,
+					vec![("min", NodeValue::Float(0.0))],
+				),
+				param(
+					"glow_enabled_in",
+					"Glow",
+					ValueType::Boolean,
+					NodeValue::Boolean(false),
+					0,
+					Vec::new(),
+				),
+				param(
+					"glow_color_in",
+					"Glow color",
+					ValueType::Color,
+					NodeValue::Color([1.0, 1.0, 0.0, 1.0]),
+					0,
+					Vec::new(),
+				),
+				param(
+					"glow_radius_in",
+					"Glow radius",
+					ValueType::Float,
+					NodeValue::Float(8.0),
+					0,
+					vec![("min", NodeValue::Float(0.0))],
+				),
+				param(
+					"valign_in",
+					"Vertical align",
+					ValueType::Combo,
+					NodeValue::Combo(0),
+					HIDDEN,
+					Vec::new(),
+				),
+				param(
+					"use_args_in",
+					"Use args",
+					ValueType::Boolean,
+					NodeValue::Boolean(true),
+					HIDDEN,
+					Vec::new(),
+				),
+				param(
+					"args_in",
+					"Args",
+					ValueType::Text,
+					NodeValue::Text(String::new()),
+					ARRAY,
+					vec![("arraystart", NodeValue::Int(1))],
+				),
+			]
 		})
 	}
 
@@ -1808,6 +2007,14 @@ impl AppEngine for MockEngine {
 		Ok(())
 	}
 
+	fn create_text_footage(&mut self, cx: &mut Context<Self>) -> Result<u64, String> {
+		let id = self.next_text_footage_id;
+		self.next_text_footage_id += 1;
+		self.text_footages.push(id);
+		cx.notify();
+		Ok(id)
+	}
+
 	fn drop_footage(
 		&mut self,
 		id: u64,
@@ -2351,6 +2558,16 @@ impl AppEngine for MockEngine {
 		cx.notify();
 	}
 
+	fn add_default_transition(
+		&mut self,
+		_clips: Vec<ClipId>,
+		_cx: &mut Context<Self>,
+	) -> Result<usize, String> {
+		// The demo timeline is not backed by a module graph, so there is no
+		// block to hang a transition on.
+		Err("default transition: the demo engine has no project".to_string())
+	}
+
 	fn multicam_switch_to(&mut self, source: i32, split_clip: bool, cx: &mut Context<Self>) {
 		let guard = self.ensure_demo_multicam();
 		let Some(demo) = guard.as_ref() else {
@@ -2559,12 +2776,19 @@ impl NodeGraphDataSource for MockEngine {
 
 impl ProjectDataSource for MockEngine {
 	fn roots(&self) -> Vec<ProjectEntry> {
-		vec![
+		let mut roots = vec![
 			ProjectEntry::new(1, crate::i18n::tr("bin.footage"), true),
 			ProjectEntry::new(2, crate::i18n::tr("bin.music"), true),
 			ProjectEntry::new(3, "第一稿.mp4", false),
 			ProjectEntry::new(4, "aaa.ove", false),
-		]
+		];
+		// Text generators created through the "添加文本素材" action (the
+		// shared node label `文本`, so the demo's browser stays locale-
+		// independent).
+		for &id in &self.text_footages {
+			roots.push(ProjectEntry::new(id, super::graphops::TEXT_FOOTAGE_LABEL, false));
+		}
+		roots
 	}
 
 	fn children(&self, parent_id: u64) -> Vec<ProjectEntry> {
@@ -2805,6 +3029,13 @@ impl MockEngine {
 		&self.footage_drops
 	}
 
+	/// The text-generator entries created via
+	/// [`AppEngine::create_text_footage`] so far (mock state; the entry ids
+	/// are also listed by [`ProjectDataSource::roots`]).
+	pub fn text_footages(&self) -> &[u64] {
+		&self.text_footages
+	}
+
 	/// The display name of the project-explorer entry with `id`, if any.
 	fn footage_entry_name(&self, id: u64) -> Option<String> {
 		for entry in self.roots() {
@@ -3583,4 +3814,35 @@ mod tests {
 			assert_eq!(engine.read(app).workarea(), Some((Frame(30), Frame(90))));
 		});
 	}
+
+	/// 添加文本素材 on the mock: every call gets a fresh entry id, the ids
+	/// are listed by the bin as leaves labelled with the shared `文本`
+	/// node label (the drop path resolves them by name).
+	#[gpui::test]
+	async fn create_text_footage_lists_a_text_entry(cx: &mut TestAppContext) {
+		cx.update(|app| {
+			let engine = demo_engine(app);
+			let first = engine.update(app, |engine, cx| {
+				engine.create_text_footage(cx).expect("create text footage")
+			});
+			let second = engine.update(app, |engine, cx| {
+				engine.create_text_footage(cx).expect("create text footage")
+			});
+			assert_ne!(first, second, "each action gets a fresh entry id");
+			assert_eq!(engine.read(app).text_footages(), &[first, second]);
+			let listed: Vec<_> = engine
+				.read(app)
+				.roots()
+				.into_iter()
+				.filter(|e| e.id == first || e.id == second)
+				.collect();
+			assert_eq!(listed.len(), 2, "both text entries are listed in the bin");
+			assert!(
+				listed.iter().all(|e| {
+					e.name == crate::oakui::graphops::TEXT_FOOTAGE_LABEL && !e.is_dir
+				}),
+				"the entries carry the shared 文本 label as leaves"
+			);
+		});
+	}
 }
diff --git a/crates/oak-app/src/oakui/real.rs b/crates/oak-app/src/oakui/real.rs
index 98258d1dd..7fa66a4e9 100644
--- a/crates/oak-app/src/oakui/real.rs
+++ b/crates/oak-app/src/oakui/real.rs
@@ -868,6 +868,12 @@ pub struct RealClip {
 	/// The clip's texture chain contains a multicam node (the timeline
 	/// overlay + the panel's detection read the same truth).
 	multicam: bool,
+	/// Duration in frames of the transition attached to the clip's head
+	/// (the seam shared with the previous clip), if any.
+	in_transition: Option<Frame>,
+	/// Duration in frames of the transition attached to the clip's tail
+	/// (the seam shared with the next clip), if any.
+	out_transition: Option<Frame>,
 }
 
 impl ClipData for RealClip {
@@ -894,6 +900,14 @@ impl ClipData for RealClip {
 	fn is_multicam(&self) -> bool {
 		self.multicam
 	}
+
+	fn in_transition(&self) -> Option<Frame> {
+		self.in_transition
+	}
+
+	fn out_transition(&self) -> Option<Frame> {
+		self.out_transition
+	}
 }
 
 /// A running proxy transcode the engine drains on the tick loop (the
@@ -1018,6 +1032,20 @@ impl TrackData for RealTrack {
 	}
 }
 
+/// The color of an adjustment layer's timeline band
+/// ([`RealClip::color`]): a fixed slate that sits outside the per-clip
+/// palette, so a band that carries no media is recognizable at a glance.
+const ADJUSTMENT_LAYER_COLOR: Hsla = Hsla {
+	h: 0.72,
+	s: 0.3,
+	l: 0.58,
+	a: 1.0,
+};
+
+/// The default label of an adjustment layer (the block itself stores no
+/// label; the snapshot synthesizes one like it does for unnamed clips).
+const ADJUSTMENT_LAYER_LABEL: &str = "调整图层";
+
 /// A deterministic clip color from a stable per-clip index (the module
 /// graph exposes no clip color).
 fn clip_color(index: u64) -> Hsla {
@@ -1528,6 +1556,10 @@ impl RealEngine {
 
 	/// Builds an engine with no project open.
 	pub fn new(cx: &mut Context<Self>) -> Self {
+		// Text nodes reach their layout/rasterizer through the process-global
+		// hooks of `oak_node::nodes::textbackend` (oak-node links no font or
+		// shaping crate); fill them in before anything can run a node.
+		super::textengine::install();
 		let rate = VideoFormat::hd_1080p25().rate;
 		let (full_res_tx, full_res_rx) = mpsc::channel::<FullResEvent>();
 		let (thumb_tx, thumb_rx) = mpsc::channel::<ThumbEvent>();
@@ -3419,6 +3451,9 @@ impl RealEngine {
 		let height = graphops::track_behavior(graph, track)
 			.map(|t| px(oak_node::track::internal_height_to_pixel_height(t.height).max(24) as f32))
 			.unwrap_or(px(64.0));
+		// The transition wedges the track's clips carry (left = head
+		// transition, right = tail transition), keyed by clip block.
+		let widths = graphops::clip_transition_widths(graph, track);
 		let clips = graphops::clip_ids(graph, track)
 			.iter()
 			.enumerate()
@@ -3427,17 +3462,30 @@ impl RealEngine {
 				let to_ts = |r: oak_core::Rational| {
 					tb.map(|tb| graphops::rational_to_ts(r, tb)).unwrap_or(0)
 				};
+				let wedge = |w: Option<oak_core::Rational>| w.map(|r| Frame(to_ts(r)));
+				let (head, tail) = widths.get(&block).copied().unwrap_or((None, None));
+				let entry = graph.get(block)?;
+				let is_adjustment = entry
+					.behavior
+					.as_any()
+					.map(|a| a.is::<oak_node::block::AdjustmentBlockBehavior>())
+					.unwrap_or(false);
 				// The clip's color is locked in at creation time (its
 				// `override_color`); clips without one (older projects)
 				// fall back to the track-relative palette so their color
-				// still varies.
-				let entry = graph.get(block)?;
-				let color = if entry.core.override_color >= 0 {
+				// still varies. Adjustment layers wear their own fixed
+				// color instead of a palette entry: their band carries no
+				// media and must read as an effect layer, not a clip.
+				let color = if is_adjustment {
+					ADJUSTMENT_LAYER_COLOR
+				} else if entry.core.override_color >= 0 {
 					clip_color(entry.core.override_color as u64)
 				} else {
 					clip_color(clip_index as u64)
 				};
-				let label = if entry.core.label.is_empty() {
+				let label = if is_adjustment && entry.core.label.is_empty() {
+					ADJUSTMENT_LAYER_LABEL.to_string()
+				} else if entry.core.label.is_empty() {
 					format!("Clip {}", clip_index + 1)
 				} else {
 					entry.core.label.clone()
@@ -3452,6 +3500,8 @@ impl RealEngine {
 					color,
 					block,
 					multicam,
+					in_transition: wedge(head),
+					out_transition: wedge(tail),
 				})
 			})
 			.collect();
@@ -3518,7 +3568,7 @@ impl RealEngine {
 		let block = graphops::id_of(id.0)?;
 		let project = self.project_ref()?;
 		let guard = graphops::lock(project);
-		graphops::clip_behavior(&guard.graph, block)?;
+		graphops::block_core_of(&guard.graph, block)?;
 		// The clip must be on the CURRENT timeline (a stale id of a
 		// removed clip must not resolve).
 		self.tracks
@@ -3559,7 +3609,7 @@ impl RealEngine {
 		};
 		let guard = graphops::lock(project);
 		// A clip block node on the current timeline: its stack is the target.
-		if graphops::clip_behavior(&guard.graph, node).is_some()
+		if graphops::block_core_of(&guard.graph, node).is_some()
 			&& self
 				.tracks
 				.iter()
@@ -3574,7 +3624,7 @@ impl RealEngine {
 				let Some(block) = graphops::id_of(clip.id.0) else {
 					continue;
 				};
-				if graphops::clip_behavior(&guard.graph, block).is_none() {
+				if graphops::block_core_of(&guard.graph, block).is_none() {
 					continue;
 				}
 				if let Some(effect) = super::effectchain::chain(&guard.graph, block)
@@ -4194,6 +4244,34 @@ impl AppEngine for RealEngine {
 		);
 	}
 
+	fn add_adjustment_layer(
+		&mut self,
+		index: usize,
+		frame: Frame,
+		cx: &mut Context<Self>,
+	) -> Result<(), String> {
+		let Some(project) = self.project.clone() else {
+			return Err("add adjustment layer: no project is open".to_string());
+		};
+		let Some(seq) = self.sequence else {
+			return Err("add adjustment layer: no sequence is open".to_string());
+		};
+		let Some(tb) = self.time_base() else {
+			return Err("add adjustment layer: the sequence has no frame rate".to_string());
+		};
+		let Some(track) = self.tracks.get(index).map(|t| t.track) else {
+			return Err(format!("add adjustment layer: no track at index {index}"));
+		};
+		// The default span: five seconds from the clicked frame (the
+		// snapshot's non-snapping approximation of the C++ default).
+		let fps = tb.1 as f64 / tb.0.max(1) as f64;
+		let length = (5.0 * fps).round().max(1.0) as i64;
+		let in_ts = frame.0.max(0);
+		let result = graphops::create_adjustment_layer(&project, seq, track, in_ts, in_ts + length);
+		self.apply_edit(result.clone().map(|_| ()), "add adjustment layer", cx);
+		result.map(|_| ())
+	}
+
 	fn set_track_height(&mut self, height: Pixels, cx: &mut Context<Self>) {
 		let (Some(project), Some(seq)) = (self.project.clone(), self.sequence) else {
 			return;
@@ -4676,11 +4754,40 @@ impl AppEngine for RealEngine {
 				}
 				cx.notify();
 			}
-			// Selection / zoom / transition / track-selected / context-menu:
+			TimelineEvent::TransitionChanged {
+				clip,
+				edge,
+				new_length,
+			} => {
+				let Some(block) = self.clip_block(*clip) else {
+					return;
+				};
+				if self.clip_track_locked(block) {
+					return;
+				}
+				let Some(project) = self.project.clone() else {
+					return;
+				};
+				let Some(tb) = self.time_base() else {
+					return;
+				};
+				// The widget works in the sequence's frame unit, the
+				// command layer in seconds.
+				let frame = graphops::ts_to_rational(1, tb);
+				let requested = graphops::ts_to_rational(new_length.0.max(1), tb);
+				let result = graphops::set_transition_length(
+					&project,
+					block,
+					matches!(edge, TrimEdge::Start),
+					frame,
+					requested,
+				);
+				self.apply_edit(result.map(|_| ()), "transition length", cx);
+			}
+			// Selection / zoom / track-selected / context-menu:
 			// not editable (the right-click is answered by the panel's popup).
 			TimelineEvent::SelectionChanged
 			| TimelineEvent::TrackSelected { .. }
-			| TimelineEvent::TransitionChanged { .. }
 			| TimelineEvent::ContextMenuRequested { .. }
 			| TimelineEvent::ZoomChanged(_) => {}
 			TimelineEvent::TrackToggleRequested { track, toggle } => {
@@ -5147,14 +5254,24 @@ impl AppEngine for RealEngine {
 		// sequence node (TEXTURE_INPUT, like `clip_connected_sequence`
 		// resolves), so a multicam source sequence drops as a single
 		// multi-cam clip the panel detects.
-		let is_sequence = {
+		//
+		// Text entry drop: a GENERATOR clip — the clip reads the text
+		// node's output through `tex_in` instead of decoding media.
+		let (is_sequence, is_text) = {
 			let guard = graphops::lock(&project);
-			graphops::sequence_behavior(&guard.graph, footage).is_some()
+			(
+				graphops::sequence_behavior(&guard.graph, footage).is_some(),
+				graphops::node_type_id(&guard.graph, footage) == graphops::TEXT_FOOTAGE_TYPE_ID,
+			)
 		};
 		if is_sequence {
 			self.drop_sequence_entry(&project, footage, track_index, time, cx);
 			return;
 		}
+		if is_text {
+			self.drop_text_entry(&project, footage, track_index, time, cx);
+			return;
+		}
 		let (filename, video_streams, total_streams, seconds) = {
 			let guard = graphops::lock(&project);
 			let Some(f) = graphops::footage_behavior(&guard.graph, footage) else {
@@ -5738,6 +5855,15 @@ impl AppEngine for RealEngine {
 		Ok(id.identity())
 	}
 
+	fn create_text_footage(&mut self, cx: &mut Context<Self>) -> Result<u64, String> {
+		let Some(project) = self.project.clone() else {
+			return Err(crate::i18n::tr("seqprops.error.no_project").to_string());
+		};
+		let id = graphops::create_text_footage_node(&project)?;
+		self.apply_edit(Ok(()), "add text footage", cx);
+		Ok(id.identity())
+	}
+
 	fn update_sequence_parameters(
 		&mut self,
 		id: u64,
@@ -6427,6 +6553,45 @@ impl AppEngine for RealEngine {
 		self.apply_edit(result, "multicam enable/disable", cx);
 	}
 
+	fn add_default_transition(
+		&mut self,
+		clips: Vec<ClipId>,
+		cx: &mut Context<Self>,
+	) -> Result<usize, String> {
+		let Some(project) = self.project.clone() else {
+			return Err("default transition: no project is open".to_string());
+		};
+		let Some(seq) = self.sequence else {
+			return Err("default transition: no sequence is open".to_string());
+		};
+		let Some(tb) = self.time_base() else {
+			return Err("default transition: the sequence has no frame rate".to_string());
+		};
+		// The preference is the WHOLE transition length; a default
+		// transition is symmetric, so each wedge takes half of it (at least
+		// one frame — `ts_to_rational` needs a positive frame count).
+		let seconds: f64 = config_get_string(CONFIG_KEY_DEFAULT_TRANSITION_SEC)
+			.parse()
+			.unwrap_or_else(|_| {
+				DEFAULT_TRANSITION_SEC
+					.parse()
+					.expect("the compiled-in default parses")
+			});
+		let fps = tb.1 as f64 / tb.0.max(1) as f64;
+		let half_frames = (seconds * fps / 2.0).round().max(1.0) as i64;
+		let half = graphops::ts_to_rational(half_frames, tb);
+		let blocks: Vec<NodeId> = clips
+			.iter()
+			.filter_map(|id| {
+				let block = self.clip_block(*id)?;
+				(!self.clip_track_locked(block)).then_some(block)
+			})
+			.collect();
+		let result = graphops::add_default_transition(&project, seq, &blocks, half);
+		self.apply_edit(result.clone().map(|_| ()), "add default transition", cx);
+		result
+	}
+
 	fn multicam_switch_to(&mut self, source: i32, split_clip: bool, cx: &mut Context<Self>) {
 		let Some(project) = self.project.clone() else {
 			return;
@@ -6726,6 +6891,64 @@ impl RealEngine {
 		Ok(audio_clip)
 	}
 
+	/// Drops a text-project entry on the timeline as a GENERATOR clip: the
+	/// clip spans five seconds from the drop point and its `tex_in` reads
+	/// the text node, so the generator re-renders at whatever time the
+	/// clip spans (no media to decode and no footage length to clamp to).
+	/// The target track policy is the sequence drop's: the pointed display
+	/// row when it is a video track, else the first video track. Undoable
+	/// as one "Add Clip".
+	fn drop_text_entry(
+		&mut self,
+		project: &ProjectRef,
+		text: NodeId,
+		track_index: usize,
+		time: Frame,
+		cx: &mut Context<Self>,
+	) {
+		let Some(host_seq) = self.sequence else {
+			println!("[real engine] drop text: no sequence open");
+			return;
+		};
+		let video_target = if self
+			.tracks
+			.get(track_index)
+			.is_some_and(|t| t.kind == TrackKind::Video)
+		{
+			track_index
+		} else if let Some(index) = self.tracks.iter().position(|t| t.kind == TrackKind::Video)
+		{
+			index
+		} else {
+			println!("[real engine] drop text: no video track");
+			return;
+		};
+		let video_index = self.tracks[video_target].track_index;
+		// Five seconds of the host sequence's frames (the drop's default
+		// extent — the text clip has no media length to take one from).
+		let (in_ts, length_ts) = {
+			let guard = graphops::lock(project);
+			let Ok(tb) = graphops::sequence_time_base(&guard.graph, host_seq)
+				.ok_or_else(|| "host sequence has no frame rate".to_string())
+			else {
+				println!("[real engine] drop text: host has no frame rate");
+				return;
+			};
+			let fps_f = tb.1.max(1) as f64 / tb.0.max(1) as f64;
+			((time.0.max(0)), ((5.0 * fps_f).round() as i64).max(1))
+		};
+		let placed = graphops::place_text_clip(
+			project,
+			host_seq,
+			text,
+			video_index,
+			in_ts,
+			in_ts + length_ts,
+		);
+		let result = placed.map(|_| ());
+		self.apply_edit(result, "drop text", cx);
+	}
+
 	/// Drops a sequence-project entry on the timeline as a NESTED-SEQUENCE
 	/// clip: ONE video clip on the video track (not the sequence's tracks —
 	/// a frame, not a folder, of tracks), fed from the sequence node. This
@@ -10258,6 +10481,315 @@ mod tests {
 		);
 	}
 
+	/// 添加文本素材 through the engine facade: `create_text_footage`
+	/// creates a `text3` generator labelled "文本" mounted under the root
+	/// (the explorer lists it; it is NOT a sequence), and dropping it lands
+	/// a five-second generator clip on the host sequence's first video
+	/// track, wired through `tex_in` and undoable as one row.
+	#[gpui::test]
+	async fn engine_creates_text_footage_and_drops_it_on_the_timeline(
+		cx: &mut gpui::TestAppContext,
+	) {
+		let _media = media_lock();
+		let engine = cx.update(|cx| cx.new(|cx| RealEngine::create(cx)));
+		cx.update(|app| engine.update(app, |engine, cx| engine.new_project(cx)));
+
+		let text = cx.update(|app| {
+			engine
+				.update(app, |engine, cx| engine.create_text_footage(cx))
+				.expect("create text footage")
+		});
+		let project = cx.read(|app| engine.read(app).project.clone().expect("project"));
+		let node = graphops::id_of(text).expect("text node");
+		{
+			let guard = graphops::lock(&project);
+			assert_eq!(
+				graphops::node_type_id(&guard.graph, node),
+				graphops::TEXT_FOOTAGE_TYPE_ID
+			);
+			assert_eq!(
+				graphops::node_label(&guard.graph, node),
+				graphops::TEXT_FOOTAGE_LABEL
+			);
+			let children = guard
+				.graph
+				.get(guard.root)
+				.and_then(|e| e.behavior.as_any())
+				.and_then(|a| a.downcast_ref::<oak_node::folder::FolderBehavior>())
+				.map(|f| f.children.clone())
+				.unwrap_or_default();
+			assert!(
+				children.contains(&node),
+				"the text entry mounts under the root folder"
+			);
+		}
+		let is_seq = cx.read(|app| engine.read(app).entry_is_sequence(text));
+		assert!(!is_seq, "a text entry is not a sequence");
+
+		// Open a host sequence; the drop targets its first video track.
+		let seq = cx.update(|app| {
+			engine
+				.update(app, |engine, cx| {
+					engine
+						.create_sequence_with_params(
+							"Text Drop".to_string(),
+							VideoFormat {
+								width: 1920,
+								height: 1080,
+								rate: FrameRate::new(30000, 1001),
+							},
+							false,
+							cx,
+						)
+						.expect("create sequence")
+				})
+		});
+		cx.update(|app| {
+			engine.update(app, |engine, cx| {
+				engine.drop_footage(text, TrackKind::Video, 0, Frame(25), cx)
+			})
+		});
+
+		let seq_node = graphops::id_of(seq).expect("sequence node");
+		// The drop row is a display row: the engine's track view runs
+		// top-to-bottom, so row 0 is the topmost video track of the
+		// sequence's list (V2 with the default layout).
+		let target_row = cx.read(|app| {
+			let row = &engine.read(app).tracks[0];
+			assert_eq!(row.kind, TrackKind::Video, "the drop row is a video track");
+			row.track_index
+		});
+		let (tb, tracks) = {
+			let guard = graphops::lock(&project);
+			(
+				graphops::sequence_time_base(&guard.graph, seq_node).expect("time base"),
+				graphops::track_ids(&guard.graph, seq_node, TrackType::Video),
+			)
+		};
+		let track = tracks[target_row];
+		{
+			let guard = graphops::lock(&project);
+			let clips = graphops::clip_ids(&guard.graph, track);
+			assert_eq!(clips.len(), 1, "the drop lands exactly one clip: {clips:?}");
+			let clip = clips[0];
+			assert_eq!(
+				graphops::clip_track(&guard.graph, clip),
+				Some(track),
+				"the clip lands on the drop row's video track"
+			);
+			for (index, other) in tracks.iter().enumerate() {
+				if index == target_row {
+					continue;
+				}
+				assert!(
+					graphops::clip_ids(&guard.graph, *other).is_empty(),
+					"the other video tracks stay untouched"
+				);
+			}
+			let (in_r, out_r, _) = graphops::clip_range(&guard.graph, clip).expect("clip range");
+			let fps = tb.1 as f64 / tb.0 as f64;
+			let length = (5.0 * fps).round().max(1.0) as i64;
+			assert_eq!(in_r, graphops::ts_to_rational(25, tb));
+			assert_eq!(
+				out_r,
+				graphops::ts_to_rational(25 + length, tb),
+				"the clip spans five seconds from the drop point"
+			);
+			assert_eq!(
+				guard
+					.graph
+					.connected_output(clip, oak_node::block::clip_input::TEXTURE_INPUT, -1),
+				Some(node),
+				"the clip reads the generator through tex_in"
+			);
+		}
+
+		// One undo removes the clip (and its edge) again.
+		oak_undo::global::undo().unwrap();
+		{
+			let guard = graphops::lock(&project);
+			assert!(
+				graphops::clip_ids(&guard.graph, track).is_empty(),
+				"the undo removes the dropped clip"
+			);
+			assert!(
+				graphops::track_behavior(&guard.graph, track)
+					.map(|t| t.blocks.is_empty())
+					.unwrap_or(false),
+				"the undo also drops the gap the placement inserted"
+			);
+		}
+		oak_undo::global::clear().unwrap();
+	}
+
+	/// 调整图层 through the engine facade: `add_adjustment_layer` places a
+	/// five-second empty block at the clicked frame on a video track (an
+	/// `AdjustmentBlockBehavior` with no footage and no texture wiring),
+	/// refuses audio tracks, resolves through the timeline selection, and is
+	/// undone by one undo step.
+	#[gpui::test]
+	async fn engine_adds_and_undoes_adjustment_layer(cx: &mut gpui::TestAppContext) {
+		let _media = media_lock();
+		let engine = cx.update(|cx| cx.new(|cx| RealEngine::create(cx)));
+		cx.update(|app| engine.update(app, |engine, cx| engine.new_project(cx)));
+
+		let seq = cx.update(|app| {
+			engine.update(app, |engine, cx| {
+				engine
+					.create_sequence_with_params(
+						"Adjustment Layer".to_string(),
+						VideoFormat {
+							width: 1920,
+							height: 1080,
+							rate: FrameRate::new(25, 1),
+						},
+						false,
+						cx,
+					)
+					.expect("create sequence")
+			})
+		});
+		let seq_node = graphops::id_of(seq).expect("sequence node");
+		let project = cx.read(|app| engine.read(app).project.clone().expect("project"));
+		let tb = {
+			let guard = graphops::lock(&project);
+			graphops::sequence_time_base(&guard.graph, seq_node).expect("time base")
+		};
+		let fps = tb.1 as f64 / tb.0.max(1) as f64;
+		let length = (5.0 * fps).round().max(1.0) as i64;
+
+		// An audio row is not a legal host.
+		let audio_row = cx.read(|app| {
+			engine
+				.read(app)
+				.tracks
+				.iter()
+				.position(|row| row.kind == TrackKind::Audio)
+				.expect("the default layout has audio rows")
+		});
+		let rejected = cx.update(|app| {
+			engine.update(app, |engine, cx| {
+				engine.add_adjustment_layer(audio_row, Frame(0), cx)
+			})
+		});
+		let error = rejected.expect_err("audio tracks reject adjustment layers");
+		assert!(
+			error.contains("video track"),
+			"the rejection names the constraint: {error}"
+		);
+
+		// Row 0 is the topmost video row; frame 12 anchors the new layer.
+		let target_row = cx.read(|app| {
+			let row = &engine.read(app).tracks[0];
+			assert_eq!(row.kind, TrackKind::Video, "row 0 is a video track");
+			row.track_index
+		});
+		cx.update(|app| {
+			engine.update(app, |engine, cx| {
+				engine.add_adjustment_layer(0, Frame(12), cx)
+			})
+		})
+		.expect("add adjustment layer");
+
+		// The snapshot carries the new layer: the fixed slate (outside the
+		// per-clip palette), the synthesized label, and the five-second span
+		// from the clicked frame.
+		let (track, clip) = cx.read(|app| {
+			let engine = engine.read(app);
+			let row = &engine.tracks[0];
+			assert_eq!(row.clips.len(), 1, "the add lands exactly one block");
+			(row.track, row.clips[0].clone())
+		});
+		assert_eq!(clip.range, FrameRange::new(Frame(12), Frame(12 + length)));
+		assert_eq!(clip.media_in, Frame(0));
+		assert_eq!(clip.label, SharedString::from("调整图层"));
+		assert_eq!(clip.color, ADJUSTMENT_LAYER_COLOR);
+		assert!(
+			(0..16).all(|index| clip.color != clip_color(index)),
+			"the adjustment slate sits outside the per-clip palette"
+		);
+		let block = clip.block;
+
+		// The graph holds an adjustment block on the row's track, spanning
+		// the same range, with no texture source.
+		{
+			let guard = graphops::lock(&project);
+			let videos = graphops::track_ids(&guard.graph, seq_node, TrackType::Video);
+			assert_eq!(
+				videos[target_row], track,
+				"the display row maps to the graph track"
+			);
+			assert_eq!(graphops::clip_track(&guard.graph, block), Some(track));
+			assert_eq!(
+				graphops::clip_ids(&guard.graph, track),
+				vec![block],
+				"the placed block is the track's only clip"
+			);
+			for (index, other) in videos.iter().enumerate() {
+				if index == target_row {
+					continue;
+				}
+				assert!(
+					graphops::clip_ids(&guard.graph, *other).is_empty(),
+					"the other video tracks stay untouched"
+				);
+			}
+			let entry = guard.graph.get(block).expect("block node");
+			assert!(
+				entry
+					.behavior
+					.as_any()
+					.map(|a| a.is::<oak_node::block::AdjustmentBlockBehavior>())
+					.unwrap_or(false),
+				"the placed block carries the adjustment behavior"
+			);
+			let (in_r, out_r, media_in) =
+				graphops::clip_range(&guard.graph, block).expect("clip range");
+			assert_eq!(in_r, graphops::ts_to_rational(12, tb));
+			assert_eq!(
+				out_r,
+				graphops::ts_to_rational(12 + length, tb),
+				"the layer spans five seconds from the clicked frame"
+			);
+			assert_eq!(media_in, oak_core::Rational::new(0, 1));
+			assert!(
+				!guard.graph.is_input_connected(
+					block,
+					oak_node::block::adjustment_input::TEXTURE_INPUT,
+					-1
+				),
+				"the adjustment layer carries no texture source"
+			);
+		}
+
+		// The timeline selection resolves the snapshot id to the block (the
+		// inspector targets adjustment layers like any clip).
+		cx.update(|app| {
+			engine.update(app, |engine, cx| {
+				engine.set_selected_clips(vec![ClipId(block.identity())], cx)
+			})
+		});
+		let selected = cx.read(|app| engine.read(app).selected_clip_node());
+		assert_eq!(selected, Some(block), "the selected layer resolves to its block");
+
+		// One undo removes the placement (and the leading gap) again.
+		oak_undo::global::undo().unwrap();
+		{
+			let guard = graphops::lock(&project);
+			assert!(
+				graphops::clip_ids(&guard.graph, track).is_empty(),
+				"the undo removes the placed adjustment layer"
+			);
+			assert!(
+				graphops::track_behavior(&guard.graph, track)
+					.map(|t| t.blocks.is_empty())
+					.unwrap_or(false),
+				"the undo also drops the gap the placement inserted"
+			);
+		}
+		oak_undo::global::clear().unwrap();
+	}
+
 	/// `update_sequence_parameters` rewrites the name, the format and the
 	/// interlace flag in one call; `sequence_parameters` reads the new
 	/// state back (the properties dialog's commit path).
diff --git a/crates/oak-app/src/oakui/renderops.rs b/crates/oak-app/src/oakui/renderops.rs
index 8a702702b..81f00f53e 100644
--- a/crates/oak-app/src/oakui/renderops.rs
+++ b/crates/oak-app/src/oakui/renderops.rs
@@ -35,7 +35,9 @@ use oak_node::id::NodeId;
 use oak_node::track::TrackType;
 use oak_render::manager::RenderManager;
 use oak_render::procpool::ShmFrameRef;
-use oak_render::ticket::{AudioTicketParams, MontageClip, TicketPayload, VideoTicketParams};
+use oak_render::ticket::{
+	AdjustmentSpan, AudioTicketParams, MontageClip, TicketPayload, VideoTicketParams,
+};
 
 use super::engine::{ExportEvent, ExportSession};
 use super::frames::{bgra_bytes_to_render_image, f32_rgba_to_bgra_image};
@@ -240,17 +242,29 @@ fn clip_effects(g: &oak_node::graph::Graph, block_id: NodeId) -> Vec<oak_render:
 		.collect()
 }
 
-/// The video montage at sequence time `time`: every clip covering `time`
+/// The video montage at sequence time `time`, together with the
+/// adjustment layers over that time (M14 W3): every clip covering `time`
 /// on video tracks, ordered bottom-to-top (the highest-numbered track —
 /// the list's last — is topmost, so it is composited last; NLE stacking
 /// matches the timeline UI, which shows the highest-numbered track on
 /// top). Hidden tracks (the muted flag doubles as the video visibility
 /// toggle, Olive parity) contribute nothing.
-pub fn video_montage(p: &ProjectRef, seq: NodeId, time: Rational) -> Vec<MontageClip> {
+///
+/// The adjustment spans come out in the same bottom-up walk order (their
+/// `track_index` boundaries are therefore non-decreasing) and, exactly
+/// like the graph path, an enabled adjustment block covering `time`
+/// takes over its track: the span is emitted and the track's own clips
+/// are skipped, whether or not the adjustment's chain has any effects.
+pub fn video_montage_with_adjustments(
+	p: &ProjectRef,
+	seq: NodeId,
+	time: Rational,
+) -> (Vec<MontageClip>, Vec<AdjustmentSpan>) {
 	let g = lock(p);
 	let mut clips = Vec::new();
+	let mut adjustments = Vec::new();
 	let Some(s) = sequence_behavior(&g.graph, seq) else {
-		return clips;
+		return (clips, adjustments);
 	};
 	for &list_id in &s.track_lists {
 		let Some(list) = track_list_behavior(&g.graph, list_id) else {
@@ -266,6 +280,13 @@ pub fn video_montage(p: &ProjectRef, seq: NodeId, time: Rational) -> Vec<Montage
 			if track.muted {
 				continue;
 			}
+			// Everything composited so far sits under this track, so the
+			// span's boundary is the clip count before the track's own
+			// clips join the list.
+			if let Some(span) = covering_adjustment(&g.graph, &track.blocks, time, clips.len()) {
+				adjustments.push(span);
+				continue;
+			}
 			for &block_id in &track.blocks {
 				let Some(clip) = clip_behavior(&g.graph, block_id) else {
 					continue;
@@ -305,7 +326,51 @@ pub fn video_montage(p: &ProjectRef, seq: NodeId, time: Rational) -> Vec<Montage
 			}
 		}
 	}
-	clips
+	(clips, adjustments)
+}
+
+/// The video montage at sequence time `time`. The adjustment layers are
+/// dropped; callers that render (as opposed to previewing the clip list)
+/// want [`video_montage_with_adjustments`].
+pub fn video_montage(p: &ProjectRef, seq: NodeId, time: Rational) -> Vec<MontageClip> {
+	video_montage_with_adjustments(p, seq, time).0
+}
+
+/// The adjustment span `blocks` contributes at `time`, if any: the first
+/// enabled adjustment block covering `time` — the graph path's track
+/// takeover rule (`render_graph_frame`), so both paths agree on which
+/// blocks are layers and which track each takes over. `below` is the
+/// number of clips already composited from the tracks underneath (the
+/// span's montage-list boundary).
+fn covering_adjustment(
+	g: &oak_node::graph::Graph,
+	blocks: &[NodeId],
+	time: Rational,
+	below: usize,
+) -> Option<AdjustmentSpan> {
+	for &block_id in blocks {
+		let Some(block) = g.get(block_id) else {
+			continue;
+		};
+		let Some(adjustment) = block
+			.behavior
+			.as_any()
+			.and_then(|a| a.downcast_ref::<oak_node::block::AdjustmentBlockBehavior>())
+		else {
+			continue;
+		};
+		let in_ = adjustment.core.in_();
+		let out = adjustment.core.out();
+		if adjustment.core.enabled && time >= in_ && time < out {
+			return Some(AdjustmentSpan {
+				in_time: in_,
+				out_time: out,
+				track_index: below,
+				effects: clip_effects(g, block_id),
+			});
+		}
+	}
+	None
 }
 
 /// The video montage at sequence time `time` containing ONLY the clip on
@@ -403,6 +468,7 @@ pub fn multicam_angle_frame_params(
 		cache_timebase: None,
 		footage: None,
 		montage: single_track_video_montage(p, seq, track, time),
+		adjustments: Vec::new(),
 	})
 }
 
@@ -810,9 +876,10 @@ pub fn sequence_frame_params(
 	// 720p proxy up to a 1080p target.
 	let (width, height) = clamp_render_size(width, height, &proxy_limits_at(p, seq, time));
 	// Bind the uuid before the literal: an inline `lock(p)` temporary in the
-	// block-tail struct expression would still be alive when `video_montage`
-	// re-locks the project, deadlocking the same thread.
+	// block-tail struct expression would still be alive when the montage
+	// builder re-locks the project, deadlocking the same thread.
 	let project = lock(p).uuid.clone();
+	let (montage, adjustments) = video_montage_with_adjustments(p, seq, time);
 	Ok(VideoTicketParams {
 		viewer: seq.identity(),
 		project,
@@ -824,7 +891,8 @@ pub fn sequence_frame_params(
 		cache_id: None,
 		cache_timebase: None,
 		footage: None,
-		montage: video_montage(p, seq, time),
+		montage,
+		adjustments,
 	})
 }
 
@@ -891,6 +959,7 @@ pub fn footage_frame_params(
 		cache_timebase: None,
 		footage: Some((filename, stream_index)),
 		montage: Vec::new(),
+		adjustments: Vec::new(),
 	})
 }
 
@@ -1412,6 +1481,7 @@ mod tests {
 			cache_timebase: None,
 			footage: None,
 			montage,
+			adjustments: Vec::new(),
 		};
 		let mut dst = vec![0u8; 64 * 64 * 16];
 		oak_render::eval::render_montage_frame_into(time, &params, (64, 64), &mut dst, 64 * 16)
@@ -1488,6 +1558,7 @@ mod tests {
 				cache_timebase: None,
 				footage: None,
 				montage,
+				adjustments: Vec::new(),
 			};
 			let mut dst = vec![0u8; 64 * 64 * 16];
 			oak_render::eval::render_montage_frame_into(time, &params, (64, 64), &mut dst, 64 * 16)
@@ -1555,6 +1626,349 @@ mod tests {
 		let _ = std::fs::remove_file(&media);
 	}
 
+	/// An adjustment-layer block spanning `[in_ts, out_ts)` on `track`
+	/// carrying one Opacity effect at `opacity` — the timeline's
+	/// adjustment-layer drop plus the effect library's append, wired the
+	/// way the inspector does it. Returns `(block, effect)`.
+	fn adjustment_layer_with_opacity(
+		p: &ProjectRef,
+		seq: NodeId,
+		track: NodeId,
+		in_ts: i64,
+		out_ts: i64,
+		opacity: f64,
+	) -> (NodeId, NodeId) {
+		let (tb, in_r) = {
+			let g = lock(p);
+			let tb = graphops::sequence_time_base(&g.graph, seq).expect("the sequence has a timebase");
+			(tb, graphops::ts_to_rational(in_ts, tb))
+		};
+		let block = {
+			let mut g = lock(p);
+			let (core, behavior) = oak_node::block::adjustment_create();
+			let id = g.graph.add_node(core, behavior);
+			let entry = g.graph.get_mut(id).expect("the new block");
+			let adjustment = entry
+				.behavior
+				.as_any_mut()
+				.and_then(|a| a.downcast_mut::<oak_node::block::AdjustmentBlockBehavior>())
+				.expect("an adjustment behavior");
+			adjustment.core.range = TimeRange::new(in_r, graphops::ts_to_rational(out_ts, tb));
+			id
+		};
+		let (list, index) = {
+			let g = lock(p);
+			let list = graphops::track_behavior(&g.graph, track)
+				.and_then(|t| t.track_list)
+				.expect("the track has a list");
+			let index = graphops::track_list_behavior(&g.graph, list)
+				.and_then(|l| l.tracks.iter().position(|&t| t == track))
+				.expect("the track is in its list") as i32;
+			(list, index)
+		};
+		graphops::push_command(
+			oak_timeline::undopointer::TrackPlaceBlockCommand::new(
+				graphops::node_ref(p, list),
+				index,
+				graphops::node_ref(p, block),
+				in_r,
+			)
+			.to_command(),
+			"Add Adjustment Layer",
+		)
+		.expect("place the adjustment layer");
+		let fx = crate::oakui::effectchain::insert(
+			p,
+			block,
+			usize::MAX,
+			"org.olivevideoeditor.Olive.opacity",
+		)
+		.expect("append the opacity effect");
+		crate::oakui::effectchain::set_input_value(
+			p,
+			fx,
+			"opacity_in",
+			oak_node::value::NodeValue::Float(opacity),
+		)
+		.expect("set the opacity value");
+		(block, fx)
+	}
+
+	/// M14 W3 montage path: an adjustment layer ABOVE the clip stack
+	/// grades the frame accumulated underneath it — a 50% Opacity on the
+	/// layer halves every channel (alpha included) of the accumulated
+	/// composite, and only inside the layer's range.
+	#[test]
+	fn montage_adjustment_span_halves_the_accumulated_frame() {
+		let _media = media_lock();
+		// Pin the legacy sRGB pass-through (see the effect-stack test) so the
+		// pixel assertions below do not depend on the ACEScg default.
+		oak_core::color::set_pipeline_color_settings(
+			oak_core::colormath::WorkingColorSpace::SrgbLegacy,
+			oak_core::colormath::OutputColorSpec::default(),
+		);
+		oak_undo::global::clear().unwrap();
+		let red = std::env::temp_dir().join(format!("oakapp_adj_red_{}.mp4", std::process::id()));
+		let blue = std::env::temp_dir().join(format!("oakapp_adj_blue_{}.mp4", std::process::id()));
+		oak_codec::testmedia::write_test_clip_solid(&red, 64, 64, 10, 10, [0.9, 0.1, 0.1, 1.0])
+			.expect("generate the red media");
+		oak_codec::testmedia::write_test_clip_solid(&blue, 64, 64, 10, 10, [0.1, 0.1, 0.9, 1.0])
+			.expect("generate the blue media");
+
+		let project = graphops::create_project();
+		let seq = graphops::create_sequence(&project, "Adjustment Montage");
+		let red_footage = graphops::import_footage(&project, &red).expect("import the red media");
+		let blue_footage = graphops::import_footage(&project, &blue).expect("import the blue media");
+		graphops::place_footage_clip(&project, seq, red_footage, TrackType::Video, 0, 0, 10, 0)
+			.expect("place the V1 (red) clip");
+		graphops::place_footage_clip(&project, seq, blue_footage, TrackType::Video, 1, 0, 10, 0)
+			.expect("place the V2 (blue) clip");
+		// A third video track (V3, above both clips) carries the layer.
+		graphops::add_track(&project, seq, TrackType::Video).expect("add the adjustment track");
+		let adj_track = {
+			let g = lock(&project);
+			graphops::track_ids(&g.graph, seq, TrackType::Video)[2]
+		};
+		let (adj, _fx) = adjustment_layer_with_opacity(&project, seq, adj_track, 3, 7, 0.5);
+
+		let tb = graphops::sequence_time_base(&lock(&project).graph, seq).unwrap();
+		let at = |frame: i64| graphops::ts_to_rational(frame, tb);
+		let render = |montage: Vec<MontageClip>, adjustments: Vec<AdjustmentSpan>, at_time: Rational| {
+			let params = VideoTicketParams {
+				viewer: 0,
+				project: String::new(),
+				time: at_time,
+				force_size: Some((64, 64)),
+				force_format: Some(oak_core::PixelFormat::F32),
+				cache: None,
+				cache_dir: None,
+				cache_id: None,
+				cache_timebase: None,
+				footage: None,
+				montage,
+				adjustments,
+			};
+			let mut dst = vec![0u8; 64 * 64 * 16];
+			oak_render::eval::render_montage_frame_into(
+				params.time,
+				&params,
+				(64, 64),
+				&mut dst,
+				64 * 16,
+			)
+			.expect("montage render");
+			dst
+		};
+		let pixel = |frame: &[u8]| {
+			let off = (8 * 64 + 8) * 16;
+			[0, 1, 2, 3]
+				.map(|i| f32::from_le_bytes(frame[off + i * 4..off + i * 4 + 4].try_into().unwrap()))
+		};
+
+		// Inside the layer's range the span rides the ticket, above the stack.
+		let (montage, spans) = video_montage_with_adjustments(&project, seq, at(5));
+		assert_eq!(montage.len(), 2, "both clips cover frame 5");
+		assert_eq!(spans.len(), 1, "the layer emits one span");
+		assert_eq!(
+			spans[0].track_index,
+			montage.len(),
+			"the layer sits above every clip (it grades the whole composite)"
+		);
+		assert_eq!(spans[0].effects.len(), 1, "the layer's stack rides the span");
+		assert_eq!(spans[0].effects[0].type_id, "org.olivevideoeditor.Olive.opacity");
+
+		let baseline = render(montage.clone(), Vec::new(), at(5));
+		let graded = render(montage.clone(), spans, at(5));
+		let (b, g) = (pixel(&baseline), pixel(&graded));
+		assert!(
+			b[2] > 0.6 && b[3] > 0.9,
+			"the baseline shows V2's opaque blue on top (b={}, a={})",
+			b[2],
+			b[3]
+		);
+		for i in 0..4 {
+			assert!(
+				(g[i] - b[i] * 0.5).abs() < 0.06,
+				"the layer halves channel {i} ({}, expected {})",
+				g[i],
+				b[i] * 0.5
+			);
+		}
+
+		// Outside the layer's range no span is emitted and the render is
+		// byte-for-byte the plain montage.
+		for frame in [1, 8] {
+			let (m, s) = video_montage_with_adjustments(&project, seq, at(frame));
+			assert_eq!(m.len(), 2, "both clips still cover frame {frame}");
+			assert!(s.is_empty(), "frame {frame} is outside the layer's range");
+			assert_eq!(
+				render(m.clone(), s, at(frame)),
+				render(m, Vec::new(), at(frame)),
+				"frame {frame} renders ungraded"
+			);
+		}
+
+		// A disabled layer is not a layer at all: no span, and the frame is
+		// the plain composite (the graph path's enabled check).
+		{
+			let mut g = lock(&project);
+			let entry = g.graph.get_mut(adj).expect("the layer block");
+			entry
+				.behavior
+				.as_any_mut()
+				.and_then(|a| a.downcast_mut::<oak_node::block::AdjustmentBlockBehavior>())
+				.expect("an adjustment behavior")
+				.core
+				.enabled = false;
+		}
+		let (m, s) = video_montage_with_adjustments(&project, seq, at(5));
+		assert_eq!(m.len(), 2, "the clips stay in the montage");
+		assert!(s.is_empty(), "a disabled layer emits no span");
+
+		oak_undo::global::clear().unwrap();
+		let _ = std::fs::remove_file(&red);
+		let _ = std::fs::remove_file(&blue);
+	}
+
+	/// M14 W3 montage path: a layer BELOW a clip grades only the tracks
+	/// underneath it — the span's boundary is the montage index reached
+	/// when the clips under the layer are composited, so V3's clip lands
+	/// over the graded result instead of being graded itself.
+	#[test]
+	fn montage_adjustment_span_grades_only_the_tracks_below_it() {
+		let _media = media_lock();
+		oak_core::color::set_pipeline_color_settings(
+			oak_core::colormath::WorkingColorSpace::SrgbLegacy,
+			oak_core::colormath::OutputColorSpec::default(),
+		);
+		oak_undo::global::clear().unwrap();
+		let red = std::env::temp_dir().join(format!("oakapp_adjlow_red_{}.mp4", std::process::id()));
+		let blue =
+			std::env::temp_dir().join(format!("oakapp_adjlow_blue_{}.mp4", std::process::id()));
+		oak_codec::testmedia::write_test_clip_solid(&red, 64, 64, 10, 10, [0.9, 0.1, 0.1, 1.0])
+			.expect("generate the red media");
+		oak_codec::testmedia::write_test_clip_solid(&blue, 64, 64, 10, 10, [0.1, 0.1, 0.9, 1.0])
+			.expect("generate the blue media");
+
+		let project = graphops::create_project();
+		let seq = graphops::create_sequence(&project, "Mid Adjustment Montage");
+		let red_footage = graphops::import_footage(&project, &red).expect("import the red media");
+		let blue_footage = graphops::import_footage(&project, &blue).expect("import the blue media");
+		graphops::place_footage_clip(&project, seq, red_footage, TrackType::Video, 0, 0, 10, 0)
+			.expect("place the V1 (red) clip");
+		graphops::add_track(&project, seq, TrackType::Video).expect("add the adjustment track");
+		let adj_track = {
+			let g = lock(&project);
+			graphops::track_ids(&g.graph, seq, TrackType::Video)[1]
+		};
+		let (_adj, _fx) = adjustment_layer_with_opacity(&project, seq, adj_track, 0, 5, 0.5);
+		// V3's clip carries its own 50% Opacity so the composite keeps both
+		// contributions visible (an opaque top clip would hide the grading).
+		let blue_clip =
+			graphops::place_footage_clip(&project, seq, blue_footage, TrackType::Video, 2, 0, 10, 0)
+				.expect("place the V3 (blue) clip");
+		let clip_fx = crate::oakui::effectchain::insert(
+			&project,
+			blue_clip,
+			usize::MAX,
+			"org.olivevideoeditor.Olive.opacity",
+		)
+		.expect("append the clip's opacity effect");
+		crate::oakui::effectchain::set_input_value(
+			&project,
+			clip_fx,
+			"opacity_in",
+			oak_node::value::NodeValue::Float(0.5),
+		)
+		.expect("set the clip opacity value");
+
+		let tb = graphops::sequence_time_base(&lock(&project).graph, seq).unwrap();
+		let time = graphops::ts_to_rational(0, tb);
+		let render = |montage: Vec<MontageClip>, adjustments: Vec<AdjustmentSpan>, at_time: Rational| {
+			let params = VideoTicketParams {
+				viewer: 0,
+				project: String::new(),
+				time: at_time,
+				force_size: Some((64, 64)),
+				force_format: Some(oak_core::PixelFormat::F32),
+				cache: None,
+				cache_dir: None,
+				cache_id: None,
+				cache_timebase: None,
+				footage: None,
+				montage,
+				adjustments,
+			};
+			let mut dst = vec![0u8; 64 * 64 * 16];
+			oak_render::eval::render_montage_frame_into(
+				at_time,
+				&params,
+				(64, 64),
+				&mut dst,
+				64 * 16,
+			)
+			.expect("montage render");
+			dst
+		};
+		let pixel = |frame: &[u8]| {
+			let off = (8 * 64 + 8) * 16;
+			[0, 1, 2, 3]
+				.map(|i| f32::from_le_bytes(frame[off + i * 4..off + i * 4 + 4].try_into().unwrap()))
+		};
+
+		let (montage, spans) = video_montage_with_adjustments(&project, seq, time);
+		// V2 is taken over by the layer (no clips there), so the montage is
+		// V1's red then V3's blue, and the boundary sits between them.
+		assert_eq!(montage.len(), 2, "the two clips cover frame 0");
+		assert!(montage[0].effects.is_empty(), "the red clip has no stack");
+		assert_eq!(montage[1].effects.len(), 1, "the blue clip's own stack rides it");
+		assert_eq!(spans.len(), 1, "the layer emits one span");
+		assert_eq!(spans[0].track_index, 1, "the layer grades only V1's clip");
+		assert!(!spans[0].effects.is_empty(), "the layer carries its opacity");
+
+		let baseline = pixel(&render(montage.clone(), Vec::new(), time));
+		let graded = pixel(&render(montage.clone(), spans.clone(), time));
+		// Blue at 50% alpha over red: r = 0.05*0.5 + 0.9*0.5 = 0.475,
+		// a = 0.5 + 1.0*0.5 = 1.0. With the layer the red underneath is
+		// halved first: r = 0.05*0.5 + 0.45*0.5 = 0.25, a = 0.5 + 0.5*0.5 = 0.75.
+		assert!(
+			(baseline[0] - 0.475).abs() < 0.06 && (baseline[3] - 1.0).abs() < 0.06,
+			"the plain composite is blue over red (r={}, a={})",
+			baseline[0],
+			baseline[3]
+		);
+		assert!(
+			(graded[0] - 0.25).abs() < 0.06,
+			"only the red underneath is halved (r={}, expected 0.25)",
+			graded[0]
+		);
+		assert!(
+			(graded[3] - 0.75).abs() < 0.06,
+			"the layer's alpha halving shows through the 50% top clip (a={}, expected 0.75)",
+			graded[3]
+		);
+
+		// The layer is a layer only through the range it covers: past frame 5
+		// the clips still composite but no span is emitted, and a stale span
+		// (one carried outside its range) is inert.
+		let late = graphops::ts_to_rational(6, tb);
+		let stale = spans.clone();
+		let (m, s) = video_montage_with_adjustments(&project, seq, late);
+		assert_eq!(m.len(), 2, "the clips still cover frame 6");
+		assert!(s.is_empty(), "the layer's range ends at frame 5");
+		assert_eq!(m[0].effects.len(), 0, "the red clip still has no stack");
+		let stale_render = render(m.clone(), stale, late);
+		assert_eq!(
+			stale_render,
+			render(m, Vec::new(), late),
+			"a span whose range misses the ticket's time renders ungraded"
+		);
+
+		oak_undo::global::clear().unwrap();
+		let _ = std::fs::remove_file(&red);
+		let _ = std::fs::remove_file(&blue);
+	}
+
 	/// A multi-cam HOST clip's montage media is the CURRENT source's angle
 	/// clip (footage + media-in mapped from the host position) — without
 	/// this the host clip's texture chain has no footage and the sequence
@@ -1662,6 +2076,7 @@ mod tests {
 			cache_timebase: None,
 			footage: None,
 			montage,
+			adjustments: Vec::new(),
 		};
 		let mut dst = vec![0u8; 64 * 64 * 16];
 		oak_render::eval::render_montage_frame_into(
@@ -1872,6 +2287,7 @@ mod tests {
 						cache_timebase: None,
 						footage: None,
 						montage,
+						adjustments: Vec::new(),
 					};
 					let mut dst = vec![0u8; 64 * 64 * 16];
 					oak_render::eval::render_montage_frame_into(
diff --git a/crates/oak-app/src/panels/commands.rs b/crates/oak-app/src/panels/commands.rs
index 648b651dd..616c27f1f 100644
--- a/crates/oak-app/src/panels/commands.rs
+++ b/crates/oak-app/src/panels/commands.rs
@@ -135,6 +135,12 @@ pub trait PanelCommandHandler: Sized {
 	fn speed_duration(&mut self, _cx: &mut Context<Self>) -> bool {
 		false
 	}
+	/// The clip menu's "Default Transition" item / the Ctrl+Shift+D action:
+	/// add a transition of the configured default length at every seam
+	/// around the selected clips.
+	fn default_transition(&mut self, _cx: &mut Context<Self>) -> bool {
+		false
+	}
 	fn toggle_links(&mut self, _cx: &mut Context<Self>) -> bool {
 		false
 	}
@@ -259,6 +265,7 @@ pub fn dispatch_to<P: PanelCommandHandler>(
 		ActionId::RippleDelete => panel.ripple_delete(cx),
 		ActionId::SplitAtPlayhead => panel.split_at_playhead(cx),
 		ActionId::SpeedDuration => panel.speed_duration(cx),
+		ActionId::DefaultTransition => panel.default_transition(cx),
 		ActionId::LinkUnlink => panel.toggle_links(cx),
 		ActionId::EnableDisable => panel.toggle_selected_enabled(cx),
 		ActionId::Insert => panel.insert(cx),
diff --git a/crates/oak-app/src/panels/timeline.rs b/crates/oak-app/src/panels/timeline.rs
index fcd42ddae..4626010e3 100644
--- a/crates/oak-app/src/panels/timeline.rs
+++ b/crates/oak-app/src/panels/timeline.rs
@@ -103,6 +103,11 @@ pub struct TimelinePanel<E: AppEngine> {
 	/// The track behind the currently open track-head menu (the "Delete"
 	/// item's target); `None` when a different menu is open.
 	context_track: Option<usize>,
+	/// The empty-area click behind the currently open empty-area menu: the
+	/// display track under the cursor plus the clicked frame ("Add
+	/// Adjustment Layer" anchors there); `None` when a different menu is
+	/// open.
+	context_empty: Option<(usize, Frame)>,
 }
 
 /// A footage drop target resolved from the cursor: the display track under
@@ -198,6 +203,7 @@ impl<E: AppEngine> TimelinePanel<E> {
 			footage_drop: None,
 			context_menu,
 			context_track: None,
+			context_empty: None,
 		}
 	}
 
@@ -240,7 +246,10 @@ impl<E: AppEngine> TimelinePanel<E> {
 				};
 				clip_menu(sync, &proxy, Some(multicam))
 			}
-			TimelineHit::Empty { .. } => empty_area_menu(),
+			TimelineHit::Empty { track, frame } => {
+				self.context_empty = Some((*track, *frame));
+				empty_area_menu()
+			}
 			TimelineHit::TrackHead(track) => {
 				self.context_track = Some(*track);
 				track_head_menu()
@@ -251,6 +260,9 @@ impl<E: AppEngine> TimelinePanel<E> {
 		if !matches!(hit, TimelineHit::TrackHead(_)) {
 			self.context_track = None;
 		}
+		if !matches!(hit, TimelineHit::Empty { .. }) {
+			self.context_empty = None;
+		}
 		self.context_menu.show(position, menu, cx);
 	}
 
@@ -282,6 +294,16 @@ impl<E: AppEngine> TimelinePanel<E> {
 				self.engine
 					.update(cx, |engine, cx| engine.delete_empty_tracks(cx));
 			}
+			LOCAL_ADD_ADJUSTMENT_LAYER => {
+				if let Some((track, frame)) = self.context_empty {
+					if let Err(err) = self
+						.engine
+						.update(cx, |engine, cx| engine.add_adjustment_layer(track, frame, cx))
+					{
+						println!("[timeline] add adjustment layer failed: {err}");
+					}
+				}
+			}
 			LOCAL_CACHE_ALL | LOCAL_CACHE_IN_OUT | LOCAL_CACHE_DISCARD => {
 				println!("[timeline] cache action {item} (not implemented yet)");
 			}
@@ -667,6 +689,18 @@ impl<E: AppEngine> PanelCommandHandler for TimelinePanel<E> {
 			.update(cx, |engine, cx| engine.split_at_playhead(cx));
 		true
 	}
+	/// 编辑 → 设为默认转场 / Ctrl+Shift+D: adds a transition of the
+	/// configured default length at every seam around the selected clips.
+	fn default_transition(&mut self, cx: &mut Context<Self>) -> bool {
+		let ids: Vec<ClipId> = self.timeline.read(cx).selection().iter().copied().collect();
+		let result = self
+			.engine
+			.update(cx, |engine, cx| engine.add_default_transition(ids, cx));
+		if let Err(error) = result {
+			println!("[timeline] add default transition failed: {error}");
+		}
+		true
+	}
 	fn set_marker(&mut self, cx: &mut Context<Self>) -> bool {
 		self.engine
 			.update(cx, |engine, cx| engine.add_marker_at_playhead(cx));
@@ -1109,6 +1143,7 @@ const LOCAL_TIMECODE_FRAMES: usize = 2126;
 const LOCAL_TIMECODE_MILLISECONDS: usize = 2127;
 const LOCAL_ADD_VIDEO_TRACK: usize = 2130;
 const LOCAL_ADD_AUDIO_TRACK: usize = 2131;
+const LOCAL_ADD_ADJUSTMENT_LAYER: usize = 2132;
 
 /// A registry-backed item shown under a "Properties" label (the C++ clip
 /// and sequence "Properties" entries open the Speed/Duration and Sequence
@@ -1294,7 +1329,11 @@ pub(crate) fn empty_area_menu() -> Menu {
 			LOCAL_SHOW_WAVEFORMS,
 			i18n::tr("timeline.context.show_waveforms"),
 		)
-		.with_checked(false)
+		.with_checked(false),
+		MenuItem::new(
+			LOCAL_ADD_ADJUSTMENT_LAYER,
+			i18n::tr("timeline.context.add_adjustment_layer"),
+		)
 		.separated(),
 		properties_item(ActionId::SequenceSettings),
 	])
@@ -1376,7 +1415,8 @@ pub(crate) fn ruler_menu() -> Menu {
 mod tests {
 	use super::*;
 	use crate::oakui::MockEngine;
-	use gpui::{px, size, TestAppContext, VisualTestContext};
+	use gpui::timeline::{FrameRange, TimelineDataSource};
+	use gpui::{px, size, Hsla, TestAppContext, VisualTestContext};
 
 	/// Builds a `TimelinePanel` in a window of the given logical size and
 	/// returns a `VisualTestContext` for bounds assertions.
@@ -1710,8 +1750,9 @@ mod tests {
 		assert!(!item.checked.unwrap_or(false));
 	}
 
-	/// The empty-area menu exposes the view toggles plus the sequence
-	/// settings "Properties" entry.
+	/// The empty-area menu exposes the view toggles, the "Add Adjustment
+	/// Layer" creation entry (§3.5) and the sequence settings "Properties"
+	/// entry.
 	#[test]
 	fn empty_area_menu_toggles_and_properties() {
 		let menu = empty_area_menu();
@@ -1732,10 +1773,89 @@ mod tests {
 		);
 		assert!(sub.iter().all(|item| item.checked == Some(false)));
 
+		// "Add Adjustment Layer" sits between the view toggles and
+		// Properties, carries the i18n label and closes its section so the
+		// tail reads as the properties block.
+		let adjustment = menu
+			.items
+			.iter()
+			.position(|item| item.id == LOCAL_ADD_ADJUSTMENT_LAYER)
+			.expect("add adjustment layer item");
+		let item = &menu.items[adjustment];
+		assert!(item.label == i18n::tr("timeline.context.add_adjustment_layer"));
+		assert!(item.enabled);
+		assert!(item.separator_after);
+		assert_eq!(
+			menu.items[adjustment + 1].id,
+			ActionId::SequenceSettings.entry().menu_id(),
+			"Properties follows the adjustment item"
+		);
+
 		let properties = menu.items.last().expect("properties tail");
 		assert_eq!(properties.id, ActionId::SequenceSettings.entry().menu_id());
 	}
 
+	/// Choosing "Add Adjustment Layer" from the empty-area menu lands the
+	/// demo's five-second layer on the video track under the cursor, at the
+	/// clicked frame and in a color of its own (§3.5).
+	#[gpui::test]
+	async fn empty_area_menu_adds_an_adjustment_layer_at_the_clicked_frame(
+		cx: &mut TestAppContext,
+	) {
+		let (cx, panel) = panel_window(cx, 1600.0, 900.0);
+		// The demo's track 0 is a video track with one clip (标题.mov, frames
+		// 120..300); the created layer must not borrow that clip's color.
+		let demo_colors: Vec<Hsla> = cx.read(|app| {
+			let engine = panel.read(app).engine.clone();
+			let engine = engine.read(app);
+			engine
+				.track(0)
+				.expect("the demo's first track")
+				.clips()
+				.iter()
+				.filter_map(|clip| clip.color())
+				.collect()
+		});
+		assert_eq!(demo_colors.len(), 1, "the demo track has one colored clip");
+
+		// The right-click anchors the menu on (track 0, frame 50); choosing
+		// the item creates the layer there.
+		cx.update(|_window, cx| {
+			panel.update(cx, |panel, cx| {
+				panel.open_context_menu(
+					gpui::point(px(10.0), px(10.0)),
+					TimelineHit::Empty {
+						track: 0,
+						frame: Frame(50),
+					},
+					cx,
+				);
+				panel.on_local_menu_item(LOCAL_ADD_ADJUSTMENT_LAYER, cx);
+			});
+		});
+
+		cx.read(|app| {
+			let engine = panel.read(app).engine.clone();
+			let engine = engine.read(app);
+			let track = engine.track(0).expect("the demo's first track");
+			let clips = track.clips();
+			assert_eq!(clips.len(), 2, "the layer joins the track's demo clip");
+			let layer = &clips[0];
+			assert_eq!(
+				layer.range(),
+				FrameRange::new(Frame(50), Frame(175)),
+				"five seconds (125 frames at 25fps) from the clicked frame"
+			);
+			assert_eq!(layer.label(), SharedString::from("调整图层"));
+			assert_eq!(layer.media_in(), Frame(0));
+			let color = layer.color().expect("the layer carries its own color");
+			assert!(
+				demo_colors.iter().all(|demo| *demo != color),
+				"the layer color is distinct from the demo clips"
+			);
+		});
+	}
+
 	/// The track-header menu is exactly the two delete entries.
 	#[test]
 	fn track_head_menu_offers_add_then_delete_entries() {
diff --git a/crates/oak-cli/src/engine.rs b/crates/oak-cli/src/engine.rs
index 356145d09..a4abb4050 100644
--- a/crates/oak-cli/src/engine.rs
+++ b/crates/oak-cli/src/engine.rs
@@ -657,6 +657,7 @@ pub fn render_frame(
 		cache_timebase: None,
 		footage: None,
 		montage,
+		adjustments: Vec::new(),
 	};
 	let id = m.tickets.next_id();
 	m.tickets.submit_video_with_id(id, params, Box::new(|_| {}));
diff --git a/crates/oak-node/src/block.rs b/crates/oak-node/src/block.rs
index b68ef7736..4d656a571 100644
--- a/crates/oak-node/src/block.rs
+++ b/crates/oak-node/src/block.rs
@@ -136,6 +136,16 @@ pub struct TransitionBlockBehavior {
 	pub out_offset: Rational,
 }
 
+/// Adjustment block behavior: a block spanning a timeline range that hosts
+/// an effect chain (the "adjustment layer"). Its effect chain is attached
+/// exactly like a clip's, through `tex_in`; the source texture is supplied
+/// by the renderer (the composite of the tracks below), so `value()` only
+/// passes a connected texture through and a bare adjustment block is inert.
+pub struct AdjustmentBlockBehavior {
+	/// Block core.
+	pub core: BlockCore,
+}
+
 /// ClipBlock input ids (C++ `clip.cpp`).
 pub mod clip_input {
 	/// `tex_in` (texture, static) — the clip's effect input (C++
@@ -160,6 +170,18 @@ pub mod transition_input {
 	pub const OUT_BLOCK: &str = "out_block_in";
 	/// `in_block_in` (the incoming side).
 	pub const IN_BLOCK: &str = "in_block_in";
+	/// The transition style combo (`crate::nodes::transitions` owns the
+	/// names and the shader ids they map to). Not part of the C++ input
+	/// set — the C++ transition block is cross-dissolve only.
+	pub const TYPE_INPUT: &str = "type_in";
+}
+
+/// AdjustmentBlock input ids.
+pub mod adjustment_input {
+	/// `tex_in` (texture, static) — the adjustment layer's effect input
+	/// (same id every effect node and the clip use, so the effect chain
+	/// machinery works unchanged).
+	pub const TEXTURE_INPUT: &str = "tex_in";
 }
 
 /// Save the shared [`BlockCore`] custom fields (C++ persists the
@@ -270,6 +292,15 @@ impl TransitionBlockBehavior {
 	}
 }
 
+impl AdjustmentBlockBehavior {
+	/// New adjustment layer with a default length of one second.
+	pub fn new() -> Self {
+		AdjustmentBlockBehavior {
+			core: BlockCore::default(),
+		}
+	}
+}
+
 fn block_categories() -> &'static [Category] {
 	&[Category::Timeline]
 }
@@ -498,8 +529,10 @@ impl NodeBehavior for TransitionBlockBehavior {
 		true
 	}
 
-	/// No video output yet (C++ transition crossfades are not ported; a
-	/// transition renders as a hole for now).
+	/// No video output of its own: the renderer blends the two blocks
+	/// this one joins at the track level (`oak_render::eval`'s transition
+	/// step), so the block node itself never produces a texture and a
+	/// track carrying only a transition renders as a hole.
 	fn value(
 		&self,
 		_core: &NodeCore,
@@ -510,6 +543,75 @@ impl NodeBehavior for TransitionBlockBehavior {
 	}
 }
 
+impl NodeBehavior for AdjustmentBlockBehavior {
+	fn name(&self) -> &str {
+		"Adjustment Layer"
+	}
+
+	fn type_id(&self) -> &str {
+		"org.olivevideoeditor.Olive.adjustment"
+	}
+
+	fn categories(&self) -> &[Category] {
+		block_categories()
+	}
+
+	fn as_any(&self) -> Option<&dyn std::any::Any> {
+		Some(self)
+	}
+
+	fn as_any_mut(&mut self) -> Option<&mut dyn std::any::Any> {
+		Some(self)
+	}
+
+	fn duplicate(&self, _core: &NodeCore) -> Option<Box<dyn NodeBehavior>> {
+		Some(Box::new(AdjustmentBlockBehavior {
+			core: self.core.clone(),
+		}))
+	}
+
+	/// Custom project save: the shared block span/state only (the effect
+	/// chain lives in the graph edges, like a clip's).
+	fn save_custom(&self, core: &NodeCore, writer: &mut dyn crate::serializer::XmlWrite) {
+		let _ = core;
+		save_block_core(writer, &self.core);
+	}
+
+	/// Custom project load; the track reference resolves in the
+	/// serializer's post-load pass.
+	fn load_custom(
+		&mut self,
+		_core: &mut NodeCore,
+		reader: &mut dyn crate::serializer::XmlRead,
+	) -> bool {
+		load_block_core(reader, &mut self.core, &mut |_, _| false);
+		true
+	}
+
+	/// Pass the connected texture through (same shape as the clip's
+	/// `value()`): an unconnected `tex_in` yields nothing, so a bare
+	/// adjustment layer is inert. The renderer drives the adjustment by
+	/// feeding the lower composition into the head of the effect chain.
+	fn value(
+		&self,
+		_core: &NodeCore,
+		inputs: &NodeValueRow,
+		_time: Rational,
+		table: &mut NodeValueTable,
+	) {
+		if !self.core.enabled {
+			return;
+		}
+		let Some(value) = inputs.get(adjustment_input::TEXTURE_INPUT) else {
+			return;
+		};
+		// `NodeValue::clone` addrefs the texture handle so the table row
+		// owns its own reference (released on drop); a plain handle copy
+		// would double-release the input's reference.
+		table.push(ValueType::Texture, value.clone(), None);
+	}
+}
+
 /// Constructor for a clip block (C++ `ClipBlock::ClipBlock()`): adds the
 /// static clip inputs (`media_in_in`, `speed_in`, `reverse_in`,
 /// `maintain_audio_pitch_in`, `autocache_in`, `loop_in`) and the texture
@@ -583,7 +685,10 @@ pub fn gap_create() -> (NodeCore, Box<dyn NodeBehavior>) {
 }
 
 /// Constructor for a transition block (C++ `TransitionBlock`): adds the
-/// `out_block_in`/`in_block_in` node-typed connection inputs.
+/// `out_block_in`/`in_block_in` node-typed connection inputs plus the
+/// transition style combo (the C++ block is cross-dissolve only; the
+/// combo lets the composite driver pick between the four shaders in
+/// [`crate::nodes::transitions`]).
 pub fn transition_create() -> (NodeCore, Box<dyn NodeBehavior>) {
 	let mut core = NodeCore::new();
 	let mut out = Input::new(
@@ -604,9 +709,43 @@ pub fn transition_create() -> (NodeCore, Box<dyn NodeBehavior>) {
 	inn.display_name = "To".to_string();
 	core.add_input(inn);
 
+	let mut ty = Input::new(
+		transition_input::TYPE_INPUT,
+		ValueType::Combo,
+		NodeValue::Combo(0),
+	);
+	ty.flags |= crate::input::flags::NOT_CONNECTABLE | crate::input::flags::NOT_KEYFRAMABLE;
+	ty.properties = vec![(
+		"combobox_strings".to_string(),
+		NodeValue::Binary(crate::nodes::transitions::TYPE_NAMES.join(",").into_bytes()),
+	)];
+	core.add_input(ty);
+
 	(core, Box::new(TransitionBlockBehavior::new()))
 }
 
+/// Constructor for an adjustment block (the "adjustment layer"): a single
+/// connectable, non-keyframable `tex_in` texture input that doubles as the
+/// block's effect input. No footage link and no media/speed/reverse inputs
+/// — the block owns only its timeline span.
+pub fn adjustment_create() -> (NodeCore, Box<dyn NodeBehavior>) {
+	let mut core = NodeCore::new();
+
+	// Same convention as `clip_create`: the texture input sits right after
+	// the inherited `enabled_in` so the effect chain attaches at position
+	// one and the renderer can swap the whole chain in and out.
+	let mut tex = Input::new(
+		adjustment_input::TEXTURE_INPUT,
+		ValueType::Texture,
+		NodeValue::None,
+	);
+	tex.flags |= crate::input::flags::NOT_KEYFRAMABLE;
+	core.inputs.insert(1, tex);
+	core.effect_input = adjustment_input::TEXTURE_INPUT.to_string();
+
+	(core, Box::new(AdjustmentBlockBehavior::new()))
+}
+
 #[cfg(test)]
 mod tests {
 	use super::*;
@@ -770,4 +909,71 @@ mod tests {
 			Some(effect_id)
 		);
 	}
+
+	/// The adjustment layer declares one connectable, non-keyframable
+	/// `tex_in` texture input and uses it as the effect input (the clip's
+	/// convention, without the media/speed inputs).
+	#[test]
+	fn adjustment_effect_input_and_texture_input() {
+		let (core, behavior) = adjustment_create();
+		assert_eq!(behavior.name(), "Adjustment Layer");
+		assert_eq!(behavior.type_id(), "org.olivevideoeditor.Olive.adjustment");
+		assert_eq!(core.effect_input, adjustment_input::TEXTURE_INPUT);
+
+		let tex = core
+			.get_input(adjustment_input::TEXTURE_INPUT)
+			.expect("adjustment declares a texture input");
+		assert_eq!(tex.value_type, ValueType::Texture);
+		assert_eq!(tex.default, NodeValue::None);
+		assert_ne!(tex.flags & crate::input::flags::NOT_KEYFRAMABLE, 0);
+		assert!(
+			tex.is_connectable(),
+			"effects attach through the texture input"
+		);
+
+		// `tex_in` sits right after the inherited `enabled_in`, and the
+		// adjustment declares no further inputs.
+		let ids: Vec<&str> = core.inputs.iter().map(|i| i.id.as_str()).collect();
+		assert_eq!(
+			ids,
+			vec![crate::node::ENABLED_INPUT, adjustment_input::TEXTURE_INPUT,]
+		);
+	}
+
+	/// The adjustment layer copies the connected `tex_in` texture to its
+	/// output; disabled and unconnected adjustment layers emit nothing
+	/// (the renderer feeds the chain's head when it drives the block).
+	#[test]
+	fn adjustment_value_passes_connected_texture_only() {
+		let mut inputs = NodeValueRow::new();
+		let handle = crate::handle::make_owned(43i32);
+		inputs.insert(
+			adjustment_input::TEXTURE_INPUT.to_string(),
+			NodeValue::Texture(handle),
+		);
+
+		let behavior = AdjustmentBlockBehavior::new();
+		let mut table = NodeValueTable::default();
+		behavior.value(&NodeCore::empty(), &inputs, Rational::new(0, 1), &mut table);
+		assert_eq!(table.count(), 1);
+		let NodeValue::Texture(out) = table.get(ValueType::Texture).unwrap() else {
+			unreachable!()
+		};
+		assert_eq!(out.ctx, handle.ctx, "the same texture box passes through");
+
+		let mut disabled = AdjustmentBlockBehavior::new();
+		disabled.core.enabled = false;
+		let mut table = NodeValueTable::default();
+		disabled.value(&NodeCore::empty(), &inputs, Rational::new(0, 1), &mut table);
+		assert_eq!(table.count(), 0, "disabled adjustment emits nothing");
+
+		let mut table = NodeValueTable::default();
+		behavior.value(
+			&NodeCore::empty(),
+			&NodeValueRow::new(),
+			Rational::new(0, 1),
+			&mut table,
+		);
+		assert_eq!(table.count(), 0, "unconnected adjustment emits nothing");
+	}
 }
diff --git a/crates/oak-node/src/nodes/compositesource.rs b/crates/oak-node/src/nodes/compositesource.rs
new file mode 100644
index 000000000..b4a93b870
--- /dev/null
+++ b/crates/oak-node/src/nodes/compositesource.rs
@@ -0,0 +1,184 @@
+// Oak Video Editor - Non-Linear Video Editor
+// Copyright (C) 2026 Oak Team
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+//! Composite source (Rust-only render seam; no C++ counterpart).
+//!
+//! The graph renderer builds one of these per adjustment-layer flush: it
+//! prefills `tex_in` with the composite of the tracks below the layer
+//! (a standard value, so nothing has to be wired upstream) and connects
+//! the node to the head of the layer's effect chain for the duration of
+//! one [`oak_node::traverser::Traverser::evaluate`] pass. `value()`
+//! pushes that texture straight into the output table — the effect chain
+//! downstream sees an ordinary texture input, so nested shader jobs,
+//! plugin jobs and the pass-through case all work unchanged.
+//!
+//! It is hidden from the create menu (users never place one; the renderer
+//! owns its lifetime — see `oakrender::eval::render_graph_frame`) and has
+//! no shader of its own. It is deliberately absent from the factory table:
+//! `factory_smoke_test` pins that table to the C++ registration order, and
+//! this type has no C++ counterpart — the renderer reaches it through
+//! [`create`] directly.
+
+use crate::node::{Category, NodeBehavior, NodeCore};
+
+/// Stable type id (`org.olivevideoeditor.Olive.composite_source`).
+pub const TYPE_ID: &str = "org.olivevideoeditor.Olive.composite_source";
+
+/// Texture input id: the prefilled source texture (same id every effect
+/// node uses, so it can feed an effect chain's head input directly).
+pub const TEXTURE_INPUT: &str = "tex_in";
+
+/// The texture source node. Has no member fields.
+pub struct CompositeSource;
+
+impl NodeBehavior for CompositeSource {
+	/// Human-readable name.
+	fn name(&self) -> &str {
+		"Composite Source"
+	}
+
+	/// Stable type id.
+	fn type_id(&self) -> &str {
+		TYPE_ID
+	}
+
+	/// Categories (never used for menus — the node carries
+	/// `DONT_SHOW_IN_CREATE_MENU`).
+	fn categories(&self) -> &[Category] {
+		&[Category::Generator]
+	}
+
+	/// Description.
+	fn description(&self) -> &str {
+		"Supplies a renderer-composed texture to an effect chain."
+	}
+
+	/// Localized input names: `tex_in` -> "Texture".
+	fn input_name<'a>(&self, id: &'a str) -> &'a str {
+		match id {
+			TEXTURE_INPUT => "Texture",
+			_ => id,
+		}
+	}
+
+	/// Evaluate outputs: push the prefilled texture. The render driver
+	/// sets it as the standard value of the (unconnected) `tex_in`, which
+	/// the traverser's row builder hands over as an ordinary texture row
+	/// entry; a bare node (no prefill) yields nothing.
+	fn value(
+		&self,
+		_core: &NodeCore,
+		inputs: &crate::value::NodeValueRow,
+		_time: oak_core::Rational,
+		table: &mut crate::value::NodeValueTable,
+	) {
+		let Some(value) = inputs.get(TEXTURE_INPUT) else {
+			return;
+		};
+		// `NodeValue::clone` addrefs the texture handle so the table owns
+		// its own reference (released on drop); a plain handle copy would
+		// double-release the row's reference.
+		table.push(crate::value::ValueType::Texture, value.clone(), None);
+	}
+
+	/// Deep copy.
+	fn duplicate(&self, _core: &NodeCore) -> Option<Box<dyn NodeBehavior>> {
+		Some(Box::new(CompositeSource))
+	}
+}
+
+/// Constructor: one connectable, non-keyframable `tex_in` texture input
+/// (default `None` — the renderer always prefills it) and the
+/// create-menu hiding flag.
+pub fn create() -> (NodeCore, Box<dyn NodeBehavior>) {
+	let mut core = NodeCore::new();
+	let mut tex = crate::input::Input::new(
+		TEXTURE_INPUT,
+		crate::value::ValueType::Texture,
+		crate::value::NodeValue::None,
+	);
+	tex.flags |= crate::input::flags::NOT_KEYFRAMABLE;
+	core.add_input(tex);
+	core.flags |= crate::node::flags::DONT_SHOW_IN_CREATE_MENU;
+	core.effect_input = TEXTURE_INPUT.to_string();
+	(core, Box::new(CompositeSource))
+}
+
+#[cfg(test)]
+mod tests {
+	use super::*;
+	use crate::input::flags::NOT_KEYFRAMABLE;
+	use crate::node::flags::DONT_SHOW_IN_CREATE_MENU;
+	use crate::value::{NodeValue, NodeValueRow, NodeValueTable, ValueType};
+
+	#[test]
+	fn create_wires_inputs_and_flags() {
+		let (core, behavior) = create();
+		assert_eq!(behavior.type_id(), TYPE_ID);
+		assert_eq!(behavior.name(), "Composite Source");
+		let tex = core.get_input(TEXTURE_INPUT).expect("tex_in");
+		assert_eq!(tex.value_type, ValueType::Texture);
+		assert_eq!(tex.default, NodeValue::None);
+		assert_eq!(tex.flags & NOT_KEYFRAMABLE, NOT_KEYFRAMABLE);
+		assert_eq!(
+			core.flags & DONT_SHOW_IN_CREATE_MENU,
+			DONT_SHOW_IN_CREATE_MENU
+		);
+		assert_eq!(core.effect_input, TEXTURE_INPUT);
+	}
+
+	#[test]
+	fn input_names() {
+		let n = CompositeSource;
+		assert_eq!(n.input_name(TEXTURE_INPUT), "Texture");
+	}
+
+	#[test]
+	fn value_pushes_row_texture() {
+		let (core, behavior) = create();
+		let texture = oak_core::texture::Texture::dummy();
+		let inputs = NodeValueRow::from([(
+			TEXTURE_INPUT.to_string(),
+			NodeValue::Texture(crate::handle::make_owned(texture)),
+		)]);
+		let mut table = NodeValueTable::default();
+		behavior.value(&core, &inputs, oak_core::Rational::new(0, 1), &mut table);
+		assert!(matches!(
+			table.get(ValueType::Texture),
+			Some(NodeValue::Texture(_))
+		));
+	}
+
+	#[test]
+	fn value_without_texture_pushes_nothing() {
+		let (core, behavior) = create();
+		let mut table = NodeValueTable::default();
+		behavior.value(
+			&core,
+			&NodeValueRow::new(),
+			oak_core::Rational::new(0, 1),
+			&mut table,
+		);
+		assert!(table.get(ValueType::Texture).is_none());
+	}
+
+	#[test]
+	fn duplicate_clones() {
+		let (core, behavior) = create();
+		let dup = behavior.duplicate(&core).unwrap();
+		assert_eq!(dup.type_id(), TYPE_ID);
+	}
+}
diff --git a/crates/oak-node/src/nodes/mod.rs b/crates/oak-node/src/nodes/mod.rs
index 1ed5c29d0..d35e75e8b 100644
--- a/crates/oak-node/src/nodes/mod.rs
+++ b/crates/oak-node/src/nodes/mod.rs
@@ -22,6 +22,7 @@
 pub mod blur;
 mod chromakey;
 mod colordifferencekey;
+pub mod compositesource;
 mod cornerpindistortnode;
 mod cropdistortnode;
 mod despill;
@@ -52,7 +53,7 @@ mod shapenodebase;
 mod solid;
 mod stroke;
 mod swirldistortnode;
-mod textbackend;
+pub mod textbackend;
 mod textv1;
 mod textv2;
 mod textv3;
@@ -62,12 +63,35 @@ mod timeformat;
 mod timeinput;
 mod timeoffsetnode;
 mod timeremap;
+pub mod transitionfx;
+pub mod transitions;
 mod transformdistortnode;
 mod trigonometry;
 mod valuenode;
 mod volume;
 mod wavedistortnode;
 mod whitebalance;
+mod colormatrix;
+mod dilate;
+mod edgedetect;
+mod erode;
+mod colorcorrect;
+mod gamma;
+mod saturation;
+mod invert;
+mod clamp;
+mod grade;
+mod dirblur;
+mod sharpen;
+mod dissolve;
+mod keymix;
+mod premult;
+mod unpremult;
+mod position;
+mod mirror;
+mod checkerboard;
+mod colorbars;
+mod ramp;
 
 use crate::node::{NodeBehavior, NodeCore};
 
@@ -123,6 +147,8 @@ pub fn register_all() {
 	flipdistortnode::register(&mut meta);
 	noise::register(&mut meta);
 	timeoffsetnode::register(&mut meta);
+	transitions::register(&mut meta);
+	transitionfx::register(&mut meta);
 	cornerpindistortnode::register(&mut meta);
 	displaytransform::register(&mut meta);
 	ociogradingtransformlinear::register(&mut meta);
@@ -138,6 +164,30 @@ pub fn register_all() {
 	tiledistortnode::register(&mut meta);
 	swirldistortnode::register(&mut meta);
 	rippledistortnode::register(&mut meta);
+	colormatrix::register(&mut meta);
+	dilate::register(&mut meta);
+	edgedetect::register(&mut meta);
+	erode::register(&mut meta);
+	// The OpenFX-Misc cleanroom GPU ports (see
+	// docs/zh/plans/ofx-misc-gpu-cleanroom.md), registered after the C++
+	// menu order ends and before multicam.
+	colorcorrect::register(&mut meta);
+	gamma::register(&mut meta);
+	saturation::register(&mut meta);
+	invert::register(&mut meta);
+	clamp::register(&mut meta);
+	grade::register(&mut meta);
+	dirblur::register(&mut meta);
+	sharpen::register(&mut meta);
+	dissolve::register(&mut meta);
+	keymix::register(&mut meta);
+	premult::register(&mut meta);
+	unpremult::register(&mut meta);
+	position::register(&mut meta);
+	mirror::register(&mut meta);
+	checkerboard::register(&mut meta);
+	colorbars::register(&mut meta);
+	ramp::register(&mut meta);
 	multicamnode::register(&mut meta);
 
 	// OpenFX plugins have no static type ids (C++
diff --git a/crates/oak-node/src/nodes/transitionfx.rs b/crates/oak-node/src/nodes/transitionfx.rs
new file mode 100644
index 000000000..8e5cb52ed
--- /dev/null
+++ b/crates/oak-node/src/nodes/transitionfx.rs
@@ -0,0 +1,766 @@
+// Oak Video Editor - Non-Linear Video Editor
+// Copyright (C) 2026 Oak Team
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+//! Layer transition node: the adjustment-layer form of a transition
+//! (`docs/zh/plans/adjustment-layers-and-transitions.md` §4.2), the
+//! counterpart of the timeline transition *block* whose node lives in
+//! [`crate::nodes::transitions`].
+//!
+//! The node sits in the effect chain of an adjustment layer covering the
+//! seam. `tex_in` is the chain input (`core.effect_input`), so the render
+//! driver feeds it the composited lower layers; `blend_in` is the second
+//! texture — the pre-seam block's output by convention — bound by name
+//! from the job row, the same dual-texture mode [`crate::nodes::merge`]
+//! uses. [`crate::nodes::transitions`] boxes its job the same way; the
+//! difference is the chain input, which the timeline transition block
+//! does not have (its two sides arrive as a pair).
+//!
+//! `progress_in` is the layer's position across the adjustment block's
+//! own span. The renderer fills it in for the duration of an adjustment
+//! sweep (`oakrender::eval`'s `hooks.layer_progress`, which pre-fills the
+//! row of any job whose shader declares the uniform and whose params do
+//! not already carry it), so the node writes `progress_in` into the job
+//! row **only when the user set it explicitly** — a row carrying the
+//! default `0.0` would otherwise pin the node to the very start of the
+//! span. [`TransitionFxNode::value`] documents the "explicit" test.
+//!
+//! One behavior carries all four styles ([`TYPE_NAMES`]), dispatched by
+//! the `type_in` combo through its shader id ([`SHADER_IDS`]) — the
+//! `node->get_shader_code(shader_id)` path the math nodes and the
+//! timeline transition node use. The fragment sources are the timeline
+//! node's, adapted to the one-picture chain form: the fade mixes against
+//! its own `color_in` (the layer form has a single picture, so there is
+//! no dip between two of them), and the wipe/slide styles take a
+//! `direction_in` sweep direction.
+
+use crate::factory::NodeMeta;
+use crate::jobs::ShaderJobPayload;
+use crate::node::{Category, NodeBehavior, NodeCore};
+
+/// Chain texture input id. Type: texture; flags: not-keyframable. This
+/// is the node's effect input.
+pub const TEXTURE_INPUT: &str = "tex_in";
+
+/// Second texture input id (the pre-seam picture by convention). Type:
+/// texture; flags: not-keyframable.
+pub const BLEND_INPUT: &str = "blend_in";
+
+/// Transition progress input id. Type: float; default `0.0`;
+/// properties: `min = 0.0`, `max = 1.0`, `view = percentage`. `0.0`
+/// yields `tex_in`, `1.0` yields `blend_in`. Left out of the job row
+/// unless the user set it, so the renderer can fill in the adjustment
+/// layer's progress instead.
+pub const PROGRESS_INPUT: &str = "progress_in";
+
+/// Fade target color input id. Type: color; default opaque black. Only
+/// the fade style samples it (`progress_in = 0` is all `color_in`,
+/// `1` is the input untouched).
+pub const COLOR_INPUT: &str = "color_in";
+
+/// Wipe/slide sweep direction input id. Type: combo; flags:
+/// not-connectable, not-keyframable; properties: `combobox_strings` =
+/// [`DIRECTION_NAMES`].
+pub const DIRECTION_INPUT: &str = "direction_in";
+
+/// Transition style input id. Type: combo; flags: not-connectable,
+/// not-keyframable; properties: `combobox_strings` = [`TYPE_NAMES`].
+pub const TYPE_INPUT: &str = "type_in";
+
+/// Localized style names, in combo-index order.
+pub const TYPE_NAMES: [&str; 4] = ["Cross Dissolve", "Fade", "Wipe", "Slide"];
+
+/// Shader ids for [`TYPE_NAMES`], by combo index. The strings match the
+/// timeline transition block's (`crate::nodes::transitions::SHADER_IDS`),
+/// so both transition forms ask the factory for the same style.
+pub const SHADER_IDS: [&str; 4] = ["crossdissolve", "fade", "wipe", "slide"];
+
+/// Sweep direction names for [`DIRECTION_INPUT`], in combo-index order:
+/// the side the boundary sweeps towards. The incoming picture is
+/// revealed from the opposite edge (`Left to Right` wipes the `blend_in`
+/// picture in from the left, `blend_in` replacing `tex_in` behind the
+/// boundary).
+pub const DIRECTION_NAMES: [&str; 4] = [
+	"Left to Right",
+	"Right to Left",
+	"Top to Bottom",
+	"Bottom to Top",
+];
+
+/// The shader id for a combo index, clamping out-of-range values onto
+/// cross dissolve (index `0`), the C++ default transition.
+pub fn shader_id_for(index: i64) -> &'static str {
+	SHADER_IDS[index.clamp(0, SHADER_IDS.len() as i64 - 1) as usize]
+}
+
+/// Cross dissolve: a straight mix, `50/50` at the midpoint.
+const SHADER_CROSSDISSOLVE: &str = r#"uniform sampler2D tex_in;
+uniform sampler2D blend_in;
+uniform bool tex_in_enabled;
+uniform bool blend_in_enabled;
+uniform float progress_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled && !blend_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    if (!tex_in_enabled) {
+        frag_color = texture(blend_in, ove_texcoord);
+        return;
+    }
+
+    if (!blend_in_enabled) {
+        frag_color = texture(tex_in, ove_texcoord);
+        return;
+    }
+
+    vec4 tex_col = texture(tex_in, ove_texcoord);
+    vec4 blend_col = texture(blend_in, ove_texcoord);
+
+    frag_color = mix(tex_col, blend_col, progress_in);
+}
+"#;
+
+/// Fade: the input fades up out of `color_in` (black by default) —
+/// `progress_in = 0` is the fade color, `1` the picture untouched. The
+/// layer form has one picture (its `blend_in` is the optional pre-seam
+/// reference), so the two-sided dip of the timeline block's fade has no
+/// pair to dip through here.
+const SHADER_FADE: &str = r#"uniform sampler2D tex_in;
+uniform bool tex_in_enabled;
+uniform vec4 color_in;
+uniform float progress_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    vec4 tex_col = texture(tex_in, ove_texcoord);
+
+    frag_color = mix(color_in, tex_col, progress_in);
+}
+"#;
+
+/// Wipe: a soft-edged boundary sweeps across the frame, the incoming
+/// picture following behind it. The direction combo picks the axis and
+/// the sign with two steps and two mixes — no branch on the index, so
+/// all four directions are one path.
+const SHADER_WIPE: &str = r#"uniform sampler2D tex_in;
+uniform sampler2D blend_in;
+uniform bool tex_in_enabled;
+uniform bool blend_in_enabled;
+uniform float progress_in;
+uniform float direction_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled && !blend_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    if (!tex_in_enabled) {
+        frag_color = texture(blend_in, ove_texcoord);
+        return;
+    }
+
+    if (!blend_in_enabled) {
+        frag_color = texture(tex_in, ove_texcoord);
+        return;
+    }
+
+    vec4 tex_col = texture(tex_in, ove_texcoord);
+    vec4 blend_col = texture(blend_in, ove_texcoord);
+
+    // Direction 0..3: 0/1 are horizontal, 2/3 vertical (the axis), and
+    // the odd entries sweep the other way (the sign).
+    float d = clamp(direction_in, 0.0, 3.0);
+    float horiz = 1.0 - step(1.5, d);
+    float neg = step(1.0, mod(d, 2.0));
+    float sweep = mix(ove_texcoord.y, ove_texcoord.x, horiz);
+    float boundary = mix(sweep, 1.0 - sweep, neg);
+
+    float soft = 0.02;
+    float mask = smoothstep(progress_in - soft, progress_in + soft, boundary);
+
+    frag_color = mix(blend_col, tex_col, mask);
+}
+"#;
+
+/// Slide: the incoming picture pushes the outgoing one off the frame,
+/// both travelling the direction the combo names. The offsets mirror
+/// each other (`-e * progress` against `+e * (1 - progress)`), so the
+/// pair stays adjacent and neither picture is stretched.
+const SHADER_SLIDE: &str = r#"uniform sampler2D tex_in;
+uniform sampler2D blend_in;
+uniform bool tex_in_enabled;
+uniform bool blend_in_enabled;
+uniform float progress_in;
+uniform float direction_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled && !blend_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    if (!tex_in_enabled) {
+        frag_color = texture(blend_in, ove_texcoord);
+        return;
+    }
+
+    if (!blend_in_enabled) {
+        frag_color = texture(tex_in, ove_texcoord);
+        return;
+    }
+
+    // Direction 0..3: 0/1 are horizontal, 2/3 vertical (the axis), and
+    // the odd entries travel the other way (the sign).
+    float d = clamp(direction_in, 0.0, 3.0);
+    float horiz = 1.0 - step(1.5, d);
+    float neg = step(1.0, mod(d, 2.0));
+    vec2 e = vec2(horiz, 1.0 - horiz) * (1.0 - 2.0 * neg);
+
+    float sweep = mix(ove_texcoord.y, ove_texcoord.x, horiz);
+    float boundary = mix(sweep, 1.0 - sweep, neg);
+
+    vec2 from_uv = ove_texcoord - e * progress_in;
+    vec2 to_uv = ove_texcoord + e * (1.0 - progress_in);
+
+    vec4 tex_col = texture(tex_in, from_uv);
+    vec4 blend_col = texture(blend_in, to_uv);
+
+    if (boundary < progress_in) {
+        frag_color = blend_col;
+    } else {
+        frag_color = tex_col;
+    }
+}
+"#;
+
+/// Layer transition node. Unit-like: the style rides in `type_in`, so
+/// there is no per-instance state.
+pub struct TransitionFxNode;
+
+impl TransitionFxNode {
+	/// The fragment source for a shader id; an unknown id falls back to
+	/// cross dissolve.
+	fn shader_frag(shader_id: &str) -> &'static str {
+		match shader_id {
+			"fade" => SHADER_FADE,
+			"wipe" => SHADER_WIPE,
+			"slide" => SHADER_SLIDE,
+			_ => SHADER_CROSSDISSOLVE,
+		}
+	}
+}
+
+impl NodeBehavior for TransitionFxNode {
+	/// Human-readable name.
+	fn name(&self) -> &str {
+		"Transition FX"
+	}
+
+	/// Stable type id.
+	fn type_id(&self) -> &str {
+		"org.olivevideoeditor.Olive.transitionfx"
+	}
+
+	/// Categories. Filed under effect: the `Category` enum has no
+	/// transition group, and this is a video effect on a chain (the
+	/// timeline transition block carries `Category::Timeline` instead).
+	fn categories(&self) -> &[Category] {
+		&[Category::Effect]
+	}
+
+	/// Description.
+	fn description(&self) -> &str {
+		"Blend a picture with a second one across an adjustment layer's span."
+	}
+
+	/// Localized input names: `tex_in` -> "From", `blend_in` -> "To",
+	/// `progress_in` -> "Progress", `color_in` -> "Color",
+	/// `direction_in` -> "Direction", `type_in` -> "Type".
+	fn input_name<'a>(&self, id: &'a str) -> &'a str {
+		match id {
+			TEXTURE_INPUT => "From",
+			BLEND_INPUT => "To",
+			PROGRESS_INPUT => "Progress",
+			COLOR_INPUT => "Color",
+			DIRECTION_INPUT => "Direction",
+			TYPE_INPUT => "Type",
+			_ => id,
+		}
+	}
+
+	/// Evaluate outputs: if only one texture is present, push it as-is
+	/// (a transition against nothing is the input itself); if both are
+	/// present, push one shader job over the whole input row; if
+	/// neither, push nothing.
+	///
+	/// The shader id the `type_in` combo selects is written into the job
+	/// row. `progress_in` is deliberately **not** written when the user
+	/// never set it: the renderer pre-fills the row from the adjustment
+	/// layer's own position (`hooks.layer_progress`) and only a missing
+	/// param lets that through. "Set" means the row value differs from
+	/// the `0.0` default or the input is keyframed — the standard-value
+	/// map cannot answer it, because the project serializer writes a
+	/// `<standard>` element for every input, so a reloaded node always
+	/// has one.
+	fn value(
+		&self,
+		core: &NodeCore,
+		inputs: &crate::value::NodeValueRow,
+		time: oak_core::Rational,
+		table: &mut crate::value::NodeValueTable,
+	) {
+		use crate::value::{NodeValue, ValueType};
+
+		let tex = inputs.get(TEXTURE_INPUT);
+		let blend = inputs.get(BLEND_INPUT);
+
+		match (tex, blend) {
+			(Some(NodeValue::Texture(_)), Some(NodeValue::Texture(_))) => {
+				let ty = match inputs.get(TYPE_INPUT) {
+					Some(v) => v.to_double() as i64,
+					None => core.value_at_time(TYPE_INPUT, -1, time).to_double() as i64,
+				};
+
+				let mut params = inputs.clone();
+				params.insert(TYPE_INPUT.to_string(), NodeValue::Combo(ty));
+
+				let progress_set = inputs
+					.get(PROGRESS_INPUT)
+					.map(|v| v.to_double() != 0.0)
+					.unwrap_or(false)
+					|| core.is_input_keyframing(PROGRESS_INPUT, -1);
+				if !progress_set {
+					params.remove(PROGRESS_INPUT);
+				}
+
+				table.push(
+					ValueType::Texture,
+					NodeValue::Texture(crate::handle::make_owned(ShaderJobPayload {
+						node_id: crate::id::NodeId::INVALID,
+						time,
+						iterations: 1,
+						type_id: self.type_id().to_string(),
+						shader_id: shader_id_for(ty).to_string(),
+						effect_input: core.effect_input.clone(),
+						params,
+						iterative_input: String::new(),
+					})),
+					None,
+				);
+			}
+			(Some(t @ NodeValue::Texture(_)), _) => {
+				table.push(ValueType::Texture, t.clone(), None);
+			}
+			(_, Some(b @ NodeValue::Texture(_))) => {
+				table.push(ValueType::Texture, b.clone(), None);
+			}
+			_ => {}
+		}
+	}
+
+	/// Shader code request: the fragment source for the requested style
+	/// id (see [`SHADER_IDS`]).
+	fn shader_code(&self, request: &str) -> Option<String> {
+		Some(Self::shader_frag(request).to_string())
+	}
+
+	/// Deep copy.
+	fn duplicate(&self, _core: &NodeCore) -> Option<Box<dyn NodeBehavior>> {
+		Some(Box::new(TransitionFxNode))
+	}
+}
+
+/// Constructor: adds the two textures, `progress_in`, `color_in`,
+/// `direction_in` and `type_in` with the defaults and properties
+/// documented on the constants, and sets the video-effect flag with
+/// `tex_in` as the effect input (the chain hangs off it; `blend_in`
+/// binds by name alongside).
+pub fn create() -> (NodeCore, Box<dyn NodeBehavior>) {
+	let mut core = NodeCore::new();
+
+	let mut tex = crate::input::Input::new(
+		TEXTURE_INPUT,
+		crate::value::ValueType::Texture,
+		crate::value::NodeValue::None,
+	);
+	tex.flags |= crate::input::flags::NOT_KEYFRAMABLE;
+	core.add_input(tex);
+
+	let mut blend = crate::input::Input::new(
+		BLEND_INPUT,
+		crate::value::ValueType::Texture,
+		crate::value::NodeValue::None,
+	);
+	blend.flags |= crate::input::flags::NOT_KEYFRAMABLE;
+	core.add_input(blend);
+
+	let mut progress = crate::input::Input::new(
+		PROGRESS_INPUT,
+		crate::value::ValueType::Float,
+		crate::value::NodeValue::Float(0.0),
+	);
+	progress.properties = vec![
+		("min".to_string(), crate::value::NodeValue::Float(0.0)),
+		("max".to_string(), crate::value::NodeValue::Float(1.0)),
+		(
+			"view".to_string(),
+			crate::value::NodeValue::Text("percentage".into()),
+		),
+	];
+	core.add_input(progress);
+
+	let color = crate::input::Input::new(
+		COLOR_INPUT,
+		crate::value::ValueType::Color,
+		crate::value::NodeValue::Color([0.0, 0.0, 0.0, 1.0]),
+	);
+	core.add_input(color);
+
+	let mut direction = crate::input::Input::new(
+		DIRECTION_INPUT,
+		crate::value::ValueType::Combo,
+		crate::value::NodeValue::Combo(0),
+	);
+	direction.flags |= crate::input::flags::NOT_CONNECTABLE | crate::input::flags::NOT_KEYFRAMABLE;
+	direction.properties = vec![(
+		"combobox_strings".to_string(),
+		crate::value::NodeValue::Binary(DIRECTION_NAMES.join(",").into_bytes()),
+	)];
+	core.add_input(direction);
+
+	let mut ty = crate::input::Input::new(
+		TYPE_INPUT,
+		crate::value::ValueType::Combo,
+		crate::value::NodeValue::Combo(0),
+	);
+	ty.flags |= crate::input::flags::NOT_CONNECTABLE | crate::input::flags::NOT_KEYFRAMABLE;
+	ty.properties = vec![(
+		"combobox_strings".to_string(),
+		crate::value::NodeValue::Binary(TYPE_NAMES.join(",").into_bytes()),
+	)];
+	core.add_input(ty);
+
+	core.flags |= crate::node::flags::VIDEO_EFFECT;
+	core.effect_input = TEXTURE_INPUT.to_string();
+
+	(core, Box::new(TransitionFxNode))
+}
+
+#[cfg(test)]
+mod tests {
+	use super::*;
+	use crate::keyframe::{Interpolation, Keyframe};
+	use crate::value::{NodeValue, NodeValueRow, NodeValueTable, ValueType};
+	use oak_core::Rational;
+
+	fn tex() -> NodeValue {
+		NodeValue::Texture(crate::handle::CHandle::null())
+	}
+
+	fn run(core: &NodeCore, inputs: &NodeValueRow) -> ShaderJobPayload {
+		let behavior = TransitionFxNode;
+		let mut table = NodeValueTable::default();
+		behavior.value(core, inputs, Rational::new(0, 1), &mut table);
+		let Some(NodeValue::Texture(h)) = table.get(ValueType::Texture) else {
+			panic!("shader job expected");
+		};
+		unsafe { crate::handle::get_checked::<ShaderJobPayload>(h) }
+			.expect("shader job payload boxed")
+			.clone()
+	}
+
+	/// The row the graph traverser builds for an unconnected node with
+	/// both textures: every input is present, `progress_in` carrying the
+	/// input default.
+	fn both() -> NodeValueRow {
+		NodeValueRow::from([
+			(TEXTURE_INPUT.to_string(), tex()),
+			(BLEND_INPUT.to_string(), tex()),
+			(PROGRESS_INPUT.to_string(), NodeValue::Float(0.0)),
+			(COLOR_INPUT.to_string(), NodeValue::Color([0.0, 0.0, 0.0, 1.0])),
+			(DIRECTION_INPUT.to_string(), NodeValue::Combo(0)),
+		])
+	}
+
+	#[test]
+	fn input_names() {
+		let n = TransitionFxNode;
+		assert_eq!(n.input_name(TEXTURE_INPUT), "From");
+		assert_eq!(n.input_name(BLEND_INPUT), "To");
+		assert_eq!(n.input_name(PROGRESS_INPUT), "Progress");
+		assert_eq!(n.input_name(COLOR_INPUT), "Color");
+		assert_eq!(n.input_name(DIRECTION_INPUT), "Direction");
+		assert_eq!(n.input_name(TYPE_INPUT), "Type");
+		assert_eq!(n.input_name("other"), "other");
+	}
+
+	#[test]
+	fn create_wires_inputs_and_flags() {
+		let (core, behavior) = create();
+		assert_eq!(
+			behavior.type_id(),
+			"org.olivevideoeditor.Olive.transitionfx"
+		);
+		assert_eq!(behavior.name(), "Transition FX");
+		assert_eq!(behavior.categories(), &[Category::Effect]);
+		assert_ne!(core.flags & crate::node::flags::VIDEO_EFFECT, 0);
+		assert_eq!(core.effect_input, TEXTURE_INPUT, "the chain hangs off tex_in");
+		assert_eq!(
+			core.get_input(TEXTURE_INPUT).map(|i| i.value_type),
+			Some(ValueType::Texture)
+		);
+
+		for id in [TEXTURE_INPUT, BLEND_INPUT] {
+			let input = core.get_input(id).expect("texture input");
+			assert_ne!(input.flags & crate::input::flags::NOT_KEYFRAMABLE, 0);
+		}
+
+		let progress = core.get_input(PROGRESS_INPUT).expect("progress input");
+		assert_eq!(progress.default, NodeValue::Float(0.0));
+		assert!(progress
+			.properties
+			.iter()
+			.any(|(k, v)| k == "min" && *v == NodeValue::Float(0.0)));
+		assert!(progress
+			.properties
+			.iter()
+			.any(|(k, v)| k == "max" && *v == NodeValue::Float(1.0)));
+
+		let color = core.get_input(COLOR_INPUT).expect("color input");
+		assert_eq!(color.default, NodeValue::Color([0.0, 0.0, 0.0, 1.0]));
+
+		let direction = core.get_input(DIRECTION_INPUT).expect("direction input");
+		assert_eq!(direction.default, NodeValue::Combo(0));
+		assert_ne!(direction.flags & crate::input::flags::NOT_CONNECTABLE, 0);
+		assert_eq!(
+			direction.properties.first(),
+			Some(&(
+				"combobox_strings".to_string(),
+				NodeValue::Binary(DIRECTION_NAMES.join(",").into_bytes())
+			))
+		);
+
+		let ty = core.get_input(TYPE_INPUT).expect("type input");
+		assert_eq!(ty.default, NodeValue::Combo(0));
+		assert_ne!(ty.flags & crate::input::flags::NOT_CONNECTABLE, 0);
+		assert_eq!(
+			ty.properties.first(),
+			Some(&(
+				"combobox_strings".to_string(),
+				NodeValue::Binary("Cross Dissolve,Fade,Wipe,Slide".as_bytes().to_vec())
+			))
+		);
+	}
+
+	#[test]
+	fn shader_ids_cover_every_type_name() {
+		assert_eq!(TYPE_NAMES.len(), SHADER_IDS.len());
+		assert_eq!(shader_id_for(0), "crossdissolve");
+		assert_eq!(shader_id_for(1), "fade");
+		assert_eq!(shader_id_for(2), "wipe");
+		assert_eq!(shader_id_for(3), "slide");
+		assert_eq!(shader_id_for(-1), "crossdissolve");
+		assert_eq!(shader_id_for(7), "slide");
+	}
+
+	#[test]
+	fn direction_names_cover_every_index() {
+		assert_eq!(DIRECTION_NAMES.len(), 4);
+		for (i, name) in DIRECTION_NAMES.iter().enumerate() {
+			assert!(!name.is_empty(), "{i} has a label");
+		}
+	}
+
+	#[test]
+	fn value_no_texture_pushes_nothing() {
+		let (core, behavior) = create();
+		let mut table = NodeValueTable::default();
+		behavior.value(&core, &NodeValueRow::default(), Rational::new(0, 1), &mut table);
+		assert!(table.is_empty());
+	}
+
+	#[test]
+	fn value_tex_only_pushes_tex() {
+		let (core, behavior) = create();
+		let input = tex();
+		let inputs = NodeValueRow::from([(TEXTURE_INPUT.to_string(), input.clone())]);
+		let mut table = NodeValueTable::default();
+		behavior.value(&core, &inputs, Rational::new(0, 1), &mut table);
+		assert_eq!(table.get(ValueType::Texture), Some(&input));
+	}
+
+	#[test]
+	fn value_blend_only_pushes_blend() {
+		let (core, behavior) = create();
+		let blend = tex();
+		let inputs = NodeValueRow::from([(BLEND_INPUT.to_string(), blend.clone())]);
+		let mut table = NodeValueTable::default();
+		behavior.value(&core, &inputs, Rational::new(0, 1), &mut table);
+		assert_eq!(table.get(ValueType::Texture), Some(&blend));
+	}
+
+	#[test]
+	fn value_both_pushes_job_payload_with_both_textures() {
+		let (core, _) = create();
+		let payload = run(&core, &both());
+		assert_eq!(payload.type_id, "org.olivevideoeditor.Olive.transitionfx");
+		assert_eq!(payload.shader_id, "crossdissolve", "combo default is index 0");
+		assert_eq!(payload.iterations, 1);
+		assert_eq!(payload.iterative_input, "");
+		assert_eq!(payload.effect_input, TEXTURE_INPUT);
+		assert!(payload.params.contains_key(TEXTURE_INPUT));
+		assert!(payload.params.contains_key(BLEND_INPUT));
+		assert!(
+			!payload.params.contains_key(PROGRESS_INPUT),
+			"the default progress stays out of the row so the layer can fill it"
+		);
+		assert_eq!(
+			payload.params.get(COLOR_INPUT),
+			Some(&NodeValue::Color([0.0, 0.0, 0.0, 1.0])),
+			"the style parameters ride along"
+		);
+	}
+
+	#[test]
+	fn value_takes_type_from_row_and_core() {
+		let (mut core, _) = create();
+		let mut inputs = both();
+		inputs.insert(TYPE_INPUT.to_string(), NodeValue::Combo(2));
+		assert_eq!(run(&core, &inputs).shader_id, "wipe");
+
+		core.set_standard_value(TYPE_INPUT, -1, NodeValue::Combo(3));
+		assert_eq!(run(&core, &both()).shader_id, "slide");
+	}
+
+	#[test]
+	fn value_explicit_progress_stays_in_the_job() {
+		let (core, _) = create();
+		let mut inputs = both();
+		inputs.insert(PROGRESS_INPUT.to_string(), NodeValue::Float(0.75));
+		let payload = run(&core, &inputs);
+		assert_eq!(
+			payload.params.get(PROGRESS_INPUT),
+			Some(&NodeValue::Float(0.75)),
+			"a user-set progress overrides the layer sweep"
+		);
+	}
+
+	#[test]
+	fn value_keyframed_progress_stays_in_the_job() {
+		let (mut core, _) = create();
+		core.keyframe_track_mut(PROGRESS_INPUT, -1)
+			.set_key(Keyframe {
+				time: Rational::new(0, 1),
+				value: NodeValue::Float(0.0),
+				interpolation: Interpolation::Linear,
+				bezier_in: (0.0, 0.0),
+				bezier_out: (0.0, 0.0),
+			});
+		let mut inputs = both();
+		inputs.insert(
+			PROGRESS_INPUT.to_string(),
+			core.value_at_time(PROGRESS_INPUT, -1, Rational::new(0, 1)),
+		);
+		let payload = run(&core, &inputs);
+		assert_eq!(
+			payload.params.get(PROGRESS_INPUT),
+			Some(&NodeValue::Float(0.0)),
+			"a keyframed progress is authored, even when its value is 0"
+		);
+	}
+
+	#[test]
+	fn shader_code_selects_variant_without_switch() {
+		let n = TransitionFxNode;
+		let sources: Vec<String> = SHADER_IDS
+			.iter()
+			.map(|id| n.shader_code(id).expect("shader source"))
+			.collect();
+		for (i, glsl) in sources.iter().enumerate() {
+			assert!(glsl.contains("uniform sampler2D tex_in;"), "{i} binds tex_in");
+			assert!(glsl.contains("uniform bool tex_in_enabled;"), "{i}");
+			assert!(glsl.contains("uniform float progress_in;"), "{i}");
+			assert!(!glsl.contains("switch"));
+		}
+		for i in [0usize, 2, 3] {
+			assert!(
+				sources[i].contains("uniform sampler2D blend_in;"),
+				"{i} binds the second picture"
+			);
+			assert!(sources[i].contains("uniform bool blend_in_enabled;"), "{i}");
+		}
+		assert!(sources[1].contains("uniform vec4 color_in;"), "fade fades");
+		for i in [2usize, 3] {
+			assert!(
+				sources[i].contains("uniform float direction_in;"),
+				"{i} sweeps"
+			);
+		}
+		for i in 0..sources.len() {
+			for j in (i + 1)..sources.len() {
+				assert_ne!(sources[i], sources[j], "types {i} and {j} differ");
+			}
+		}
+		assert!(sources[0].contains("mix(tex_col, blend_col, progress_in)"));
+		assert!(sources[1].contains("mix(color_in, tex_col, progress_in)"));
+	}
+
+	#[test]
+	fn shader_code_unknown_id_falls_back_to_cross_dissolve() {
+		let n = TransitionFxNode;
+		assert_eq!(
+			n.shader_code("nonsense"),
+			n.shader_code(SHADER_IDS[0]),
+			"an unknown style id renders as cross dissolve"
+		);
+	}
+
+	#[test]
+	fn duplicate_clones_behavior() {
+		let (core, behavior) = create();
+		let dup = behavior.duplicate(&core).unwrap();
+		assert_eq!(dup.name(), "Transition FX");
+	}
+}
+
+/// Register this node type.
+pub fn register(meta: &mut Vec<NodeMeta>) {
+	meta.push(NodeMeta {
+		type_id: "org.olivevideoeditor.Olive.transitionfx",
+		name: "Transition FX",
+		categories: &[Category::Effect],
+		create,
+	});
+}
diff --git a/crates/oak-node/src/nodes/transitions.rs b/crates/oak-node/src/nodes/transitions.rs
new file mode 100644
index 000000000..673f0d42c
--- /dev/null
+++ b/crates/oak-node/src/nodes/transitions.rs
@@ -0,0 +1,608 @@
+// Oak Video Editor - Non-Linear Video Editor
+// Copyright (C) 2026 Oak Team
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+//! Transition node: blends the two textures a timeline transition block
+//! sits between, driven by a `progress_in` factor.
+//!
+//! One behavior carries all four transition styles
+//! ([`TYPE_NAMES`]); the `type_in` combo picks the fragment shader
+//! through its shader id ([`SHADER_IDS`]) — the same
+//! `node->get_shader_code(shader_id)` dispatch the math nodes use. The
+//! two texture inputs are named `tex_in` ("from", the clip before the
+//! cut) and `blend_in` ("to", the clip after it), matching the timeline
+//! transition block's `out_block_in`/`in_block_in` wiring; `progress_in`
+//! is the block's position across its own span, `0.0` at the cut's
+//! outgoing side and `1.0` at its incoming side.
+//!
+//! Like [`crate::nodes::dissolve`], the node never sets a
+//! `core.effect_input`: the both-present case boxes one
+//! [`ShaderJobPayload`] whose param row carries both textures, and the
+//! renderer binds `tex_in`/`blend_in` by name.
+
+use crate::factory::NodeMeta;
+use crate::jobs::ShaderJobPayload;
+use crate::node::{Category, NodeBehavior, NodeCore};
+
+/// Outgoing ("from") texture input id. Type: texture; flags:
+/// not-keyframable.
+pub const TEXTURE_INPUT: &str = "tex_in";
+
+/// Incoming ("to") texture input id. Type: texture; flags:
+/// not-keyframable.
+pub const BLEND_INPUT: &str = "blend_in";
+
+/// Transition progress input id. Type: float; default `0.0`;
+/// properties: `min = 0.0`, `max = 1.0`, `view = percentage`. `0.0`
+/// yields `tex_in`, `1.0` yields `blend_in`.
+pub const PROGRESS_INPUT: &str = "progress_in";
+
+/// Transition style input id (the combo the block and the inspector
+/// both edit). Type: combo; flags: not-connectable, not-keyframable;
+/// properties: `combobox_strings` = [`TYPE_NAMES`].
+pub const TYPE_INPUT: &str = "type_in";
+
+/// Localized style names, in combo-index order.
+pub const TYPE_NAMES: [&str; 4] = ["Cross Dissolve", "Fade", "Wipe", "Slide"];
+
+/// Shader ids for [`TYPE_NAMES`], by combo index. These are the strings
+/// the timeline transition block stores so the composite driver can ask
+/// the factory for the matching fragment source without instantiating
+/// this node in the project graph.
+pub const SHADER_IDS: [&str; 4] = ["crossdissolve", "fade", "wipe", "slide"];
+
+/// The shader id for a combo index, clamping out-of-range values onto
+/// cross dissolve (index `0`), the C++ default transition.
+pub fn shader_id_for(index: i64) -> &'static str {
+	SHADER_IDS[index.clamp(0, SHADER_IDS.len() as i64 - 1) as usize]
+}
+
+/// Cross dissolve: a straight mix, `50/50` at the midpoint.
+const SHADER_CROSSDISSOLVE: &str = r#"uniform sampler2D tex_in;
+uniform sampler2D blend_in;
+uniform bool tex_in_enabled;
+uniform bool blend_in_enabled;
+uniform float progress_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled && !blend_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    if (!tex_in_enabled) {
+        frag_color = texture(blend_in, ove_texcoord);
+        return;
+    }
+
+    if (!blend_in_enabled) {
+        frag_color = texture(tex_in, ove_texcoord);
+        return;
+    }
+
+    vec4 tex_col = texture(tex_in, ove_texcoord);
+    vec4 blend_col = texture(blend_in, ove_texcoord);
+
+    frag_color = mix(tex_col, blend_col, progress_in);
+}
+"#;
+
+/// Fade: a dip through the fade color (black), outgoing image down over
+/// the first half, incoming image up over the second — the classic
+/// "fade to color" cut. The other end of the dip (white) is not exposed
+/// yet: W5 exposes no fade-color parameter.
+const SHADER_FADE: &str = r#"uniform sampler2D tex_in;
+uniform sampler2D blend_in;
+uniform bool tex_in_enabled;
+uniform bool blend_in_enabled;
+uniform float progress_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled && !blend_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    if (!tex_in_enabled) {
+        frag_color = texture(blend_in, ove_texcoord);
+        return;
+    }
+
+    if (!blend_in_enabled) {
+        frag_color = texture(tex_in, ove_texcoord);
+        return;
+    }
+
+    vec4 tex_col = texture(tex_in, ove_texcoord);
+    vec4 blend_col = texture(blend_in, ove_texcoord);
+    vec4 fade_col = vec4(0.0, 0.0, 0.0, 1.0);
+
+    if (progress_in < 0.5) {
+        frag_color = mix(tex_col, fade_col, progress_in * 2.0);
+    } else {
+        frag_color = mix(fade_col, blend_col, (progress_in - 0.5) * 2.0);
+    }
+}
+"#;
+
+/// Wipe: a soft-edged boundary sweeps left to right, the incoming image
+/// following behind it.
+const SHADER_WIPE: &str = r#"uniform sampler2D tex_in;
+uniform sampler2D blend_in;
+uniform bool tex_in_enabled;
+uniform bool blend_in_enabled;
+uniform float progress_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled && !blend_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    if (!tex_in_enabled) {
+        frag_color = texture(blend_in, ove_texcoord);
+        return;
+    }
+
+    if (!blend_in_enabled) {
+        frag_color = texture(tex_in, ove_texcoord);
+        return;
+    }
+
+    vec4 tex_col = texture(tex_in, ove_texcoord);
+    vec4 blend_col = texture(blend_in, ove_texcoord);
+
+    float soft = 0.02;
+    float boundary = progress_in;
+    float mask = smoothstep(boundary - soft, boundary + soft, ove_texcoord.x);
+
+    frag_color = mix(blend_col, tex_col, mask);
+}
+"#;
+
+/// Slide: the outgoing image is pushed off to the left while the
+/// incoming image slides in from the right (the C++ "push" direction).
+const SHADER_SLIDE: &str = r#"uniform sampler2D tex_in;
+uniform sampler2D blend_in;
+uniform bool tex_in_enabled;
+uniform bool blend_in_enabled;
+uniform float progress_in;
+
+in vec2 ove_texcoord;
+out vec4 frag_color;
+
+void main(void) {
+    if (!tex_in_enabled && !blend_in_enabled) {
+        frag_color = vec4(0.0);
+        return;
+    }
+
+    if (!tex_in_enabled) {
+        frag_color = texture(blend_in, ove_texcoord);
+        return;
+    }
+
+    if (!blend_in_enabled) {
+        frag_color = texture(tex_in, ove_texcoord);
+        return;
+    }
+
+    vec2 from_uv = ove_texcoord + vec2(progress_in, 0.0);
+    vec2 to_uv = ove_texcoord - vec2(1.0 - progress_in, 0.0);
+
+    vec4 tex_col = texture(tex_in, from_uv);
+    vec4 blend_col = texture(blend_in, to_uv);
+
+    if (ove_texcoord.x < 1.0 - progress_in) {
+        frag_color = tex_col;
+    } else {
+        frag_color = blend_col;
+    }
+}
+"#;
+
+/// Transition node. Unit-like: the style rides in `type_in`, so there is
+/// no per-instance state.
+pub struct TransitionNode;
+
+impl TransitionNode {
+	/// The fragment source for a shader id; an unknown id falls back to
+	/// cross dissolve (the only style the timeline creates by default).
+	fn shader_frag(shader_id: &str) -> &'static str {
+		match shader_id {
+			"fade" => SHADER_FADE,
+			"wipe" => SHADER_WIPE,
+			"slide" => SHADER_SLIDE,
+			_ => SHADER_CROSSDISSOLVE,
+		}
+	}
+}
+
+impl NodeBehavior for TransitionNode {
+	/// Human-readable name.
+	fn name(&self) -> &str {
+		"Transition"
+	}
+
+	/// Stable type id.
+	fn type_id(&self) -> &str {
+		"org.olivevideoeditor.Olive.transition"
+	}
+
+	/// Categories. Filed under timeline (the `Category` enum has no
+	/// transition group; the transition *block* sits in the same one).
+	fn categories(&self) -> &[Category] {
+		&[Category::Timeline]
+	}
+
+	/// Description.
+	fn description(&self) -> &str {
+		"Blend the two sides of a cut with a transition style."
+	}
+
+	/// Localized input names: `tex_in` -> "From", `blend_in` -> "To",
+	/// `progress_in` -> "Progress", `type_in` -> "Type" (the ids the
+	/// transition block's `out_block_in`/`in_block_in` display names
+	/// mirror).
+	fn input_name<'a>(&self, id: &'a str) -> &'a str {
+		match id {
+			TEXTURE_INPUT => "From",
+			BLEND_INPUT => "To",
+			PROGRESS_INPUT => "Progress",
+			TYPE_INPUT => "Type",
+			_ => id,
+		}
+	}
+
+	/// Evaluate outputs: if only one texture is present, push it as-is
+	/// (a transition against nothing is the input itself); if both are
+	/// present, push one shader job over the whole input row; if
+	/// neither, push nothing.
+	///
+	/// The resolved progress and the shader id the `type_in` combo
+	/// selects are written into the job row, so the `progress_in`
+	/// uniform is always present (the composite driver relies on that,
+	/// and `run_effect`'s row is otherwise only as complete as the
+	/// caller's).
+	fn value(
+		&self,
+		core: &NodeCore,
+		inputs: &crate::value::NodeValueRow,
+		time: oak_core::Rational,
+		table: &mut crate::value::NodeValueTable,
+	) {
+		use crate::value::{NodeValue, ValueType};
+
+		let tex = inputs.get(TEXTURE_INPUT);
+		let blend = inputs.get(BLEND_INPUT);
+
+		match (tex, blend) {
+			(Some(NodeValue::Texture(_)), Some(NodeValue::Texture(_))) => {
+				let progress = match inputs.get(PROGRESS_INPUT) {
+					Some(v) => v.to_double(),
+					None => core.value_at_time(PROGRESS_INPUT, -1, time).to_double(),
+				};
+				let ty = match inputs.get(TYPE_INPUT) {
+					Some(v) => v.to_double() as i64,
+					None => core.value_at_time(TYPE_INPUT, -1, time).to_double() as i64,
+				};
+
+				let mut params = inputs.clone();
+				params.insert(PROGRESS_INPUT.to_string(), NodeValue::Float(progress));
+				params.insert(TYPE_INPUT.to_string(), NodeValue::Combo(ty));
+
+				table.push(
+					ValueType::Texture,
+					NodeValue::Texture(crate::handle::make_owned(ShaderJobPayload {
+						node_id: crate::id::NodeId::INVALID,
+						time,
+						iterations: 1,
+						type_id: self.type_id().to_string(),
+						shader_id: shader_id_for(ty).to_string(),
+						effect_input: core.effect_input.clone(),
+						params,
+						iterative_input: String::new(),
+					})),
+					None,
+				);
+			}
+			(Some(t @ NodeValue::Texture(_)), _) => {
+				table.push(ValueType::Texture, t.clone(), None);
+			}
+			(_, Some(b @ NodeValue::Texture(_))) => {
+				table.push(ValueType::Texture, b.clone(), None);
+			}
+			_ => {}
+		}
+	}
+
+	/// Shader code request: the fragment source for the requested style
+	/// id (see [`SHADER_IDS`]).
+	fn shader_code(&self, request: &str) -> Option<String> {
+		Some(Self::shader_frag(request).to_string())
+	}
+
+	/// Deep copy.
+	fn duplicate(&self, _core: &NodeCore) -> Option<Box<dyn NodeBehavior>> {
+		Some(Box::new(TransitionNode))
+	}
+}
+
+/// Constructor: adds `tex_in`, `blend_in`, `progress_in` and `type_in`
+/// with the defaults and properties documented on the constants and sets
+/// the video-effect flag (no effect input: both textures bind by name).
+pub fn create() -> (NodeCore, Box<dyn NodeBehavior>) {
+	let mut core = NodeCore::new();
+
+	let mut tex = crate::input::Input::new(
+		TEXTURE_INPUT,
+		crate::value::ValueType::Texture,
+		crate::value::NodeValue::None,
+	);
+	tex.flags |= crate::input::flags::NOT_KEYFRAMABLE;
+	core.add_input(tex);
+
+	let mut blend = crate::input::Input::new(
+		BLEND_INPUT,
+		crate::value::ValueType::Texture,
+		crate::value::NodeValue::None,
+	);
+	blend.flags |= crate::input::flags::NOT_KEYFRAMABLE;
+	core.add_input(blend);
+
+	let mut progress = crate::input::Input::new(
+		PROGRESS_INPUT,
+		crate::value::ValueType::Float,
+		crate::value::NodeValue::Float(0.0),
+	);
+	progress.properties = vec![
+		("min".to_string(), crate::value::NodeValue::Float(0.0)),
+		("max".to_string(), crate::value::NodeValue::Float(1.0)),
+		(
+			"view".to_string(),
+			crate::value::NodeValue::Text("percentage".into()),
+		),
+	];
+	core.add_input(progress);
+
+	let mut ty = crate::input::Input::new(
+		TYPE_INPUT,
+		crate::value::ValueType::Combo,
+		crate::value::NodeValue::Combo(0),
+	);
+	ty.flags |= crate::input::flags::NOT_CONNECTABLE | crate::input::flags::NOT_KEYFRAMABLE;
+	ty.properties = vec![(
+		"combobox_strings".to_string(),
+		crate::value::NodeValue::Binary(TYPE_NAMES.join(",").into_bytes()),
+	)];
+	core.add_input(ty);
+
+	core.flags |= crate::node::flags::VIDEO_EFFECT;
+
+	(core, Box::new(TransitionNode))
+}
+
+#[cfg(test)]
+mod tests {
+	use super::*;
+	use crate::value::{NodeValue, NodeValueRow, NodeValueTable, ValueType};
+	use oak_core::Rational;
+
+	fn tex() -> NodeValue {
+		NodeValue::Texture(crate::handle::CHandle::null())
+	}
+
+	fn run(core: &NodeCore, inputs: &NodeValueRow) -> ShaderJobPayload {
+		let behavior = TransitionNode;
+		let mut table = NodeValueTable::default();
+		behavior.value(core, inputs, Rational::new(0, 1), &mut table);
+		let Some(NodeValue::Texture(h)) = table.get(ValueType::Texture) else {
+			panic!("shader job expected");
+		};
+		unsafe { crate::handle::get_checked::<ShaderJobPayload>(h) }
+			.expect("shader job payload boxed")
+			.clone()
+	}
+
+	fn both() -> NodeValueRow {
+		NodeValueRow::from([
+			(TEXTURE_INPUT.to_string(), tex()),
+			(BLEND_INPUT.to_string(), tex()),
+		])
+	}
+
+	#[test]
+	fn input_names() {
+		let n = TransitionNode;
+		assert_eq!(n.input_name(TEXTURE_INPUT), "From");
+		assert_eq!(n.input_name(BLEND_INPUT), "To");
+		assert_eq!(n.input_name(PROGRESS_INPUT), "Progress");
+		assert_eq!(n.input_name(TYPE_INPUT), "Type");
+		assert_eq!(n.input_name("other"), "other");
+	}
+
+	#[test]
+	fn create_wires_inputs_and_flags() {
+		let (core, behavior) = create();
+		assert_eq!(behavior.type_id(), "org.olivevideoeditor.Olive.transition");
+		assert_eq!(behavior.name(), "Transition");
+		assert_eq!(behavior.categories(), &[Category::Timeline]);
+		assert_ne!(core.flags & crate::node::flags::VIDEO_EFFECT, 0);
+		assert_eq!(core.effect_input, "");
+
+		for id in [TEXTURE_INPUT, BLEND_INPUT] {
+			let input = core.get_input(id).expect("texture input");
+			assert_ne!(input.flags & crate::input::flags::NOT_KEYFRAMABLE, 0);
+		}
+
+		let progress = core.get_input(PROGRESS_INPUT).expect("progress input");
+		assert_eq!(progress.default, NodeValue::Float(0.0));
+		assert!(progress
+			.properties
+			.iter()
+			.any(|(k, v)| k == "min" && *v == NodeValue::Float(0.0)));
+		assert!(progress
+			.properties
+			.iter()
+			.any(|(k, v)| k == "max" && *v == NodeValue::Float(1.0)));
+
+		let ty = core.get_input(TYPE_INPUT).expect("type input");
+		assert_eq!(ty.default, NodeValue::Combo(0));
+		assert_ne!(ty.flags & crate::input::flags::NOT_CONNECTABLE, 0);
+		assert_eq!(
+			ty.properties.first(),
+			Some(&(
+				"combobox_strings".to_string(),
+				NodeValue::Binary("Cross Dissolve,Fade,Wipe,Slide".as_bytes().to_vec())
+			))
+		);
+	}
+
+	#[test]
+	fn shader_ids_cover_every_type_name() {
+		assert_eq!(TYPE_NAMES.len(), SHADER_IDS.len());
+		assert_eq!(shader_id_for(0), "crossdissolve");
+		assert_eq!(shader_id_for(3), "slide");
+		assert_eq!(shader_id_for(-1), "crossdissolve");
+		assert_eq!(shader_id_for(7), "slide");
+	}
+
+	#[test]
+	fn value_no_texture_pushes_nothing() {
+		let (core, behavior) = create();
+		let mut table = NodeValueTable::default();
+		behavior.value(&core, &NodeValueRow::default(), Rational::new(0, 1), &mut table);
+		assert!(table.is_empty());
+	}
+
+	#[test]
+	fn value_tex_only_pushes_tex() {
+		let (core, behavior) = create();
+		let input = tex();
+		let inputs = NodeValueRow::from([(TEXTURE_INPUT.to_string(), input.clone())]);
+		let mut table = NodeValueTable::default();
+		behavior.value(&core, &inputs, Rational::new(0, 1), &mut table);
+		assert_eq!(table.get(ValueType::Texture), Some(&input));
+	}
+
+	#[test]
+	fn value_blend_only_pushes_blend() {
+		let (core, behavior) = create();
+		let blend = tex();
+		let inputs = NodeValueRow::from([(BLEND_INPUT.to_string(), blend.clone())]);
+		let mut table = NodeValueTable::default();
+		behavior.value(&core, &inputs, Rational::new(0, 1), &mut table);
+		assert_eq!(table.get(ValueType::Texture), Some(&blend));
+	}
+
+	#[test]
+	fn value_both_pushes_job_payload_with_both_textures() {
+		let (core, _) = create();
+		let payload = run(&core, &both());
+		assert_eq!(payload.type_id, "org.olivevideoeditor.Olive.transition");
+		assert_eq!(payload.shader_id, "crossdissolve", "combo default is index 0");
+		assert_eq!(payload.iterations, 1);
+		assert_eq!(payload.iterative_input, "");
+		assert!(payload.params.contains_key(TEXTURE_INPUT));
+		assert!(payload.params.contains_key(BLEND_INPUT));
+		assert_eq!(
+			payload.params.get(PROGRESS_INPUT),
+			Some(&NodeValue::Float(0.0)),
+			"the progress factor rides in the job params"
+		);
+	}
+
+	#[test]
+	fn value_takes_progress_and_type_from_row() {
+		let (core, _) = create();
+		let mut inputs = both();
+		inputs.insert(PROGRESS_INPUT.to_string(), NodeValue::Float(0.75));
+		inputs.insert(TYPE_INPUT.to_string(), NodeValue::Combo(2));
+		let payload = run(&core, &inputs);
+		assert_eq!(payload.params.get(PROGRESS_INPUT), Some(&NodeValue::Float(0.75)));
+		assert_eq!(payload.shader_id, "wipe");
+	}
+
+	#[test]
+	fn value_takes_type_from_core() {
+		let (mut core, _) = create();
+		core.set_standard_value(TYPE_INPUT, -1, NodeValue::Combo(3));
+		let payload = run(&core, &both());
+		assert_eq!(payload.shader_id, "slide");
+	}
+
+	#[test]
+	fn shader_code_selects_variant_without_switch() {
+		let n = TransitionNode;
+		let sources: Vec<String> = SHADER_IDS
+			.iter()
+			.map(|id| n.shader_code(id).expect("shader source"))
+			.collect();
+		for (i, glsl) in sources.iter().enumerate() {
+			assert!(glsl.contains("uniform sampler2D tex_in;"), "{i} binds tex_in");
+			assert!(
+				glsl.contains("uniform sampler2D blend_in;"),
+				"{i} binds blend_in"
+			);
+			assert!(glsl.contains("uniform bool tex_in_enabled;"));
+			assert!(glsl.contains("uniform bool blend_in_enabled;"));
+			assert!(glsl.contains("uniform float progress_in;"));
+			assert!(!glsl.contains("switch"));
+		}
+		for i in 0..sources.len() {
+			for j in (i + 1)..sources.len() {
+				assert_ne!(sources[i], sources[j], "types {i} and {j} differ");
+			}
+		}
+		assert!(sources[0].contains("mix(tex_col, blend_col, progress_in)"));
+		assert!(sources[3].contains("1.0 - progress_in"));
+	}
+
+	#[test]
+	fn shader_code_unknown_id_falls_back_to_cross_dissolve() {
+		let n = TransitionNode;
+		assert_eq!(
+			n.shader_code("nonsense"),
+			n.shader_code(SHADER_IDS[0]),
+			"an unknown style id renders as cross dissolve"
+		);
+	}
+
+	#[test]
+	fn duplicate_clones_behavior() {
+		let (core, behavior) = create();
+		let dup = behavior.duplicate(&core).unwrap();
+		assert_eq!(dup.name(), "Transition");
+	}
+}
+
+/// Register this node type.
+pub fn register(meta: &mut Vec<NodeMeta>) {
+	meta.push(NodeMeta {
+		type_id: "org.olivevideoeditor.Olive.transition",
+		name: "Transition",
+		categories: &[Category::Timeline],
+		create,
+	});
+}
diff --git a/crates/oak-node/src/serializer.rs b/crates/oak-node/src/serializer.rs
index ffd33cb74..3593ad534 100644
--- a/crates/oak-node/src/serializer.rs
+++ b/crates/oak-node/src/serializer.rs
@@ -737,6 +737,7 @@ fn create_timeline_type(
 		"org.olivevideoeditor.Olive.transitionblock" => {
 			Some(crate::block::transition_create())
 		}
+		"org.olivevideoeditor.Olive.adjustment" => Some(crate::block::adjustment_create()),
 		_ => None,
 	}
 }
@@ -985,6 +986,12 @@ fn resolve_timeline_refs(graph: &mut Graph, id_map: &std::collections::HashMap<u
 		{
 			t.core.track = t.core.track.and_then(|r| resolve(&r));
 			t.core.links = entry.core.links.clone();
+		} else if let Some(a) = behavior
+			.as_any_mut()
+			.and_then(|a| a.downcast_mut::<crate::block::AdjustmentBlockBehavior>())
+		{
+			a.core.track = a.core.track.and_then(|r| resolve(&r));
+			a.core.links = entry.core.links.clone();
 		} else if let Some(f) = behavior
 			.as_any_mut()
 			.and_then(|a| a.downcast_mut::<crate::footage::FootageBehavior>())
diff --git a/crates/oak-node/tests/factory_smoke_test.rs b/crates/oak-node/tests/factory_smoke_test.rs
index b2715faa1..ccd42cfa3 100644
--- a/crates/oak-node/tests/factory_smoke_test.rs
+++ b/crates/oak-node/tests/factory_smoke_test.rs
@@ -56,6 +56,8 @@ const EXPECTED_ORDER: &[&str] = &[
 	"org.olivevideoeditor.Olive.flip",
 	"org.olivevideoeditor.Olive.noise",
 	"org.olivevideoeditor.Olive.timeoffset",
+	"org.olivevideoeditor.Olive.transition",
+	"org.olivevideoeditor.Olive.transitionfx",
 	"org.olivevideoeditor.Olive.cornerpin",
 	"org.olivevideoeditor.Olive.displaytransform",
 	"org.olivevideoeditor.Olive.ociogradingtransformlinear",
@@ -71,6 +73,29 @@ const EXPECTED_ORDER: &[&str] = &[
 	"org.olivevideoeditor.Olive.tile",
 	"org.olivevideoeditor.Olive.swirl",
 	"org.olivevideoeditor.Olive.ripple",
+	"org.olivevideoeditor.Olive.colormatrix",
+	"org.olivevideoeditor.Olive.dilate",
+	"org.olivevideoeditor.Olive.edgedetect",
+	"org.olivevideoeditor.Olive.erode",
+	// The OpenFX-Misc cleanroom GPU ports, registered between the C++
+	// menu order and multicam (see `nodes::register_all`).
+	"org.olivevideoeditor.Olive.colorcorrect",
+	"org.olivevideoeditor.Olive.gamma",
+	"org.olivevideoeditor.Olive.saturation",
+	"org.olivevideoeditor.Olive.invert",
+	"org.olivevideoeditor.Olive.clamp",
+	"org.olivevideoeditor.Olive.grade",
+	"org.olivevideoeditor.Olive.dirblur",
+	"org.olivevideoeditor.Olive.sharpen",
+	"org.olivevideoeditor.Olive.dissolve",
+	"org.olivevideoeditor.Olive.keymix",
+	"org.olivevideoeditor.Olive.premult",
+	"org.olivevideoeditor.Olive.unpremult",
+	"org.olivevideoeditor.Olive.position",
+	"org.olivevideoeditor.Olive.mirror",
+	"org.olivevideoeditor.Olive.checkerboard",
+	"org.olivevideoeditor.Olive.colorbars",
+	"org.olivevideoeditor.Olive.ramp",
 	"org.olivevideoeditor.Olive.multicam",
 ];
 
diff --git a/crates/oak-node/tests/serializer_test.rs b/crates/oak-node/tests/serializer_test.rs
index e6b55f542..9ec89af4c 100644
--- a/crates/oak-node/tests/serializer_test.rs
+++ b/crates/oak-node/tests/serializer_test.rs
@@ -229,7 +229,10 @@ fn golden_project_with_footage_loads() {
 /// chain with keyframes, plus settings and a link.
 fn build_full_project() -> std::sync::Arc<std::sync::Mutex<oak_node::project::Project>> {
 	use oak_core::{Rational, TimeRange};
-	use oak_node::block::{clip_create, gap_create, ClipBlockBehavior, GapBlockBehavior};
+	use oak_node::block::{
+		adjustment_create, clip_create, gap_create, AdjustmentBlockBehavior, ClipBlockBehavior,
+		GapBlockBehavior,
+	};
 	use oak_node::folder::FolderBehavior;
 	use oak_node::footage::{FootageBehavior, StreamInfo};
 	use oak_node::keyframe::{Interpolation, Keyframe};
@@ -425,6 +428,18 @@ fn build_full_project() -> std::sync::Arc<std::sync::Mutex<oak_node::project::Pr
 		p.graph.add_node(core, behavior)
 	};
 
+	// A standalone adjustment layer at the end of the video track.
+	let adj_id = {
+		let (core, mut behavior) = adjustment_create();
+		let a = behavior
+			.as_any_mut()
+			.and_then(|a| a.downcast_mut::<AdjustmentBlockBehavior>())
+			.unwrap();
+		a.core.range = TimeRange::new(Rational::new(220, 25), Rational::new(260, 25));
+		a.core.track = Some(vtrack_id);
+		p.graph.add_node(core, behavior)
+	};
+
 	// Wire the hierarchy (behavior fields, the Rust model).
 	{
 		let entry = p.graph.get_mut(seq_id).unwrap();
@@ -463,7 +478,7 @@ fn build_full_project() -> std::sync::Arc<std::sync::Mutex<oak_node::project::Pr
 			.as_any_mut()
 			.and_then(|a| a.downcast_mut::<TrackBehavior>())
 			.unwrap();
-		t.blocks = vec![clip1_id, gap1_id, clip2_id];
+		t.blocks = vec![clip1_id, gap1_id, clip2_id, adj_id];
 	}
 	{
 		let entry = p.graph.get_mut(alist_id).unwrap();
@@ -675,9 +690,11 @@ fn assert_full_roundtrip_fields(orig: &oak_node::project::Project, loaded: &oak_
 	let o_clip1 = o_t.blocks[0];
 	let o_gap1 = o_t.blocks[1];
 	let o_clip2 = o_t.blocks[2];
+	let o_adj = o_t.blocks[3];
 	let l_clip1 = l_t.blocks[0];
 	let l_gap1 = l_t.blocks[1];
 	let l_clip2 = l_t.blocks[2];
+	let l_adj = l_t.blocks[3];
 
 	// Clip 1: range/media_in/speed/reversed/enabled/pitch/loop, track
 	// and footage backrefs, and the effect connection.
@@ -754,6 +771,31 @@ fn assert_full_roundtrip_fields(orig: &oak_node::project::Project, loaded: &oak_
 	assert_eq!(l_c2.core.loop_mode, o_c2.core.loop_mode, "clip2 loop");
 	assert_eq!(l_c2.footage, Some(l_footage), "clip2 footage backref");
 
+	// Adjustment layer: type, range and track backref.
+	let o_a = orig
+		.graph
+		.get(o_adj)
+		.unwrap()
+		.behavior
+		.as_any()
+		.and_then(|a| a.downcast_ref::<oak_node::block::AdjustmentBlockBehavior>())
+		.unwrap();
+	let l_a = loaded
+		.graph
+		.get(l_adj)
+		.unwrap()
+		.behavior
+		.as_any()
+		.and_then(|a| a.downcast_ref::<oak_node::block::AdjustmentBlockBehavior>())
+		.unwrap();
+	assert_eq!(
+		loaded.graph.get(l_adj).unwrap().behavior.type_id(),
+		"org.olivevideoeditor.Olive.adjustment",
+		"adjustment type id"
+	);
+	assert_eq!(l_a.core.range, o_a.core.range, "adjustment range");
+	assert_eq!(l_a.core.track, Some(l_vtrack), "adjustment track backref");
+
 	// Audio track + clip 3.
 	let o_at = orig
 		.graph
diff --git a/crates/oak-render/examples/bench_playback.rs b/crates/oak-render/examples/bench_playback.rs
index 02e6c464b..3f1d4cfee 100644
--- a/crates/oak-render/examples/bench_playback.rs
+++ b/crates/oak-render/examples/bench_playback.rs
@@ -117,6 +117,7 @@ fn main() {
 				// A single footage clip covers the whole timeline.
 				footage: Some((media_clone, 0)),
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			}),
 			audio: None,
 			produce: Arc::new(|_, _| {
diff --git a/crates/oak-render/examples/bench_process.rs b/crates/oak-render/examples/bench_process.rs
index 0e7aeb4b8..c71f36be1 100644
--- a/crates/oak-render/examples/bench_process.rs
+++ b/crates/oak-render/examples/bench_process.rs
@@ -118,6 +118,7 @@ fn main() {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			}),
 			audio: None,
 			// Never invoked on the process backend (workers render from the
diff --git a/crates/oak-render/src/autocacher.rs b/crates/oak-render/src/autocacher.rs
index 8e9c4c6dc..121f3e738 100644
--- a/crates/oak-render/src/autocacher.rs
+++ b/crates/oak-render/src/autocacher.rs
@@ -153,6 +153,7 @@ impl PreviewAutoCacher {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			},
 			Box::new(|_| {}),
 		);
@@ -191,6 +192,7 @@ impl PreviewAutoCacher {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			},
 			done,
 		);
diff --git a/crates/oak-render/src/eval.rs b/crates/oak-render/src/eval.rs
index 1b6aa5ef7..91118e226 100644
--- a/crates/oak-render/src/eval.rs
+++ b/crates/oak-render/src/eval.rs
@@ -172,7 +172,13 @@ pub struct RenderEvalHooks {
     /// size as a paused full-res frame. `None` (non-sequence renders)
     /// falls back to the render-target size.
     pub sequence_size: Option<(i32, i32)>,
-
+    /// Position of the adjustment layer currently being swept, in
+    /// `0.0..=1.0` (C++ the adjustment/transition `progress_in` uniform).
+    /// [`render_graph_frame`] sets this for the duration of one adjustment
+    /// block's evaluation, so a chain hanging off an adjustment layer can
+    /// fade its result across the layer's span. Only shaders declaring a
+    /// `progress_in` uniform consume it.
+    pub layer_progress: Option<f64>,
 }
 
 // ---------------------------------------------------------------------------
@@ -283,6 +289,7 @@ impl RenderEvalHooks {
             ticket: None,
             frame_size: None,
             sequence_size: None,
+            layer_progress: None,
         }
     }
 
@@ -813,22 +820,42 @@ impl RenderEvalHooks {
         // visibly changes size whenever the transport stops. Pre-filling
         // the row wins over `run_effect`'s frame-size auto-fill; a node
         // that inserted its own `resolution_in` keeps it.
-        let mut sequence_row;
-        let params = if self.sequence_size.is_some()
+        //
+        // `progress_in` is the same pre-fill for the adjustment-layer
+        // sweep: a shader declaring the uniform receives the layer
+        // progress the graph driver recorded for this evaluation. The
+        // row is only cloned when something actually needs inserting.
+        let mut anchored_row;
+        let needs_resolution = self.sequence_size.is_some()
             && !payload.params.contains_key("resolution_in")
             && compiled
                 .translated
                 .uniforms
                 .iter()
-                .any(|u| u.name == "resolution_in")
-        {
-            let (w, h) = self.sequence_size.unwrap();
-            sequence_row = payload.params.clone();
-            sequence_row.insert(
-                "resolution_in".to_string(),
-                NodeValue::Vec2([w as f64, h as f64]),
-            );
-            &sequence_row
+                .any(|u| u.name == "resolution_in");
+        let needs_progress = self.layer_progress.is_some()
+            && !payload.params.contains_key("progress_in")
+            && compiled
+                .translated
+                .uniforms
+                .iter()
+                .any(|u| u.name == "progress_in");
+        let params = if needs_resolution || needs_progress {
+            anchored_row = payload.params.clone();
+            if needs_resolution {
+                let (w, h) = self.sequence_size.unwrap();
+                anchored_row.insert(
+                    "resolution_in".to_string(),
+                    NodeValue::Vec2([w as f64, h as f64]),
+                );
+            }
+            if needs_progress {
+                anchored_row.insert(
+                    "progress_in".to_string(),
+                    NodeValue::Float(self.layer_progress.unwrap()),
+                );
+            }
+            &anchored_row
         } else {
             &payload.params
         };
@@ -1396,6 +1423,253 @@ fn composite_tracks(frames: Vec<Frame>, size: (i32, i32)) -> Frame {
     acc
 }
 
+/// One video track's contribution to [`render_graph_frame`], resolved
+/// before any evaluation so the project lock is only borrowed immutably
+/// (the adjustment sweep needs `&mut` on the very same graph).
+enum TrackRenderStep {
+    /// The track's enabled clips covering the frame time.
+    Clips(Vec<oak_node::id::NodeId>),
+    /// The track's enabled transition block covering the frame time: the
+    /// two blocks it joins are evaluated at `time` and blended with the
+    /// transition's shader. `progress` is the block's position across its
+    /// own span, in `0..=1` (`0.0` shows `out_block`, `1.0` shows
+    /// `in_block`); `shader` is the style id the block's `type_in` combo
+    /// selects.
+    Transition {
+        block: oak_node::id::NodeId,
+        out_block: oak_node::id::NodeId,
+        in_block: oak_node::id::NodeId,
+        progress: f64,
+        shader: &'static str,
+    },
+    /// The track's enabled adjustment block covering the frame time: run
+    /// its effect chain over every frame collected below and let the
+    /// result replace them (C++ adjustment layers affect everything
+    /// underneath). `progress` is the block's position across its own
+    /// span, in `0..=1`.
+    Adjustment {
+        block: oak_node::id::NodeId,
+        progress: f64,
+    },
+}
+
+/// Where `time` sits inside `in_..out`, in `0..=1` (the adjustment
+/// layer's `progress_in`). A degenerate span reports 0.
+fn layer_progress(in_: Rational, out: Rational, time: Rational) -> f64 {
+    let (in_, out, time) = (in_.to_f64(), out.to_f64(), time.to_f64());
+    if out > in_ {
+        ((time - in_) / (out - in_)).clamp(0.0, 1.0)
+    } else {
+        0.0
+    }
+}
+
+/// Evaluate one block at `time` and read its texture back to a CPU
+/// frame. Every non-texture outcome — the evaluator produced no texture
+/// channel, the handle is null, or the read-back failed — is `Ok(None)`
+/// so a caller can fall back instead of failing the whole frame.
+fn evaluate_block_frame(
+    graph: &oak_node::graph::Graph,
+    traverser: &mut oak_node::traverser::Traverser,
+    hooks: &mut RenderEvalHooks,
+    block: oak_node::id::NodeId,
+    time: Rational,
+) -> Result<Option<Frame>> {
+    let request = oak_node::traverser::EvalRequest::new(block, time);
+    let table = traverser.evaluate(graph, &request, hooks).map_err(|e| {
+        Error::Failed(format!(
+            "graph evaluation of block {block:?} failed: {e:?}"
+        ))
+    })?;
+    let Some(NodeValue::Texture(handle)) = table.get(oak_node::value::ValueType::Texture) else {
+        return Ok(None);
+    };
+    if handle.ctx.is_null() {
+        return Ok(None);
+    }
+    let Some(texture) = (unsafe { oak_node::handle::get_checked::<Texture>(handle) }).cloned() else {
+        return Ok(None);
+    };
+    match texture.to_frame() {
+        Ok(frame) => Ok(Some(frame)),
+        Err(err) => {
+            eprintln!("graph sequence: texture read-back failed: {err:#}");
+            Ok(None)
+        }
+    }
+}
+
+/// Blend the two sides of a transition block at `time` with the block's
+/// style shader: evaluate both sides, box a [`ShaderJobPayload`] carrying
+/// the pair and the progress factor, and run it through the render
+/// seam's shader-job path (the transition node behavior supplies the
+/// fragment source for the shader id). `None` when a side produced no
+/// texture or the job could not run (no GPU context, compile failure) —
+/// the caller then falls back to the side that natively covers `time`.
+fn blend_transition(
+    graph: &oak_node::graph::Graph,
+    traverser: &mut oak_node::traverser::Traverser,
+    hooks: &mut RenderEvalHooks,
+    out_block: oak_node::id::NodeId,
+    in_block: oak_node::id::NodeId,
+    time: Rational,
+    progress: f64,
+    shader: &str,
+) -> Result<Option<Frame>> {
+    use oak_node::nodes::transitions;
+    use oak_node::value::NodeValueRow;
+
+    let from = evaluate_block_frame(graph, traverser, hooks, out_block, time)?;
+    let to = evaluate_block_frame(graph, traverser, hooks, in_block, time)?;
+
+    // Both sides present: blend them. The side frames are boxed as CPU
+    // textures (the shader-job path uploads those into scratch textures
+    // for the pass and releases them afterwards), so the originals stay
+    // around for the fallback below.
+    let blended = match (from.as_ref(), to.as_ref()) {
+        (Some(from), Some(to)) => {
+            let payload = ShaderJobPayload {
+                node_id: oak_node::id::NodeId::INVALID,
+                time,
+                iterations: 1,
+                type_id: "org.olivevideoeditor.Olive.transition".to_string(),
+                shader_id: shader.to_string(),
+                effect_input: String::new(),
+                params: NodeValueRow::from([
+                    (
+                        transitions::TEXTURE_INPUT.to_string(),
+                        texture_value(Texture::wrap_frame(from.clone())),
+                    ),
+                    (
+                        transitions::BLEND_INPUT.to_string(),
+                        texture_value(Texture::wrap_frame(to.clone())),
+                    ),
+                    (
+                        transitions::PROGRESS_INPUT.to_string(),
+                        NodeValue::Float(progress),
+                    ),
+                ]),
+                iterative_input: String::new(),
+            };
+            match hooks.process_shader_job(&payload) {
+                Some(texture) => match texture.to_frame() {
+                    Ok(frame) => Some(frame),
+                    Err(err) => {
+                        eprintln!("graph sequence: transition read-back failed: {err:#}");
+                        None
+                    }
+                },
+                None => None,
+            }
+        }
+        _ => None,
+    };
+
+    // Fallback: without a blend (a missing side, or a shader job that
+    // could not run) show the side that natively covers `time` — the
+    // outgoing block before the cut, the incoming one at or after it.
+    Ok(blended.or_else(|| if progress < 0.5 { from } else { to }))
+}
+
+/// Walk an adjustment block's effect chain to its head: the first node
+/// whose effect input has no upstream — the node a sweep must feed the
+/// composited lower layers into. Returns that node and the input id to
+/// connect on it. `None` when there is no chain (a bare adjustment layer
+/// contributes nothing and is skipped entirely) or when the walk cannot
+/// reach a head (unconnected effect input, or a cycle).
+fn adjustment_chain_head(
+    graph: &oak_node::graph::Graph,
+    block: oak_node::id::NodeId,
+) -> Option<(oak_node::id::NodeId, String)> {
+    let mut visited = std::collections::HashSet::new();
+    let mut node = block;
+    loop {
+        if !visited.insert(node) {
+            return None;
+        }
+        let entry = graph.get(node)?;
+        let input = entry.core.effect_input.clone();
+        if input.is_empty() || entry.core.get_input(&input).is_none() {
+            return None;
+        }
+        match graph.connected_output(node, &input, -1) {
+            Some(upstream) => node = upstream,
+            // The block's own effect input is open: nothing to run.
+            None if node == block => return None,
+            None => return Some((node, input)),
+        }
+    }
+}
+
+/// Run the effect chain of the adjustment `block` over `below` (the
+/// frames of every track underneath it, topmost first): composite them,
+/// stand up a temporary texture source holding the result, wire it into
+/// the chain head, evaluate the block, and tear the temporary node back
+/// down before returning — the graph is part of the live project, so
+/// anything inspecting it concurrently must never see a half-wired
+/// sweep.
+///
+/// `Ok(None)` means the boundary changes nothing (no effect chain, or the
+/// chain produced no readable texture); `Ok(Some(frame))` is the chain's
+/// output, which replaces the lower layers.
+fn flush_adjustment_layer(
+    graph: &mut oak_node::graph::Graph,
+    traverser: &mut oak_node::traverser::Traverser,
+    hooks: &mut RenderEvalHooks,
+    block: oak_node::id::NodeId,
+    below: &[Frame],
+    size: (i32, i32),
+    time: Rational,
+    progress: f64,
+) -> Result<Option<Frame>> {
+    let Some((head, head_input)) = adjustment_chain_head(graph, block) else {
+        return Ok(None);
+    };
+    let below = composite_tracks(below.to_vec(), size);
+    let (core, behavior) = oak_node::nodes::compositesource::create();
+    let source = graph.add_node(core, behavior);
+    if let Some(entry) = graph.get_mut(source) {
+        entry.core.set_standard_value(
+            oak_node::nodes::compositesource::TEXTURE_INPUT,
+            -1,
+            texture_value(Texture::wrap_frame(below)),
+        );
+    }
+    if let Err(err) = graph.connect(source, head, &head_input, -1) {
+        let _ = graph.remove_node(source);
+        return Err(Error::Failed(format!(
+            "adjustment layer {block:?}: cannot feed {head_input} of {head:?}: {err:?}"
+        )));
+    }
+    hooks.layer_progress = Some(progress);
+    let evaluated =
+        traverser.evaluate(graph, &oak_node::traverser::EvalRequest::new(block, time), hooks);
+    hooks.layer_progress = None;
+    let _ = graph.remove_node(source);
+    let table = evaluated.map_err(|e| {
+        Error::Failed(format!(
+            "graph evaluation of adjustment layer {block:?} failed: {e:?}"
+        ))
+    })?;
+    let Some(NodeValue::Texture(handle)) = table.get(oak_node::value::ValueType::Texture) else {
+        return Ok(None);
+    };
+    if handle.ctx.is_null() {
+        return Ok(None);
+    }
+    let Some(texture) = (unsafe { oak_node::handle::get_checked::<Texture>(handle) }).cloned()
+    else {
+        return Ok(None);
+    };
+    match texture.to_frame() {
+        Ok(frame) => Ok(Some(frame)),
+        Err(err) => {
+            eprintln!("graph sequence: adjustment layer read-back failed: {err:#}");
+            Ok(None)
+        }
+    }
+}
+
 /// Render one frame of `viewer` (a sequence) at `time`: evaluate every
 /// enabled clip overlapping `time` through the node graph (one traverser
 /// pass per clip; the hooks' decoder cache is shared across clips) and
@@ -1403,6 +1677,19 @@ fn composite_tracks(frames: Vec<Frame>, size: (i32, i32)) -> Frame {
 /// list the LAST track (the highest-numbered one) is the topmost stack
 /// element (NLE stacking, matching the timeline UI).
 ///
+/// A track whose enabled adjustment block covers `time` contributes an
+/// adjustment sweep (see [`flush_adjustment_layer`]) instead of its clips:
+/// everything collected below is composited and pushed through the block's
+/// effect chain, and the result replaces the layer stack underneath, so a
+/// single block affects every lower track at once.
+///
+/// A track whose enabled transition block covers `time` contributes one
+/// blended frame (see [`blend_transition`]) instead of the clip that
+/// covers `time` on its own: both blocks the transition joins are
+/// evaluated and mixed by the style shader the block's `type_in` combo
+/// selects. An adjustment block still wins over a transition on the same
+/// track.
+///
 /// `size` is the decode target for every clip, so all frames composite
 /// without per-frame scaling. Errors: `Invalid` for a non-F32 format or a
 /// non-positive size, `NotFound` for a missing viewer or non-sequence.
@@ -1420,62 +1707,166 @@ pub fn render_graph_frame(
     if w <= 0 || h <= 0 {
         return Err(Error::Invalid);
     }
-    let graph = &project.lock().unwrap().graph;
-    let entry = graph.get(viewer).ok_or(Error::NotFound)?;
-    let sequence = entry
-        .behavior
-        .as_any()
-        .and_then(|a| a.downcast_ref::<oak_node::sequence::SequenceBehavior>())
-        .ok_or(Error::NotFound)?;
+    let mut project_guard = project.lock().unwrap();
 
-    // Collect the clips covering `time`: the sequence's track lists (video
-    // then audio — C++ `Sequence` keeps them in the `k_track_input_format`
-    // array order), the video list's tracks topmost-first (the list's last
-    // track is the top of the stack; `composite_tracks` walks the frames in
-    // reverse and draws the first frame last), then each track's blocks.
-    let mut clips: Vec<oak_node::id::NodeId> = Vec::new();
-    for tl_id in &sequence.track_lists {
-        let Some(tl) = graph.get(*tl_id) else {
-            continue;
-        };
-        let Some(tl) = tl
+    // Plan the video tracks bottommost-first, one step per track: the
+    // sequence's track lists (video then audio — C++ `Sequence` keeps them
+    // in the `k_track_input_format` array order), the video list's tracks
+    // in stacking order (the list's last track is the top of the stack;
+    // `composite_tracks` walks the frames in reverse and draws the first
+    // frame last), then each track's blocks. The plan holds plain ids so
+    // the evaluation pass below can borrow the graph mutably for an
+    // adjustment sweep's temporary source node.
+    let (steps, sequence_size) = {
+        let graph = &project_guard.graph;
+        let entry = graph.get(viewer).ok_or(Error::NotFound)?;
+        let sequence = entry
             .behavior
             .as_any()
-            .and_then(|a| a.downcast_ref::<oak_node::track::TrackListBehavior>())
-        else {
-            continue;
-        };
-        if tl.kind != oak_node::track::TrackType::Video {
-            continue;
-        }
-        for track_id in tl.tracks.iter().rev() {
-            let Some(track) = graph.get(*track_id) else {
+            .and_then(|a| a.downcast_ref::<oak_node::sequence::SequenceBehavior>())
+            .ok_or(Error::NotFound)?;
+
+        let mut steps: Vec<TrackRenderStep> = Vec::new();
+        for tl_id in &sequence.track_lists {
+            let Some(tl) = graph.get(*tl_id) else {
                 continue;
             };
-            let Some(track) = track
+            let Some(tl) = tl
                 .behavior
                 .as_any()
-                .and_then(|a| a.downcast_ref::<oak_node::track::TrackBehavior>())
+                .and_then(|a| a.downcast_ref::<oak_node::track::TrackListBehavior>())
             else {
                 continue;
             };
-            for block_id in &track.blocks {
-                let Some(block) = graph.get(*block_id) else {
+            if tl.kind != oak_node::track::TrackType::Video {
+                continue;
+            }
+            for track_id in &tl.tracks {
+                let Some(track) = graph.get(*track_id) else {
                     continue;
                 };
-                let Some(clip) = block
+                let Some(track) = track
                     .behavior
                     .as_any()
-                    .and_then(|a| a.downcast_ref::<oak_node::block::ClipBlockBehavior>())
+                    .and_then(|a| a.downcast_ref::<oak_node::track::TrackBehavior>())
                 else {
                     continue;
                 };
-                if clip.core.enabled && time >= clip.core.in_() && time < clip.core.out() {
-                    clips.push(*block_id);
+                // An enabled adjustment block covering `time` takes over
+                // the track: its sweep replaces that track's clip stack.
+                let mut step = None;
+                for block_id in &track.blocks {
+                    let Some(block) = graph.get(*block_id) else {
+                        continue;
+                    };
+                    let Some(adjustment) = block.behavior.as_any().and_then(|a| {
+                        a.downcast_ref::<oak_node::block::AdjustmentBlockBehavior>()
+                    }) else {
+                        continue;
+                    };
+                    if adjustment.core.enabled
+                        && time >= adjustment.core.in_()
+                        && time < adjustment.core.out()
+                    {
+                        step = Some(TrackRenderStep::Adjustment {
+                            block: *block_id,
+                            progress: layer_progress(
+                                adjustment.core.in_(),
+                                adjustment.core.out(),
+                                time,
+                            ),
+                        });
+                        break;
+                    }
                 }
+                // A transition block covering `time` blends the two blocks
+                // it joins. The scan runs before the clip scan so the
+                // blend replaces the plain clip read: in the first half of
+                // the span the outgoing clip alone covers `time` (the cut
+                // is its out-point), in the second the incoming one does,
+                // so the clip path would otherwise hard-cut at the cut.
+                // A block with an unconnected side falls through to the
+                // clip path, which shows whichever clip covers `time`.
+                if step.is_none() {
+                    for block_id in &track.blocks {
+                        let Some(block) = graph.get(*block_id) else {
+                            continue;
+                        };
+                        let Some(transition) = block.behavior.as_any().and_then(|a| {
+                            a.downcast_ref::<oak_node::block::TransitionBlockBehavior>()
+                        }) else {
+                            continue;
+                        };
+                        if !(transition.core.enabled
+                            && time >= transition.core.in_()
+                            && time < transition.core.out())
+                        {
+                            continue;
+                        }
+                        let (Some(out_block), Some(in_block)) = (
+                            graph.connected_output(
+                                *block_id,
+                                oak_node::block::transition_input::OUT_BLOCK,
+                                -1,
+                            ),
+                            graph.connected_output(
+                                *block_id,
+                                oak_node::block::transition_input::IN_BLOCK,
+                                -1,
+                            ),
+                        ) else {
+                            continue;
+                        };
+                        let style = block
+                            .core
+                            .value_at_time(
+                                oak_node::block::transition_input::TYPE_INPUT,
+                                -1,
+                                time,
+                            )
+                            .to_double() as i64;
+                        step = Some(TrackRenderStep::Transition {
+                            block: *block_id,
+                            out_block,
+                            in_block,
+                            progress: layer_progress(
+                                transition.core.in_(),
+                                transition.core.out(),
+                                time,
+                            ),
+                            shader: oak_node::nodes::transitions::shader_id_for(style),
+                        });
+                        break;
+                    }
+                }
+                if step.is_none() {
+                    let mut clips: Vec<oak_node::id::NodeId> = Vec::new();
+                    for block_id in &track.blocks {
+                        let Some(block) = graph.get(*block_id) else {
+                            continue;
+                        };
+                        let Some(clip) = block
+                            .behavior
+                            .as_any()
+                            .and_then(|a| a.downcast_ref::<oak_node::block::ClipBlockBehavior>())
+                        else {
+                            continue;
+                        };
+                        if clip.core.enabled && time >= clip.core.in_() && time < clip.core.out() {
+                            clips.push(*block_id);
+                        }
+                    }
+                    step = Some(TrackRenderStep::Clips(clips));
+                }
+                steps.push(step.unwrap());
             }
         }
-    }
+        let sequence_size = sequence
+            .video_params
+            .first()
+            .map(|p| (p.width.max(1), p.height.max(1)));
+        (steps, sequence_size)
+    };
 
     let mut traverser = oak_node::traverser::Traverser::new();
     let mut hooks = RenderEvalHooks::new();
@@ -1483,47 +1874,124 @@ pub fn render_graph_frame(
     // The generators' `resolution_in` anchor (C++ NodeGlobals square
     // resolution): the sequence's native size, so proxy-size playback and
     // full-res paused frames draw generated layers identically.
-    hooks.sequence_size = sequence
-        .video_params
-        .first()
-        .map(|p| (p.width.max(1), p.height.max(1)));
+    hooks.sequence_size = sequence_size;
     let perf = std::env::var_os("OAK_PERF").is_some();
     let mut perf_collect = perf.then(std::time::Instant::now);
     let mut perf_collect_ms = 0.0f64;
+    // Topmost frame first (see the track walk above).
     let mut frames: Vec<Frame> = Vec::new();
-    let mut perf_clip_hist: Vec<(oak_node::id::NodeId, f64)> = Vec::new();
-    for clip in clips.clone() {
-        let clip_sw = perf.then(std::time::Instant::now);
-        let request = oak_node::traverser::EvalRequest::new(clip, time);
-        let table = traverser.evaluate(graph, &request, &mut hooks).map_err(|e| {
-            Error::Failed(format!("graph evaluation of clip {clip:?} failed: {e:?}"))
-        })?;
-        if perf {
-            perf_clip_hist.push((
-                clip,
-                clip_sw
-                    .map(|t| t.elapsed().as_secs_f64() * 1000.0)
-                    .unwrap_or(0.0),
-            ));
-        }
-        if let Some(t0) = &perf_collect {
-            perf_collect_ms += t0.elapsed().as_secs_f64() * 1000.0;
-            perf_collect = Some(std::time::Instant::now());
-        }
-        let Some(NodeValue::Texture(handle)) = table.get(oak_node::value::ValueType::Texture)
-        else {
-            continue;
-        };
-        if handle.ctx.is_null() {
-            continue;
-        }
-        let Some(texture) = (unsafe { oak_node::handle::get_checked::<Texture>(handle) }).cloned()
-        else {
-            continue;
-        };
-        match texture.to_frame() {
-            Ok(frame) => frames.push(frame),
-            Err(err) => eprintln!("graph sequence: texture read-back failed: {err:#}"),
+    let mut perf_clip_hist: Vec<(&'static str, oak_node::id::NodeId, f64)> = Vec::new();
+    for step in steps {
+        match step {
+            TrackRenderStep::Clips(clips) => {
+                for clip in clips {
+                    let clip_sw = perf.then(std::time::Instant::now);
+                    let request = oak_node::traverser::EvalRequest::new(clip, time);
+                    let table = traverser
+                        .evaluate(&project_guard.graph, &request, &mut hooks)
+                        .map_err(|e| {
+                            Error::Failed(format!(
+                                "graph evaluation of clip {clip:?} failed: {e:?}"
+                            ))
+                        })?;
+                    if perf {
+                        perf_clip_hist.push((
+                            "clip",
+                            clip,
+                            clip_sw
+                                .map(|t| t.elapsed().as_secs_f64() * 1000.0)
+                                .unwrap_or(0.0),
+                        ));
+                    }
+                    if let Some(t0) = &perf_collect {
+                        perf_collect_ms += t0.elapsed().as_secs_f64() * 1000.0;
+                        perf_collect = Some(std::time::Instant::now());
+                    }
+                    let Some(NodeValue::Texture(handle)) =
+                        table.get(oak_node::value::ValueType::Texture)
+                    else {
+                        continue;
+                    };
+                    if handle.ctx.is_null() {
+                        continue;
+                    }
+                    let Some(texture) =
+                        (unsafe { oak_node::handle::get_checked::<Texture>(handle) }).cloned()
+                    else {
+                        continue;
+                    };
+                    match texture.to_frame() {
+                        Ok(frame) => frames.insert(0, frame),
+                        Err(err) => {
+                            eprintln!("graph sequence: texture read-back failed: {err:#}")
+                        }
+                    }
+                }
+            }
+            TrackRenderStep::Transition {
+                block,
+                out_block,
+                in_block,
+                progress,
+                shader,
+            } => {
+                let transition_sw = perf.then(std::time::Instant::now);
+                let blended = blend_transition(
+                    &project_guard.graph,
+                    &mut traverser,
+                    &mut hooks,
+                    out_block,
+                    in_block,
+                    time,
+                    progress,
+                    shader,
+                )?;
+                if perf {
+                    perf_clip_hist.push((
+                        "transition",
+                        block,
+                        transition_sw
+                            .map(|t| t.elapsed().as_secs_f64() * 1000.0)
+                            .unwrap_or(0.0),
+                    ));
+                }
+                if let Some(t0) = &perf_collect {
+                    perf_collect_ms += t0.elapsed().as_secs_f64() * 1000.0;
+                    perf_collect = Some(std::time::Instant::now());
+                }
+                if let Some(frame) = blended {
+                    frames.insert(0, frame);
+                }
+            }
+            TrackRenderStep::Adjustment { block, progress } => {
+                let sweep_sw = perf.then(std::time::Instant::now);
+                let swept = flush_adjustment_layer(
+                    &mut project_guard.graph,
+                    &mut traverser,
+                    &mut hooks,
+                    block,
+                    &frames,
+                    size,
+                    time,
+                    progress,
+                )?;
+                if perf {
+                    perf_clip_hist.push((
+                        "adjustment",
+                        block,
+                        sweep_sw
+                            .map(|t| t.elapsed().as_secs_f64() * 1000.0)
+                            .unwrap_or(0.0),
+                    ));
+                }
+                if let Some(t0) = &perf_collect {
+                    perf_collect_ms += t0.elapsed().as_secs_f64() * 1000.0;
+                    perf_collect = Some(std::time::Instant::now());
+                }
+                if let Some(frame) = swept {
+                    frames = vec![frame];
+                }
+            }
         }
     }
 
@@ -1534,8 +2002,8 @@ pub fn render_graph_frame(
         let composite_ms = composite_started
             .map(|t| t.elapsed().as_secs_f64() * 1000.0)
             .unwrap_or(0.0);
-        for (clip, ms) in &perf_clip_hist {
-            eprintln!("[perf]   clip {clip:?} evaluate {ms:.1}ms");
+        for (kind, node, ms) in &perf_clip_hist {
+            eprintln!("[perf]   {kind} {node:?} evaluate {ms:.1}ms");
         }
         eprintln!(
             "[perf] graph frame time {time:?} size {size:?}: evaluate {perf_collect_ms:.1}ms composite {composite_ms:.1}ms total {}ms",
@@ -1772,6 +2240,14 @@ fn render_montage_frame(
 /// shared-memory slot slice as `dst`, so the composited frame lands in
 /// the slot with no staging copy. `dst` is zeroed first (transparent
 /// black base).
+///
+/// Adjustment layers (`params.adjustments`, M14 W3) sit between the
+/// clips: a span's effect stack runs over the frame accumulated so far
+/// once `span.track_index` clips below it have been composited — the
+/// montage twin of the graph path's `flush_adjustment_layer`, which
+/// grades everything under the layer and leaves the tracks above
+/// compositing over that result. Spans arrive bottom-up (non-decreasing
+/// boundary); the loop consumes them in list order.
 pub fn render_montage_frame_into(
     time: Rational,
     params: &crate::ticket::VideoTicketParams,
@@ -1786,8 +2262,15 @@ pub fn render_montage_frame_into(
     }
     // Transparent-black base.
     dst[..need].fill(0);
+    let mut spans = params.adjustments.iter().peekable();
     // Decode from the bottom clip first, composite topmost-last.
-    for clip in &params.montage {
+    for (index, clip) in params.montage.iter().enumerate() {
+        // Every layer whose boundary is reached by the clips composited
+        // so far grades the accumulated frame before this clip lands.
+        while spans.peek().is_some_and(|span| span.track_index <= index) {
+            let span = spans.next().expect("peeked");
+            apply_adjustment_span(dst, dst_stride, w, h, span, time);
+        }
         if time < clip.in_time || time >= clip.out_time {
             continue;
         }
@@ -1809,9 +2292,119 @@ pub fn render_montage_frame_into(
         };
         composite_over(dst, dst_stride, w, h, src_data, src_stride, clip.gain);
     }
+    // Layers above every clip (their boundary is the montage end): they
+    // grade the final composite.
+    for span in spans {
+        apply_adjustment_span(dst, dst_stride, w, h, span, time);
+    }
     Ok(())
 }
 
+/// Apply an adjustment layer's effect stack to the frame accumulated so
+/// far — the montage twin of the graph path's `flush_adjustment_layer`.
+/// The stack runs source-first over `dst` before the clips above the
+/// layer are composited, so the layer grades exactly the picture
+/// underneath it (C++: the adjustment node's `texture_input` chain
+/// output passes through its effect chain, and that output is what the
+/// tracks above composite over; a bare adjustment layer passes the
+/// frame through unchanged).
+///
+/// Known limitation (M14 W3 — the same one the clip stacks carry): the
+/// montage evaluator only covers the built-in Opacity effect and OFX
+/// plugins; other built-ins (color management, transforms, generated
+/// textures…) log a warning once and pass the frame through. Graph mode
+/// (`render_graph_frame`, M14 W2) evaluates the full built-in set, so a
+/// worker holding a matching project snapshot stays the accurate
+/// preview and this path is the montage fallback.
+fn apply_adjustment_span(
+    dst: &mut [u8],
+    dst_stride: i32,
+    w: i32,
+    h: i32,
+    span: &crate::ticket::AdjustmentSpan,
+    time: Rational,
+) {
+    // Spans are baked for the ticket's own time; one that does not cover
+    // it (a stale or foreign list) is inert, matching the graph path's
+    // range test.
+    if time < span.in_time || time >= span.out_time {
+        return;
+    }
+    // Fast path: an all-Opacity stack (the common case) scales the
+    // accumulated frame in place, no staging copy — opacity is a pure
+    // per-channel multiply, alpha included (C++ `:/shaders/opacity.frag`).
+    let mut factors = Vec::new();
+    let mut opacity_only = true;
+    for effect in span.effects.iter().filter(|e| e.enabled) {
+        match opacity_factor(effect) {
+            Some(factor) => factors.push(factor),
+            // Unity is a pass-through; any other type needs the staged
+            // path below.
+            None if effect.type_id == OPACITY_EFFECT_TYPE_ID => {}
+            None => {
+                opacity_only = false;
+                break;
+            }
+        }
+    }
+    if opacity_only {
+        for factor in factors {
+            scale_channels_in_place(dst, dst_stride as usize, w, h, factor);
+        }
+        return;
+    }
+    // General path: the effect evaluator takes and returns an owned
+    // texture, so stage the accumulated frame into one and copy the
+    // result back.
+    let Ok(mut frame) = generate_frame(time, (w, h), PixelFormat::F32) else {
+        return;
+    };
+    let frame_stride = frame.linesize_bytes();
+    if !copy_rows(
+        &mut frame.data,
+        frame_stride,
+        dst,
+        dst_stride as usize,
+        w,
+        h,
+    ) {
+        return;
+    }
+    let tex = apply_effect_list(Texture::wrap_frame(frame), &span.effects, time);
+    if let Texture::Cpu(out) = &tex {
+        copy_rows(dst, dst_stride as usize, &out.data, out.linesize_bytes(), w, h);
+    }
+}
+
+/// Copy `h` rows of `w * 16` bytes between two F32 RGBA buffers with
+/// different strides (the montage pipeline writes packed frames, slots
+/// carry their own stride). False — nothing copied — when either buffer
+/// is too small for the requested geometry.
+fn copy_rows(
+    dst: &mut [u8],
+    dst_stride: usize,
+    src: &[u8],
+    src_stride: usize,
+    w: i32,
+    h: i32,
+) -> bool {
+    if w <= 0 || h <= 0 {
+        return false;
+    }
+    let row_bytes = (w as usize) * 16;
+    let rows = h as usize;
+    let span = |stride: usize| (rows - 1) * stride + row_bytes;
+    if src.len() < span(src_stride) || dst.len() < span(dst_stride) {
+        return false;
+    }
+    for y in 0..rows {
+        let s = y * src_stride;
+        let d = y * dst_stride;
+        dst[d..d + row_bytes].copy_from_slice(&src[s..s + row_bytes]);
+    }
+    true
+}
+
 /// `src` over `dst` (premultiplied-ish alpha compositing; F32 RGBA).
 /// `gain` scales the source RGB (audio-style volume applied to video
 /// transparency is ignored here; gain scales color). Exposed for the M15
@@ -1902,9 +2495,19 @@ fn apply_clip_effects(
     src: Texture,
     clip: &crate::ticket::MontageClip,
     time: Rational,
+) -> Texture {
+    apply_effect_list(src, &clip.effects, time)
+}
+
+/// Run an effect stack (source-first) over `src`; the clip stacks and the
+/// adjustment-layer stacks share this evaluator.
+fn apply_effect_list(
+    src: Texture,
+    effects: &[crate::ticket::MontageEffect],
+    time: Rational,
 ) -> Texture {
     let mut tex = src;
-    for effect in &clip.effects {
+    for effect in effects {
         if !effect.enabled {
             continue;
         }
@@ -1913,6 +2516,49 @@ fn apply_clip_effects(
     tex
 }
 
+/// The factor an Opacity effect would scale its input by: `None` when
+/// the effect is not the built-in Opacity or its factor is unity (C++
+/// `qFuzzyCompare(opacity, 1.0)`, a pass-through).
+fn opacity_factor(effect: &crate::ticket::MontageEffect) -> Option<f32> {
+    if effect.type_id != OPACITY_EFFECT_TYPE_ID {
+        return None;
+    }
+    let factor = effect
+        .params
+        .iter()
+        .find(|(id, _)| id == OPACITY_VALUE_INPUT)
+        .map(|(_, v)| v.to_double())
+        .unwrap_or(1.0);
+    if (factor - 1.0).abs() * 1e12 <= factor.abs().min(1.0) {
+        return None;
+    }
+    Some(factor as f32)
+}
+
+/// Scale every F32 channel (alpha included) of an F32 RGBA frame by
+/// `factor`, in place — the Opacity shader's `frag_color * opacity_in`.
+fn scale_channels_in_place(
+    data: &mut [u8],
+    stride: usize,
+    width: i32,
+    height: i32,
+    factor: f32,
+) {
+    if stride == 0 || width <= 0 || height <= 0 {
+        return;
+    }
+    let row_bytes = (width as usize) * 16;
+    for row in data.chunks_exact_mut(stride).take(height as usize) {
+        let row_end = row_bytes.min(row.len());
+        for px in row[..row_end].chunks_exact_mut(16) {
+            for c in px.chunks_exact_mut(4) {
+                let v = f32::from_le_bytes(c.try_into().unwrap());
+                c.copy_from_slice(&(v * factor).to_le_bytes());
+            }
+        }
+    }
+}
+
 /// Apply one effect to `src` (an F32 RGBA CPU frame of the montage
 /// pipeline). `time` is the sequence time the frame is rendered at (the
 /// C++ node evaluation time).
@@ -1925,30 +2571,17 @@ fn apply_montage_effect(
     // (C++ `:/shaders/opacity.frag`: `frag_color = texture(tex_in, …) *
     // opacity_in` — the shader scales the whole vec4, alpha included).
     if effect.type_id == OPACITY_EFFECT_TYPE_ID {
-        let factor = effect
-            .params
-            .iter()
-            .find(|(id, _)| id == OPACITY_VALUE_INPUT)
-            .map(|(_, v)| v.to_double())
-            .unwrap_or(1.0);
         // Unity is a pass-through (C++ `qFuzzyCompare(opacity, 1.0)`).
-        if (factor - 1.0).abs() * 1e12 <= factor.abs().min(1.0) {
+        let Some(factor) = opacity_factor(effect) else {
             return src;
-        }
+        };
         // Texture implements Drop, so scale in place through a mutable
         // borrow instead of moving the frame out.
         let mut tex = src;
         match &mut tex {
             Texture::Cpu(frame) => {
                 let stride = frame.linesize_bytes();
-                for row in frame.data.chunks_exact_mut(stride).take(frame.height.max(0) as usize) {
-                    for px in row[..(frame.width.max(0) as usize) * 16].chunks_exact_mut(16) {
-                        for c in px.chunks_exact_mut(4) {
-                            let v = f32::from_le_bytes(c.try_into().unwrap());
-                            c.copy_from_slice(&(v * factor as f32).to_le_bytes());
-                        }
-                    }
-                }
+                scale_channels_in_place(&mut frame.data, stride, frame.width, frame.height, factor);
             }
             _ => {
                 warn_unsupported_once(
@@ -2033,6 +2666,7 @@ mod tests {
             cache_timebase: None,
             footage: None,
             montage: Vec::new(),
+        	adjustments: Vec::new(),
         };
         let tex = render_produced_frame(params.time, &params).unwrap();
         assert_eq!(tex.size(), (16, 9));
diff --git a/crates/oak-render/src/ipc.rs b/crates/oak-render/src/ipc.rs
index 790940c5b..65b6ac176 100644
--- a/crates/oak-render/src/ipc.rs
+++ b/crates/oak-render/src/ipc.rs
@@ -453,6 +453,49 @@ pub fn montage_effect_from(wire: &WireMontageEffect) -> crate::ticket::MontageEf
 	}
 }
 
+/// One adjustment layer of a montage on the wire (protocol v2 additive
+/// field of [`BatchTicketSpec`]; rationals flattened to num/den pairs).
+/// `track_index` is the montage-list boundary of
+/// [`crate::ticket::AdjustmentSpan`], not a track number.
+#[derive(Serialize, Deserialize, Default, Debug, Clone, PartialEq)]
+#[serde(default)]
+pub struct WireAdjustmentSpan {
+	/// Adjustment in point numerator (sequence time).
+	pub in_num: i64,
+	/// Adjustment in point denominator.
+	pub in_den: i64,
+	/// Adjustment out point numerator (sequence time).
+	pub out_num: i64,
+	/// Adjustment out point denominator.
+	pub out_den: i64,
+	/// Montage-list boundary (clips composited below this layer).
+	pub track_index: usize,
+	/// The adjustment's effect stack (source-first).
+	pub effects: Vec<WireMontageEffect>,
+}
+
+/// Map a ticket-side adjustment span to its wire form (main process).
+pub fn wire_adjustment_from(span: &crate::ticket::AdjustmentSpan) -> WireAdjustmentSpan {
+	WireAdjustmentSpan {
+		in_num: span.in_time.numerator(),
+		in_den: span.in_time.denominator(),
+		out_num: span.out_time.numerator(),
+		out_den: span.out_time.denominator(),
+		track_index: span.track_index,
+		effects: span.effects.iter().map(wire_effect_from).collect(),
+	}
+}
+
+/// Map a wire adjustment span back to the ticket-side form (worker).
+pub fn adjustment_from_wire(wire: &WireAdjustmentSpan) -> crate::ticket::AdjustmentSpan {
+	crate::ticket::AdjustmentSpan {
+		in_time: oak_core::Rational::new(wire.in_num, wire.in_den),
+		out_time: oak_core::Rational::new(wire.out_num, wire.out_den),
+		track_index: wire.track_index,
+		effects: wire.effects.iter().map(montage_effect_from).collect(),
+	}
+}
+
 /// One frame ticket inside a [`RenderBatchMsg`] — the main process
 /// assigns the destination `slot`.
 #[derive(Serialize, Deserialize, Default, Debug, Clone, PartialEq)]
@@ -480,6 +523,9 @@ pub struct BatchTicketSpec {
 	pub footage_stream: i32,
 	/// Sequence montage (ordered topmost-last; empty = none).
 	pub montage: Vec<WireMontageClip>,
+	/// Adjustment layers over the montage (protocol v2 additive: older peers
+	/// omit the field and it defaults to no adjustment layers).
+	pub adjustments: Vec<WireAdjustmentSpan>,
 	/// Sequence viewer node identity (0 = montage mode; nonzero = render
 	/// the viewer's graph frame from the worker's loaded snapshot).
 	pub viewer_node: u64,
@@ -1976,6 +2022,7 @@ mod tests {
 					footage_file: "a.mp4".into(),
 					footage_stream: 0,
 					montage: vec![],
+					adjustments: vec![],
 					viewer_node: 0,
 					project_key: String::new(),
 				},
@@ -2010,6 +2057,22 @@ mod tests {
 							}],
 						}],
 					}],
+					adjustments: vec![WireAdjustmentSpan {
+						in_num: 0,
+						in_den: 24,
+						out_num: 48,
+						out_den: 24,
+						track_index: 1,
+						effects: vec![WireMontageEffect {
+							type_id: "org.olivevideoeditor.Olive.opacity".into(),
+							enabled: true,
+							effect_input_id: "tex_in".into(),
+							params: vec![WireEffectParam {
+								input: "opacity_in".into(),
+								value: WireNodeValue::Float(0.5),
+							}],
+						}],
+					}],
 					viewer_node: 0,
 					project_key: String::new(),
 				},
@@ -2031,6 +2094,44 @@ mod tests {
 			value["tickets"][1]["montage"][0]["effects"][0]["params"][0]["value"],
 			json!({ "t": "float", "v": 0.5 })
 		);
+		// The adjustment layer rides the ticket as an additive v2 field: its
+		// boundary is a montage-list index, not a track number.
+		assert_eq!(value["tickets"][1]["adjustments"][0]["track_index"], 1);
+		assert_eq!(value["tickets"][1]["adjustments"][0]["out_num"], 48);
+		assert_eq!(
+			value["tickets"][1]["adjustments"][0]["effects"][0]["type_id"],
+			"org.olivevideoeditor.Olive.opacity"
+		);
+		// The ticket-side conversions agree with the wire structs.
+		let parsed: WireAdjustmentSpan =
+			serde_json::from_value(value["tickets"][1]["adjustments"][0].clone()).unwrap();
+		let span = adjustment_from_wire(&parsed);
+		assert_eq!(span.in_time, oak_core::Rational::new(0, 24));
+		assert_eq!(span.out_time, oak_core::Rational::new(48, 24));
+		assert_eq!(span.track_index, 1);
+		assert_eq!(
+			span.effects[0].type_id,
+			"org.olivevideoeditor.Olive.opacity"
+		);
+		assert_eq!(span.effects[0].effect_input_id.as_deref(), Some("tex_in"));
+		assert_eq!(
+			span.effects[0].params[0].1,
+			oak_node::value::NodeValue::Float(0.5)
+		);
+		// `Rational::new` normalizes, so the wire form of the parsed span
+		// collapses 48/24 to 2/1.
+		let wire = wire_adjustment_from(&span);
+		assert_eq!(
+			wire,
+			WireAdjustmentSpan {
+				in_num: 0,
+				in_den: 1,
+				out_num: 2,
+				out_den: 1,
+				track_index: 1,
+				effects: parsed.effects.clone(),
+			}
+		);
 		// Round-trip back to the struct.
 		let round: RenderBatchMsg = serde_json::from_value(value).unwrap();
 		assert_eq!(round, batch);
@@ -2041,6 +2142,10 @@ mod tests {
 		assert_eq!(bare.slot, 2);
 		assert_eq!(bare.format, 0);
 		assert!(bare.montage.is_empty());
+		assert!(
+			bare.adjustments.is_empty(),
+			"v1 tickets carry no adjustment layers"
+		);
 	}
 
 	/// Protocol v2 compatibility: a v1-shaped montage clip (no `effects`
diff --git a/crates/oak-render/src/procpool.rs b/crates/oak-render/src/procpool.rs
index 94c56adcf..8e2e0dbea 100644
--- a/crates/oak-render/src/procpool.rs
+++ b/crates/oak-render/src/procpool.rs
@@ -2471,6 +2471,11 @@ fn build_ticket_spec(
 		footage_file,
 		footage_stream,
 		montage,
+		adjustments: params
+			.adjustments
+			.iter()
+			.map(crate::ipc::wire_adjustment_from)
+			.collect(),
 		// M16 S1 graph mode: the worker renders the viewer's graph frame
 		// when nonzero (else the montage path above) — and only when the
 		// ticket's project matches the worker's loaded snapshot (the
diff --git a/crates/oak-render/src/ticket.rs b/crates/oak-render/src/ticket.rs
index 7a449ee40..8406e0b7f 100644
--- a/crates/oak-render/src/ticket.rs
+++ b/crates/oak-render/src/ticket.rs
@@ -82,6 +82,34 @@ pub struct MontageClip {
 	pub effects: Vec<MontageEffect>,
 }
 
+/// One adjustment layer of a sequence montage (M14 W3): a time range
+/// whose effect stack applies to everything composited below it.
+///
+/// The montage is a flat list, so the layer's position is expressed as a
+/// boundary into that list rather than as a track index: the renderer
+/// applies the span's effects once the accumulated frame holds
+/// `track_index` clips — i.e. before compositing the clip at that index
+/// (a `track_index` equal to the montage length applies after the last
+/// clip, the topmost adjustment layer). Spans arrive ordered
+/// bottom-up (non-decreasing `track_index`), matching the order the
+/// builder walks the track list in.
+#[derive(Clone, Debug)]
+pub struct AdjustmentSpan {
+	/// Adjustment in point (sequence time).
+	pub in_time: Rational,
+	/// Adjustment out point (sequence time).
+	pub out_time: Rational,
+	/// Montage-list boundary: the number of clips already composited
+	/// below this layer. Effects apply to the accumulated frame just
+	/// before the clip at this index is composited; `montage.len()`
+	/// means "after every clip".
+	pub track_index: usize,
+	/// The adjustment's effect stack (source-first; empty for a bare
+	/// adjustment layer, which passes the accumulated frame through
+	/// unchanged).
+	pub effects: Vec<MontageEffect>,
+}
+
 /// Audio ticket parameters (M12 P1): the output format plus the audio
 /// montage to mix over the requested range.
 #[derive(Clone, Debug)]
@@ -129,6 +157,10 @@ pub struct VideoTicketParams {
 	/// Sequence montage (ordered topmost-last; M12 P0). When set, the
 	/// footage field is ignored.
 	pub montage: Vec<MontageClip>,
+	/// Adjustment layers over the montage (M14 W3; montage twin of the
+	/// graph path's adjustment sweep). Ordered bottom-up; empty means the
+	/// sequence has no enabled adjustment layer over the rendered time.
+	pub adjustments: Vec<AdjustmentSpan>,
 }
 
 impl VideoTicketParams {
@@ -574,6 +606,7 @@ impl TicketArena {
 					cache_timebase: None,
 					footage: None,
 					montage: Vec::new(),
+					adjustments: Vec::new(),
 				}),
 				audio: Some(ap_job),
 				produce: producer,
@@ -744,6 +777,7 @@ mod tests {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			},
 			Box::new(move |r| {
 				let _ = tx.send(r.is_ok());
@@ -795,6 +829,7 @@ mod tests {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			},
 			Box::new(move |r| {
 				let _ = tx.send(r);
@@ -881,6 +916,7 @@ mod tests {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			},
 			Box::new(|_| {}),
 		);
@@ -897,6 +933,7 @@ mod tests {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			},
 			Box::new(|_| {}),
 		);
diff --git a/crates/oak-render/src/worker.rs b/crates/oak-render/src/worker.rs
index 124493fa5..d05450597 100644
--- a/crates/oak-render/src/worker.rs
+++ b/crates/oak-render/src/worker.rs
@@ -493,6 +493,7 @@ mod tests {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			}),
 			audio: None,
 			produce,
@@ -564,6 +565,7 @@ mod tests {
 					cache_timebase: None,
 					footage: None,
 					montage: Vec::new(),
+					adjustments: Vec::new(),
 				}),
 				audio: None,
 				produce: p,
@@ -602,6 +604,7 @@ mod tests {
 			cache_timebase: None,
 			footage: None,
 			montage: Vec::new(),
+			adjustments: Vec::new(),
 		});
 		d.post(Job {
 			node_identity: 0,
diff --git a/crates/oak-render/tests/adjustment_layer.rs b/crates/oak-render/tests/adjustment_layer.rs
new file mode 100644
index 000000000..9c0473b6d
--- /dev/null
+++ b/crates/oak-render/tests/adjustment_layer.rs
@@ -0,0 +1,352 @@
+// Oak Video Editor - Non-Linear Video Editor
+// Copyright (C) 2026 Oak Team
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+//! Adjustment layers on the graph render path: a video track whose enabled
+//! adjustment block covers the request time composites every track
+//! underneath, pushes that composite through the block's effect chain and
+//! lets the chain's output replace the lower stack — one block affects all
+//! lower tracks, clip boundaries included. A bare block (no effect chain)
+//! has no head to feed and is inert.
+
+use std::sync::{Arc, Mutex};
+
+use oak_core::texture::Texture;
+use oak_core::{PixelFormat, Rational, TimeRange};
+use oak_node::block::{AdjustmentBlockBehavior, ClipBlockBehavior};
+use oak_node::footage::FootageBehavior;
+use oak_node::id::NodeId;
+use oak_node::node::NodeCore;
+use oak_node::project::Project;
+use oak_node::sequence::SequenceBehavior;
+use oak_node::track::{TrackBehavior, TrackListBehavior};
+
+mod common;
+
+/// Unique temp path per test (the process id disambiguates parallel test
+/// binaries; the tag separates tests inside one binary).
+fn clip_path(tag: &str) -> std::path::PathBuf {
+	std::env::temp_dir().join(format!("oakrender_adjust_{tag}_{}.mp4", std::process::id()))
+}
+
+/// These tests verify graph/composite MECHANICS (the adjustment sweep), not
+/// the color pipeline. Pin the working space to the legacy sRGB
+/// pass-through so the decoded pixels stay display-referred and the
+/// pixel-value assertions hold regardless of the ACEScg default. All tests
+/// in this binary set the same value, so the shared global is race-free.
+fn pin_legacy_working_space() {
+	oak_core::color::set_pipeline_color_settings(
+		oak_core::colormath::WorkingColorSpace::SrgbLegacy,
+		oak_core::colormath::OutputColorSpec::default(),
+	);
+}
+
+/// V1 (a single track carrying every clip of `clips`, in order) plus an
+/// optional V2 adjustment layer. V2 goes in last, so it is the top of the
+/// stack and sweeps V1. `adjustment` is `(in, out, opacity)`: `Some(opacity)`
+/// builds the effect chain (an Opacity node feeding the block's `tex_in`,
+/// its own `tex_in` left open as the chain head), `None` leaves the block
+/// bare.
+fn build_project(
+	clips: &[(&str, Rational, Rational)],
+	adjustment: Option<(Rational, Rational, Option<f64>)>,
+) -> (Arc<Mutex<Project>>, NodeId) {
+	pin_legacy_working_space();
+	let project = Project::new();
+	let seq;
+	{
+		let mut p = project.lock().unwrap();
+		let (score, sbehavior) = SequenceBehavior::create();
+		seq = p.graph.add_node(score, sbehavior);
+
+		let (tcore, tbehavior) = TrackListBehavior::create();
+		let tl = p.graph.add_node(tcore, tbehavior);
+
+		let (tcore, tbehavior) = TrackBehavior::create();
+		let v1 = p.graph.add_node(tcore, tbehavior);
+		for &(path, in_, out) in clips {
+			let mut footage = FootageBehavior::new(path);
+			footage.probe().expect("probe the generated clip");
+			let footage = p.graph.add_node(NodeCore::new(), Box::new(footage));
+
+			let (ccore, cbehavior) = oak_node::block::clip_create();
+			let clip = p.graph.add_node(ccore, cbehavior);
+			p.graph
+				.connect(
+					footage,
+					clip,
+					oak_node::block::clip_input::TEXTURE_INPUT,
+					-1,
+				)
+				.expect("connect footage to clip");
+
+			p.graph
+				.get_mut(clip)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<ClipBlockBehavior>()
+				.expect("clip block")
+				.core
+				.range = TimeRange::new(in_, out);
+
+			p.graph
+				.get_mut(v1)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<TrackBehavior>()
+				.expect("video track")
+				.append_block(clip);
+		}
+		p.graph
+			.get_mut(tl)
+			.unwrap()
+			.behavior
+			.as_any_mut()
+			.unwrap()
+			.downcast_mut::<TrackListBehavior>()
+			.expect("video track list")
+			.tracks
+			.push(v1);
+
+		if let Some((in_, out, opacity)) = adjustment {
+			let (tcore, tbehavior) = TrackBehavior::create();
+			let v2 = p.graph.add_node(tcore, tbehavior);
+
+			let (acore, abehavior) = oak_node::block::adjustment_create();
+			let block = p.graph.add_node(acore, abehavior);
+			p.graph
+				.get_mut(block)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<AdjustmentBlockBehavior>()
+				.expect("adjustment block")
+				.core
+				.range = TimeRange::new(in_, out);
+
+			if let Some(value) = opacity {
+				let (ecore, ebehavior) = oak_node::nodes::opacity::create();
+				let effect = p.graph.add_node(ecore, ebehavior);
+				p.graph
+					.connect(
+						effect,
+						block,
+						oak_node::block::adjustment_input::TEXTURE_INPUT,
+						-1,
+					)
+					.expect("connect opacity to the adjustment block");
+				p.graph.get_mut(effect).unwrap().core.set_standard_value(
+					oak_node::nodes::opacity::VALUE_INPUT,
+					-1,
+					oak_node::value::NodeValue::Float(value),
+				);
+			}
+
+			p.graph
+				.get_mut(v2)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<TrackBehavior>()
+				.expect("video track")
+				.append_block(block);
+			p.graph
+				.get_mut(tl)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<TrackListBehavior>()
+				.expect("video track list")
+				.tracks
+				.push(v2);
+		}
+
+		p.graph
+			.get_mut(seq)
+			.unwrap()
+			.behavior
+			.as_any_mut()
+			.unwrap()
+			.downcast_mut::<SequenceBehavior>()
+			.expect("sequence")
+			.track_lists
+			.push(tl);
+	}
+	(project, seq)
+}
+
+/// The raw CPU frame bytes of a rendered texture.
+fn frame_data(texture: &Texture) -> &[u8] {
+	let Texture::Cpu(frame) = texture else {
+		panic!("graph render produced a non-CPU texture");
+	};
+	&frame.data
+}
+
+/// Render one 64x64 F32 frame of `seq` at `time` and return its bytes.
+fn render_frame(project: &Arc<Mutex<Project>>, seq: NodeId, time: Rational) -> Vec<u8> {
+	let texture =
+		oak_render::eval::render_graph_frame(project, seq, time, (64, 64), PixelFormat::F32)
+			.expect("graph render");
+	frame_data(&texture).to_vec()
+}
+
+/// The F32 RGBA channel of a 64x64 frame at `(x, y)`.
+fn channel(data: &[u8], x: usize, y: usize, c: usize) -> f32 {
+	let off = (y * 64 + x) * 16 + c * 4;
+	f32::from_le_bytes(data[off..off + 4].try_into().unwrap())
+}
+
+/// An adjustment layer over a single V1 track whose two solid clips abut at
+/// t=1/2: inside the layer's span BOTH clips come out of the effect chain
+/// (the red clip on the left half, the blue one on the right half of the
+/// span), so one block reaches across the clip boundary; outside the span
+/// the render is byte-identical to the same project without the layer.
+///
+/// The chain is an Opacity node at 0.5, which scales the straight-alpha
+/// vec4 by 0.5; the frame then goes through the alpha-over composite once
+/// more (like `shader_job_opacity_halves_pixels`), so each color channel
+/// ends up at a 0.25 ratio of the plain render and the alpha at 0.5.
+/// Skipped (with a note) when no GPU adapter exists.
+#[test]
+fn adjustment_layer_affects_lower_tracks_across_clips() {
+	if oak_core::backend::GpuContext::shared().is_none() {
+		eprintln!("skipping adjustment_layer_affects_lower_tracks_across_clips: no GPU adapter");
+		return;
+	}
+	let red = clip_path("across_red");
+	let blue = clip_path("across_blue");
+	oak_codec::testmedia::write_test_clip_solid(&red, 64, 64, 10, 10, [0.9, 0.1, 0.1, 1.0])
+		.expect("red clip generation");
+	oak_codec::testmedia::write_test_clip_solid(&blue, 64, 64, 10, 10, [0.1, 0.1, 0.9, 1.0])
+		.expect("blue clip generation");
+
+	let red_path = red.to_string_lossy().to_string();
+	let blue_path = blue.to_string_lossy().to_string();
+	let clips: Vec<(&str, Rational, Rational)> = vec![
+		(&red_path, Rational::new(0, 1), Rational::new(1, 2)),
+		(&blue_path, Rational::new(1, 2), Rational::new(1, 1)),
+	];
+	let plain = build_project(&clips, None);
+	let adjusted = build_project(
+		&clips,
+		Some((Rational::new(1, 4), Rational::new(3, 4), Some(0.5))),
+	);
+	let nodes_before = adjusted.0.lock().unwrap().graph.node_count();
+
+	// Inside the span, a point covered by each clip in turn.
+	for (time, label) in [
+		(Rational::new(1, 4), "the red clip"),
+		(Rational::new(5, 8), "the blue clip"),
+	] {
+		let base = render_frame(&plain.0, plain.1, time);
+		let swept = render_frame(&adjusted.0, adjusted.1, time);
+
+		// Sample away from the x=32 half boundary (MPEG-2 chroma bleed and
+		// luma ringing stay within a few pixels of it).
+		let mut ratios: Vec<f32> = Vec::new();
+		let mut alphas: Vec<f32> = Vec::new();
+		for y in 4..60 {
+			for x in (4..24).chain(40..60) {
+				for c in 0..3 {
+					let a = channel(&base, x, y, c);
+					if a > 0.02 {
+						ratios.push(channel(&swept, x, y, c) / a);
+					}
+				}
+				alphas.push(channel(&swept, x, y, 3));
+			}
+		}
+		assert!(
+			ratios.len() >= 512,
+			"{label}: too few comparable samples: {}",
+			ratios.len()
+		);
+		let mean = ratios.iter().sum::<f32>() / ratios.len() as f32;
+		assert!(
+			(mean - 0.25).abs() < 0.02,
+			"{label}: a 0.5-opacity layer must leave 0.25 of each channel (shader x0.5, alpha-over x0.5), got {mean}"
+		);
+		let mean_alpha = alphas.iter().sum::<f32>() / alphas.len() as f32;
+		assert!(
+			(mean_alpha - 0.5).abs() < 0.1,
+			"{label}: the swept frame's alpha must be 0.5, got {mean_alpha}"
+		);
+	}
+
+	// Outside the span, nothing changes — not even a rounding step.
+	for time in [Rational::new(1, 8), Rational::new(7, 8)] {
+		assert_eq!(
+			render_frame(&plain.0, plain.1, time),
+			render_frame(&adjusted.0, adjusted.1, time),
+			"t={time:?} is outside the layer's span and must render unchanged"
+		);
+	}
+
+	assert_eq!(
+		adjusted.0.lock().unwrap().graph.node_count(),
+		nodes_before,
+		"the sweep must tear its temporary source node back down"
+	);
+
+	let _ = std::fs::remove_file(&red);
+	let _ = std::fs::remove_file(&blue);
+}
+
+/// A bare adjustment layer (a block with no effect chain) is inert: there
+/// is no chain head to feed, so the tracks underneath render byte for byte
+/// as if the block were absent, and the block leaves no node behind.
+#[test]
+fn bare_adjustment_layer_passes_lower_tracks_through() {
+	let red = clip_path("bare_red");
+	oak_codec::testmedia::write_test_clip_solid(&red, 64, 64, 10, 10, [0.9, 0.1, 0.1, 1.0])
+		.expect("red clip generation");
+	let red_path = red.to_string_lossy().to_string();
+	let clips: Vec<(&str, Rational, Rational)> =
+		vec![(&red_path, Rational::new(0, 1), Rational::new(1, 1))];
+
+	let plain = build_project(&clips, None);
+	let bare = build_project(
+		&clips,
+		Some((Rational::new(1, 4), Rational::new(3, 4), None)),
+	);
+	let nodes_before = bare.0.lock().unwrap().graph.node_count();
+
+	let time = Rational::new(1, 2);
+	let base = render_frame(&plain.0, plain.1, time);
+	assert!(
+		base.iter().any(|&b| b != 0),
+		"the reference frame is not black"
+	);
+	assert_eq!(
+		render_frame(&bare.0, bare.1, time),
+		base,
+		"a bare adjustment layer must not change the render"
+	);
+	assert_eq!(
+		bare.0.lock().unwrap().graph.node_count(),
+		nodes_before,
+		"a bare adjustment layer must not grow the graph"
+	);
+
+	let _ = std::fs::remove_file(&red);
+}
diff --git a/crates/oak-render/tests/ticket_worker_test.rs b/crates/oak-render/tests/ticket_worker_test.rs
index f3ab4f246..459680b20 100644
--- a/crates/oak-render/tests/ticket_worker_test.rs
+++ b/crates/oak-render/tests/ticket_worker_test.rs
@@ -67,6 +67,7 @@ fn params(time: Rational) -> VideoTicketParams {
 		cache_timebase: None,
 		footage: None,
 		montage: Vec::new(),
+		adjustments: Vec::new(),
 	}
 }
 
diff --git a/crates/oak-render/tests/transition_render.rs b/crates/oak-render/tests/transition_render.rs
new file mode 100644
index 000000000..53514646f
--- /dev/null
+++ b/crates/oak-render/tests/transition_render.rs
@@ -0,0 +1,397 @@
+// Oak Video Editor - Non-Linear Video Editor
+// Copyright (C) 2026 Oak Team
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+//! Timeline transitions on the graph render path: an enabled transition
+//! block covering the request time replaces the plain clip read with a
+//! blend of the two clips it joins, driven by the block's style combo and
+//! its position across the block's own span. Outside the span the render
+//! is byte-identical to the same project without the transition.
+
+use std::sync::{Arc, Mutex};
+
+use oak_core::texture::Texture;
+use oak_core::{PixelFormat, Rational, TimeRange};
+use oak_node::block::{ClipBlockBehavior, TransitionBlockBehavior};
+use oak_node::footage::FootageBehavior;
+use oak_node::id::NodeId;
+use oak_node::node::NodeCore;
+use oak_node::project::Project;
+use oak_node::sequence::SequenceBehavior;
+use oak_node::track::{TrackBehavior, TrackListBehavior};
+
+mod common;
+
+/// Unique temp path per test (the process id disambiguates parallel test
+/// binaries; the tag separates tests inside one binary).
+fn clip_path(tag: &str) -> std::path::PathBuf {
+	std::env::temp_dir().join(format!("oakrender_trans_{tag}_{}.mp4", std::process::id()))
+}
+
+/// The transition mechanics are color-space independent, so pin the
+/// working space to the legacy sRGB pass-through like the adjustment
+/// layer tests do and compare decoded pixel values directly.
+fn pin_legacy_working_space() {
+	oak_core::color::set_pipeline_color_settings(
+		oak_core::colormath::WorkingColorSpace::SrgbLegacy,
+		oak_core::colormath::OutputColorSpec::default(),
+	);
+}
+
+/// V1 (a single track carrying `clips`, in order) plus an optional
+/// transition block. `transition` is `(seam, in_offset, out_offset)`: the
+/// block is created with `[seam - in_offset, seam + out_offset]` and
+/// inserted on the track between the first two clips, its two inputs
+/// wired to them (`out_block_in` from the first clip, `in_block_in` from
+/// the second). Returns the transition block's id so a test can flip its
+/// style combo.
+fn build_project(
+	clips: &[(&str, Rational, Rational)],
+	transition: Option<(Rational, Rational, Rational)>,
+) -> (Arc<Mutex<Project>>, NodeId, Option<NodeId>) {
+	pin_legacy_working_space();
+	let project = Project::new();
+	let mut transition_id = None;
+	let seq;
+	{
+		let mut p = project.lock().unwrap();
+		let (score, sbehavior) = SequenceBehavior::create();
+		seq = p.graph.add_node(score, sbehavior);
+
+		let (tcore, tbehavior) = TrackListBehavior::create();
+		let tl = p.graph.add_node(tcore, tbehavior);
+
+		let (tcore, tbehavior) = TrackBehavior::create();
+		let v1 = p.graph.add_node(tcore, tbehavior);
+
+		let mut clip_ids: Vec<NodeId> = Vec::new();
+		for &(path, in_, out) in clips {
+			let mut footage = FootageBehavior::new(path);
+			footage.probe().expect("probe the generated clip");
+			let footage = p.graph.add_node(NodeCore::new(), Box::new(footage));
+
+			let (ccore, cbehavior) = oak_node::block::clip_create();
+			let clip = p.graph.add_node(ccore, cbehavior);
+			p.graph
+				.connect(footage, clip, oak_node::block::clip_input::TEXTURE_INPUT, -1)
+				.expect("connect footage to clip");
+
+			p.graph
+				.get_mut(clip)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<ClipBlockBehavior>()
+				.expect("clip block")
+				.core
+				.range = TimeRange::new(in_, out);
+
+			clip_ids.push(clip);
+		}
+
+		if let Some((seam, in_offset, out_offset)) = transition {
+			let (tcore, tbehavior) = oak_node::block::transition_create();
+			let block = p.graph.add_node(tcore, tbehavior);
+			p.graph
+				.get_mut(block)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<TransitionBlockBehavior>()
+				.expect("transition block")
+				.core
+				.range = TimeRange::new(seam - in_offset, seam + out_offset);
+			{
+				let behavior = p
+					.graph
+					.get_mut(block)
+					.unwrap()
+					.behavior
+					.as_any_mut()
+					.unwrap()
+					.downcast_mut::<TransitionBlockBehavior>()
+					.expect("transition block");
+				behavior.in_offset = in_offset;
+				behavior.out_offset = out_offset;
+			}
+			p.graph
+				.connect(
+					clip_ids[0],
+					block,
+					oak_node::block::transition_input::OUT_BLOCK,
+					-1,
+				)
+				.expect("connect the outgoing clip to the transition");
+			p.graph
+				.connect(
+					clip_ids[1],
+					block,
+					oak_node::block::transition_input::IN_BLOCK,
+					-1,
+				)
+				.expect("connect the incoming clip to the transition");
+			transition_id = Some(block);
+
+			// Track order: the outgoing clip, the transition, the incoming
+			// clip — the order `transition_commands` builds.
+			let track = p
+				.graph
+				.get_mut(v1)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<TrackBehavior>()
+				.expect("video track");
+			track.append_block(clip_ids[0]);
+			track.append_block(block);
+			track.append_block(clip_ids[1]);
+		} else {
+			let track = p
+				.graph
+				.get_mut(v1)
+				.unwrap()
+				.behavior
+				.as_any_mut()
+				.unwrap()
+				.downcast_mut::<TrackBehavior>()
+				.expect("video track");
+			for &clip in &clip_ids {
+				track.append_block(clip);
+			}
+		}
+
+		p.graph
+			.get_mut(tl)
+			.unwrap()
+			.behavior
+			.as_any_mut()
+			.unwrap()
+			.downcast_mut::<TrackListBehavior>()
+			.expect("video track list")
+			.tracks
+			.push(v1);
+
+		p.graph
+			.get_mut(seq)
+			.unwrap()
+			.behavior
+			.as_any_mut()
+			.unwrap()
+			.downcast_mut::<SequenceBehavior>()
+			.expect("sequence")
+			.track_lists
+			.push(tl);
+	}
+	(project, seq, transition_id)
+}
+
+/// Flip the transition block's style combo (index into
+/// `oak_node::nodes::transitions::TYPE_NAMES`).
+fn set_style(project: &Arc<Mutex<Project>>, block: NodeId, style: i64) {
+	project
+		.lock()
+		.unwrap()
+		.graph
+		.get_mut(block)
+		.unwrap()
+		.core
+		.set_standard_value(
+			oak_node::block::transition_input::TYPE_INPUT,
+			-1,
+			oak_node::value::NodeValue::Combo(style),
+		);
+}
+
+/// The raw CPU frame bytes of a rendered texture.
+fn frame_data(texture: &Texture) -> &[u8] {
+	let Texture::Cpu(frame) = texture else {
+		panic!("graph render produced a non-CPU texture");
+	};
+	&frame.data
+}
+
+/// Render one 64x64 F32 frame of `seq` at `time` and return its bytes.
+fn render_frame(project: &Arc<Mutex<Project>>, seq: NodeId, time: Rational) -> Vec<u8> {
+	let texture =
+		oak_render::eval::render_graph_frame(project, seq, time, (64, 64), PixelFormat::F32)
+			.expect("graph render");
+	frame_data(&texture).to_vec()
+}
+
+/// The F32 RGBA channel of a 64x64 frame at `(x, y)`.
+fn channel(data: &[u8], x: usize, y: usize, c: usize) -> f32 {
+	let off = (y * 64 + x) * 16 + c * 4;
+	f32::from_le_bytes(data[off..off + 4].try_into().unwrap())
+}
+
+/// The mean of channel `c` over `xs` x `ys` (8..56 both ways: away from
+/// the encoder's frame borders).
+fn channel_mean(data: &[u8], c: usize, xs: &[usize], ys: &[usize]) -> f32 {
+	let mut sum = 0.0;
+	let mut n = 0;
+	for &y in ys {
+		for &x in xs {
+			sum += channel(data, x, y, c);
+			n += 1;
+		}
+	}
+	sum / n as f32
+}
+
+/// Two solid clips abut at t=1/2; a transition spanning [1/4, 3/4] mixes
+/// them across the cut. At the seam (progress 0.5) a cross dissolve must
+/// show the average of the two sides, while outside the span the render
+/// is byte-identical to the same project without the transition.
+///
+/// Skipped (with a note) when no GPU adapter exists.
+#[test]
+fn cross_dissolve_blends_the_two_sides_of_a_cut() {
+	if oak_core::backend::GpuContext::shared().is_none() {
+		eprintln!("skipping cross_dissolve_blends_the_two_sides_of_a_cut: no GPU adapter");
+		return;
+	}
+	let red = clip_path("dissolve_red");
+	let blue = clip_path("dissolve_blue");
+	oak_codec::testmedia::write_test_clip_solid(&red, 64, 64, 10, 10, [0.9, 0.1, 0.1, 1.0])
+		.expect("red clip generation");
+	oak_codec::testmedia::write_test_clip_solid(&blue, 64, 64, 10, 10, [0.1, 0.1, 0.9, 1.0])
+		.expect("blue clip generation");
+	let red_path = red.to_string_lossy().to_string();
+	let blue_path = blue.to_string_lossy().to_string();
+	let clips: Vec<(&str, Rational, Rational)> = vec![
+		(&red_path, Rational::new(0, 1), Rational::new(1, 2)),
+		(&blue_path, Rational::new(1, 2), Rational::new(1, 1)),
+	];
+	let plain = build_project(&clips, None);
+	let dissolved = build_project(
+		&clips,
+		Some((Rational::new(1, 2), Rational::new(1, 4), Rational::new(1, 4))),
+	);
+	let transition = dissolved.2.expect("the transition block");
+
+	// The two sides, rendered by the same project without the transition.
+	let plain_red = render_frame(&plain.0, plain.1, Rational::new(1, 4));
+	let plain_blue = render_frame(&plain.0, plain.1, Rational::new(5, 8));
+	assert!(
+		channel_mean(&plain_red, 0, &(8..56).collect::<Vec<_>>(), &(8..56).collect::<Vec<_>>()) > 0.5,
+		"the outgoing reference must be the red clip"
+	);
+	assert!(
+		channel_mean(&plain_blue, 0, &(8..56).collect::<Vec<_>>(), &(8..56).collect::<Vec<_>>()) < 0.5,
+		"the incoming reference must be the blue clip"
+	);
+
+	// At the exact seam the block is halfway through its span, so a cross
+	// dissolve shows the average of the two sides. Sample away from the
+	// frame borders (MPEG-2 chroma bleed lives at the edges).
+	let seam = render_frame(&dissolved.0, dissolved.1, Rational::new(1, 2));
+	let xs: Vec<usize> = (8..56).collect();
+	let ys: Vec<usize> = (8..56).collect();
+	for c in 0..3 {
+		let expected = 0.5 * (channel_mean(&plain_red, c, &xs, &ys)
+			+ channel_mean(&plain_blue, c, &xs, &ys));
+		let got = channel_mean(&seam, c, &xs, &ys);
+		assert!(
+			(got - expected).abs() < 0.02,
+			"channel {c} at the seam must be the average of the two sides: expected {expected}, got {got}"
+		);
+	}
+
+	// Outside the block's span the render is byte-identical to the plain
+	// project: the transition is inert there.
+	for time in [Rational::new(1, 8), Rational::new(7, 8)] {
+		assert_eq!(
+			render_frame(&dissolved.0, dissolved.1, time),
+			render_frame(&plain.0, plain.1, time),
+			"t={time:?} is outside the transition's span and must render unchanged"
+		);
+	}
+
+	// The blend is the block's own output: the plain clip read is replaced,
+	// not added under it. Drop the block's style back to a fade and the
+	// seam goes through black (the fade's midpoint), proving the style
+	// rides on this block.
+	set_style(&dissolved.0, transition, 1);
+	let fade = render_frame(&dissolved.0, dissolved.1, Rational::new(1, 2));
+	assert!(
+		channel_mean(&fade, 0, &xs, &ys) < 0.1 && channel_mean(&fade, 2, &xs, &ys) < 0.1,
+		"a fade at progress 0.5 is fully in the fade color (black), got r={} b={}",
+		channel_mean(&fade, 0, &xs, &ys),
+		channel_mean(&fade, 2, &xs, &ys)
+	);
+
+	let _ = std::fs::remove_file(&red);
+	let _ = std::fs::remove_file(&blue);
+}
+
+/// The block's style combo selects the shader: a wipe at progress 0.5
+/// puts the incoming clip on the left of the sweeping boundary and the
+/// outgoing one on the right (the outgoing image leads the sweep).
+#[test]
+fn wipe_style_splits_the_frame_at_the_boundary() {
+	if oak_core::backend::GpuContext::shared().is_none() {
+		eprintln!("skipping wipe_style_splits_the_frame_at_the_boundary: no GPU adapter");
+		return;
+	}
+	let red = clip_path("wipe_red");
+	let blue = clip_path("wipe_blue");
+	oak_codec::testmedia::write_test_clip_solid(&red, 64, 64, 10, 10, [0.9, 0.1, 0.1, 1.0])
+		.expect("red clip generation");
+	oak_codec::testmedia::write_test_clip_solid(&blue, 64, 64, 10, 10, [0.1, 0.1, 0.9, 1.0])
+		.expect("blue clip generation");
+	let red_path = red.to_string_lossy().to_string();
+	let blue_path = blue.to_string_lossy().to_string();
+	let clips: Vec<(&str, Rational, Rational)> = vec![
+		(&red_path, Rational::new(0, 1), Rational::new(1, 2)),
+		(&blue_path, Rational::new(1, 2), Rational::new(1, 1)),
+	];
+	let project = build_project(
+		&clips,
+		Some((Rational::new(1, 2), Rational::new(1, 4), Rational::new(1, 4))),
+	);
+	set_style(&project.0, project.2.expect("the transition block"), 2);
+	let wipe = render_frame(&project.0, project.1, Rational::new(1, 2));
+
+	// Left of the boundary (x=32 at progress 0.5) trails the sweep and
+	// shows the incoming (blue) clip; right of it the outgoing (red) one.
+	// Keep 4 px clear of the boundary's soft edge (`soft = 0.02` of the
+	// width). Green is 0.1 in both clips, so only the red and blue
+	// channels tell the two sides apart.
+	let left = (8..28).collect::<Vec<_>>();
+	let right = (36..56).collect::<Vec<_>>();
+	let ys: Vec<usize> = (8..56).collect();
+	let (l_red, r_red) = (
+		channel_mean(&wipe, 0, &left, &ys),
+		channel_mean(&wipe, 0, &right, &ys),
+	);
+	let (l_blue, r_blue) = (
+		channel_mean(&wipe, 2, &left, &ys),
+		channel_mean(&wipe, 2, &right, &ys),
+	);
+	assert!(
+		l_red < 0.5 && l_blue > 0.5,
+		"left of the boundary must be the incoming clip (blue): r={l_red} b={l_blue}"
+	);
+	assert!(
+		r_red > 0.5 && r_blue < 0.5,
+		"right of the boundary must be the outgoing clip (red): r={r_red} b={r_blue}"
+	);
+
+	let _ = std::fs::remove_file(&red);
+	let _ = std::fs::remove_file(&blue);
+}
diff --git a/crates/oak-render/tests/transitionfx.rs b/crates/oak-render/tests/transitionfx.rs
new file mode 100644
index 000000000..608ef2c20
--- /dev/null
+++ b/crates/oak-render/tests/transitionfx.rs
@@ -0,0 +1,241 @@
+// Oak Video Editor - Non-Linear Video Editor
+// Copyright (C) 2026 Oak Team
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+//! GPU pixel tests for the layer-form transition node
+//! (`org.olivevideoeditor.Olive.transitionfx`): cross dissolve, fade, wipe
+//! and slide over `tex_in` + `blend_in`, with the progress either set on
+//! the row or auto-filled by the renderer from the adjustment layer's own
+//! span. Skips without a GPU adapter.
+
+use oak_core::texture::Texture;
+use oak_core::{PixelFormat, Rational};
+use oak_node::value::{NodeValue, NodeValueRow, NodeValueTable, ValueType};
+
+const TRANSITIONFX: &str = "org.olivevideoeditor.Olive.transitionfx";
+
+fn texture_value(t: Texture) -> NodeValue { NodeValue::Texture(oak_node::handle::make_owned(t)) }
+fn gpu() -> bool { oak_core::backend::GpuContext::shared().is_some() }
+fn filled_frame(size: (i32, i32), rgba: [f32; 4]) -> Texture {
+	let mut f = oak_render::eval::generate_frame(Rational::new(0, 1), size, PixelFormat::F32).unwrap();
+	for px in f.data.chunks_exact_mut(16) { for (c, v) in px.chunks_exact_mut(4).zip(rgba) { c.copy_from_slice(&v.to_le_bytes()); } }
+	Texture::wrap_frame(f)
+}
+/// An 8x8 frame whose red channel carries each column's centre `u`
+/// (`(x + 0.5) / width`), so a sample taken at `u` reads back as exactly
+/// `u`: the slide tests use it to tell which part of the picture landed
+/// where.
+fn gradient_frame_x(size: (i32, i32)) -> Texture {
+	let mut f = oak_render::eval::generate_frame(Rational::new(0, 1), size, PixelFormat::F32).unwrap();
+	let w = size.0 as usize;
+	for (i, px) in f.data.chunks_exact_mut(16).enumerate() {
+		let u = ((i % w) as f32 + 0.5) / w as f32;
+		for (c, v) in px.chunks_exact_mut(4).zip([u, 0.0, 0.0, 1.0]) { c.copy_from_slice(&v.to_le_bytes()); }
+	}
+	Texture::wrap_frame(f)
+}
+fn pixel_at(frame: &oak_core::texture::Frame, x: usize, y: usize) -> [f32; 4] {
+	let stride = frame.linesize_bytes() as usize;
+	let at = y * stride + x * 16;
+	let mut out = [0f32; 4];
+	for c in 0..4 { out[c] = f32::from_le_bytes(frame.data[at + c*4..at + c*4 + 4].try_into().unwrap()); }
+	out
+}
+/// Evaluate one node row. `layer_progress` is the adjustment-layer sweep
+/// position the graph driver records before walking an effect chain; the
+/// shader job picks it up for `progress_in` when the node left the input
+/// out of its params (the auto-fill path).
+fn eval_node_row(
+	type_id: &str,
+	inputs: NodeValueRow,
+	frame_size: Option<(i32, i32)>,
+	layer_progress: Option<f64>,
+) -> oak_core::texture::Frame {
+	use oak_node::traverser::RenderHooks;
+	let (core, behavior) = oak_node::factory::Factory::global().create_any(type_id).expect("node type registered");
+	let mut table = NodeValueTable::default();
+	behavior.value(&core, &inputs, Rational::new(0, 1), &mut table);
+	let mut hooks = oak_render::eval::RenderEvalHooks::new();
+	hooks.frame_size = frame_size;
+	hooks.layer_progress = layer_progress;
+	hooks.resolve(oak_node::id::NodeId::INVALID, &inputs, &mut table);
+	let Some(NodeValue::Texture(handle)) = table.get(ValueType::Texture) else { panic!("{type_id}: no texture produced") };
+	if handle.ctx.is_null() { panic!("{type_id}: null texture produced"); }
+	let tex = unsafe { oak_node::handle::get_checked::<Texture>(handle) }.expect("resolved texture");
+	assert!(matches!(tex, Texture::Gpu { .. }), "{type_id}: must render on the GPU");
+	tex.to_frame().expect("readback")
+}
+
+/// Assert the four channels of a pixel against the expected values.
+fn assert_pixel(px: [f32; 4], want: [f32; 4], what: &str) {
+	for c in 0..4 {
+		assert!(
+			(px[c] - want[c]).abs() < 1e-3,
+			"{what}: channel {c}: got {px:?}, want {want:?}"
+		);
+	}
+}
+
+/// The two-picture row the shader job carries: `tex_in` (From) and
+/// `blend_in` (To), each painted over the whole 8x8 frame.
+fn transition_row(tex: [f32; 4], blend: [f32; 4]) -> NodeValueRow {
+	let mut row = NodeValueRow::new();
+	row.insert("tex_in".to_string(), texture_value(filled_frame((8, 8), tex)));
+	row.insert("blend_in".to_string(), texture_value(filled_frame((8, 8), blend)));
+	row
+}
+
+/// Style combo index for a row (cross dissolve is 0, the default).
+fn style(row: &mut NodeValueRow, index: i64) {
+	row.insert("type_in".to_string(), NodeValue::Combo(index));
+}
+
+/// Halfway cross dissolve lerps red and green to olive.
+#[test]
+fn crossdissolve_half_progress_blends() {
+	if !gpu() { eprintln!("no adapter; skipping"); return; }
+	let mut row = transition_row([1.0, 0.0, 0.0, 1.0], [0.0, 1.0, 0.0, 1.0]);
+	row.insert("progress_in".to_string(), NodeValue::Float(0.5));
+	let frame = eval_node_row(TRANSITIONFX, row, Some((8, 8)), None);
+	assert_pixel(pixel_at(&frame, 2, 2), [0.5, 0.5, 0.0, 1.0], "cross dissolve 0.5");
+}
+
+/// Progress 0 and 1 are the two pictures untouched.
+#[test]
+fn crossdissolve_ends_are_the_inputs() {
+	if !gpu() { eprintln!("no adapter; skipping"); return; }
+	for (progress, want, what) in [
+		(0.0, [1.0, 0.0, 0.0, 1.0], "cross dissolve 0"),
+		(1.0, [0.0, 1.0, 0.0, 1.0], "cross dissolve 1"),
+	] {
+		let mut row = transition_row([1.0, 0.0, 0.0, 1.0], [0.0, 1.0, 0.0, 1.0]);
+		row.insert("progress_in".to_string(), NodeValue::Float(progress));
+		let frame = eval_node_row(TRANSITIONFX, row, Some((8, 8)), None);
+		assert_pixel(pixel_at(&frame, 4, 4), want, what);
+	}
+}
+
+/// Fade mixes the picture up out of `color_in`: 0 is the colour, 1 the
+/// picture, and a blue `color_in` proves the uniform binds by name (the
+/// shader declares no `blend_in` at all). The row always carries the
+/// input's default, exactly as the traverser fills every input in — a
+/// row without `color_in` would leave the uniform zero-filled, which for
+/// a colour is transparent black rather than the input's opaque default.
+#[test]
+fn fade_dips_through_the_color() {
+	if !gpu() { eprintln!("no adapter; skipping"); return; }
+	let cases: [(f64, [f64; 4], [f32; 4]); 3] = [
+		(0.0, [0.0, 0.0, 0.0, 1.0], [0.0, 0.0, 0.0, 1.0]),
+		(0.5, [0.0, 0.0, 1.0, 1.0], [0.5, 0.0, 0.5, 1.0]),
+		(1.0, [0.0, 0.0, 0.0, 1.0], [1.0, 0.0, 0.0, 1.0]),
+	];
+	for (progress, color, want) in cases {
+		let mut row = transition_row([1.0, 0.0, 0.0, 1.0], [0.0, 1.0, 0.0, 1.0]);
+		style(&mut row, 1);
+		row.insert("progress_in".to_string(), NodeValue::Float(progress));
+		row.insert("color_in".to_string(), NodeValue::Color(color));
+		let frame = eval_node_row(TRANSITIONFX, row, Some((8, 8)), None);
+		assert_pixel(pixel_at(&frame, 3, 5), want, "fade");
+	}
+}
+
+/// Wipe at half progress: the incoming picture covers the half the
+/// direction comes from, the outgoing keeps the rest. All four combo
+/// entries are the same path (axis and sign are two steps and two mixes).
+#[test]
+fn wipe_directions_place_the_incoming_picture() {
+	if !gpu() { eprintln!("no adapter; skipping"); return; }
+	let red = [1.0, 0.0, 0.0, 1.0];
+	let green = [0.0, 1.0, 0.0, 1.0];
+	// (direction, pixel, expected) — tex_in is red (From), blend_in green (To).
+	let cases: [(i64, (usize, usize), [f32; 4]); 8] = [
+		(0, (2, 4), green),
+		(0, (6, 4), red),
+		(1, (2, 4), red),
+		(1, (6, 4), green),
+		(2, (4, 2), green),
+		(2, (4, 6), red),
+		(3, (4, 2), red),
+		(3, (4, 6), green),
+	];
+	for (direction, (x, y), want) in cases {
+		let mut row = transition_row(red, green);
+		style(&mut row, 2);
+		row.insert("progress_in".to_string(), NodeValue::Float(0.5));
+		row.insert("direction_in".to_string(), NodeValue::Combo(direction));
+		let frame = eval_node_row(TRANSITIONFX, row, Some((8, 8)), None);
+		assert_pixel(
+			pixel_at(&frame, x, y),
+			want,
+			&format!("wipe direction {direction} at ({x},{y})"),
+		);
+	}
+}
+
+/// Slide: a gradient stands in for the incoming picture, so the value at
+/// a pixel names the column that landed there. At progress 0.25 the
+/// incoming enters from the left with its trailing quarter on screen (the
+/// frame shows the gradient's `[0.75, 1]` columns), while the outgoing
+/// part shows the untouched `tex_in`. The reversed direction mirrors it.
+#[test]
+fn slide_moves_both_pictures_the_named_way() {
+	if !gpu() { eprintln!("no adapter; skipping"); return; }
+	let red = [1.0, 0.0, 0.0, 1.0];
+	// (direction, pixel, expected red channel) at progress 0.25.
+	let cases: [(i64, (usize, usize), f32); 4] = [
+		(0, (0, 4), 0.8125),
+		(0, (6, 4), 1.0),
+		(1, (0, 4), 1.0),
+		(1, (7, 4), 0.1875),
+	];
+	for (direction, (x, y), red_channel) in cases {
+		let mut row = NodeValueRow::new();
+		row.insert("tex_in".to_string(), texture_value(filled_frame((8, 8), red)));
+		row.insert("blend_in".to_string(), texture_value(gradient_frame_x((8, 8))));
+		style(&mut row, 3);
+		row.insert("progress_in".to_string(), NodeValue::Float(0.25));
+		row.insert("direction_in".to_string(), NodeValue::Combo(direction));
+		let frame = eval_node_row(TRANSITIONFX, row, Some((8, 8)), None);
+		assert_pixel(
+			pixel_at(&frame, x, y),
+			[red_channel, 0.0, 0.0, 1.0],
+			&format!("slide direction {direction} at ({x},{y})"),
+		);
+	}
+}
+
+/// Auto-fill: the row carries no `progress_in` at all (the node dropped
+/// it), so the renderer inserts the adjustment layer's own progress. The
+/// same row renders differently for each hook value — without the fill the
+/// uniform would stay 0 and both calls would be pure `tex_in`.
+#[test]
+fn progress_is_filled_from_the_adjustment_layer() {
+	if !gpu() { eprintln!("no adapter; skipping"); return; }
+	let row = transition_row([1.0, 0.0, 0.0, 1.0], [0.0, 1.0, 0.0, 1.0]);
+	let frame = eval_node_row(TRANSITIONFX, row.clone(), Some((8, 8)), Some(0.5));
+	assert_pixel(pixel_at(&frame, 4, 4), [0.5, 0.5, 0.0, 1.0], "auto progress 0.5");
+	let frame = eval_node_row(TRANSITIONFX, row, Some((8, 8)), Some(0.75));
+	assert_pixel(pixel_at(&frame, 4, 4), [0.25, 0.75, 0.0, 1.0], "auto progress 0.75");
+}
+
+/// An explicit `progress_in` on the row wins over the layer sweep.
+#[test]
+fn explicit_progress_beats_the_layer_sweep() {
+	if !gpu() { eprintln!("no adapter; skipping"); return; }
+	let mut row = transition_row([1.0, 0.0, 0.0, 1.0], [0.0, 1.0, 0.0, 1.0]);
+	row.insert("progress_in".to_string(), NodeValue::Float(0.25));
+	let frame = eval_node_row(TRANSITIONFX, row, Some((8, 8)), Some(0.75));
+	assert_pixel(pixel_at(&frame, 4, 4), [0.75, 0.25, 0.0, 1.0], "explicit progress");
+}
diff --git a/crates/oak-task/src/nodeops.rs b/crates/oak-task/src/nodeops.rs
index 28e021dae..05f23cfdc 100644
--- a/crates/oak-task/src/nodeops.rs
+++ b/crates/oak-task/src/nodeops.rs
@@ -31,7 +31,10 @@ use std::sync::{Arc, Mutex, MutexGuard};
 
 use oak_core::videoparams::VideoParams as CommonVideoParams;
 use oak_core::Rational;
-use oak_node::block::{BlockCore, ClipBlockBehavior, GapBlockBehavior, TransitionBlockBehavior};
+use oak_node::block::{
+	AdjustmentBlockBehavior, BlockCore, ClipBlockBehavior, GapBlockBehavior,
+	TransitionBlockBehavior,
+};
 use oak_node::folder::FolderBehavior;
 use oak_node::footage::{FootageBehavior, StreamInfo};
 use oak_node::graph::Graph;
@@ -108,6 +111,9 @@ pub fn block_core_of(graph: &Graph, block: NodeId) -> Option<BlockCore> {
 	if let Some(b) = any.downcast_ref::<TransitionBlockBehavior>() {
 		return Some(b.core.clone());
 	}
+	if let Some(b) = any.downcast_ref::<AdjustmentBlockBehavior>() {
+		return Some(b.core.clone());
+	}
 	None
 }
 
@@ -132,6 +138,10 @@ fn set_block_core(graph: &mut Graph, block: NodeId, f: impl FnOnce(&mut BlockCor
 		f(&mut b.core);
 		return true;
 	}
+	if let Some(b) = any.downcast_mut::<AdjustmentBlockBehavior>() {
+		f(&mut b.core);
+		return true;
+	}
 	false
 }
 
@@ -728,6 +738,18 @@ pub fn block_kind(project: &ProjectRef, block: NodeId) -> BlockKind {
 	}
 }
 
+/// Whether the block is an adjustment layer (neither `BlockKind` covers
+/// it: the enum mirrors the deleted C ABI's block types).
+pub fn block_is_adjustment(project: &ProjectRef, block: NodeId) -> bool {
+	let guard = lock_project(project);
+	guard
+		.graph
+		.get(block)
+		.and_then(|e| e.behavior.as_any())
+		.map(|a| a.downcast_ref::<AdjustmentBlockBehavior>().is_some())
+		.unwrap_or(false)
+}
+
 /// The block's timeline in point (`oaknode_block_get_in`).
 pub fn block_in(project: &ProjectRef, block: NodeId) -> Rational {
 	let guard = lock_project(project);
diff --git a/crates/oak-task/src/project/saveotio.rs b/crates/oak-task/src/project/saveotio.rs
index ce6f49cf0..728a7989d 100644
--- a/crates/oak-task/src/project/saveotio.rs
+++ b/crates/oak-task/src/project/saveotio.rs
@@ -143,8 +143,10 @@ impl SaveOTIOTask {
 	}
 
 	/// Serialize one track: a `Track` whose kind matches the native track
-	/// type, one OTIO block per native block, plus a trailing `Gap` when the
-	/// track is shorter than `max_track_length`.
+	/// type, one OTIO block per native block (an adjustment layer, which has
+	/// no OTIO equivalent, becomes an equal-length `Gap` so later children
+	/// keep their timeline positions), plus a trailing `Gap` when the track
+	/// is shorter than `max_track_length`.
 	///
 	/// CPP-PARITY: saveotio.cpp (SaveOTIOTask::serialize_track)
 	fn serialize_track(
@@ -258,7 +260,28 @@ impl SaveOTIOTask {
 
 					Some(Composable::Transition(otio_transition))
 				}
-				nodeops::BlockKind::Other => None,
+				nodeops::BlockKind::Other => {
+					// Adjustment layers have no OTIO equivalent. OTIO places
+					// children sequentially, so dropping the block would shift
+					// every following child left; keeping its span as a Gap
+					// holds the timeline in place (and round-trips as a gap,
+					// which is what "no adjustment applied" means here).
+					nodeops::block_is_adjustment(project, block).then(|| {
+						Composable::Gap(Gap::new(
+							TimeRange::new(
+								RationalTime::from_rational(
+									block_in_of(project, block),
+									sequence_rate,
+								),
+								RationalTime::from_rational(
+									block_length_of(project, block),
+									sequence_rate,
+								),
+							),
+							nodeops::node_label(project, block),
+						))
+					})
+				}
 			};
 
 			let Some(otio_block) = otio_block else {
diff --git a/crates/oak-task/src/render.rs b/crates/oak-task/src/render.rs
index acaf5682e..29e03a425 100644
--- a/crates/oak-task/src/render.rs
+++ b/crates/oak-task/src/render.rs
@@ -542,6 +542,7 @@ impl RenderTask {
 				cache_timebase: None,
 				footage: Some((filename, 0)),
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			}),
 			// Sequence viewer: the montage is resolved without the lock.
 			None => {
@@ -558,6 +559,7 @@ impl RenderTask {
 					cache_timebase: None,
 					footage: None,
 					montage,
+					adjustments: Vec::new(),
 				})
 			}
 		}
diff --git a/crates/oak-timeline/src/undogeneral.rs b/crates/oak-timeline/src/undogeneral.rs
index 5fea8e024..71b79ca8e 100644
--- a/crates/oak-timeline/src/undogeneral.rs
+++ b/crates/oak-timeline/src/undogeneral.rs
@@ -25,7 +25,8 @@
 //! pointers (see `undocommon`). Detached blocks are held as owned arena
 //! entries on the command that detached them.
 
-use oak_core::Rational;
+use oak_core::{Rational, TimeRange};
+use oak_node::block::transition_input::{IN_BLOCK, OUT_BLOCK};
 use oak_node::graph::NodeEntry;
 use oak_node::track::TrackType;
 use oak_undo::undocommand::UndoCommand;
@@ -33,13 +34,15 @@ use oak_undo::undocommand::UndoCommand;
 use crate::undocommon::{box_command, create_block_remove_command, Command};
 use crate::undosplit::BlockSplitPreservingLinksCommand;
 use crate::util::{
-	block_add_to_graph, block_enabled, block_gap_create, block_in, block_kind, block_length,
-	block_next, block_out, block_previous, block_remove_from_graph, block_set_enabled,
-	block_set_in, block_set_length_and_media_in, block_set_length_and_media_out, block_track,
+	block_add_to_graph, block_connect, block_connected_input, block_disconnect_input,
+	block_enabled, block_gap_create, block_in, block_kind, block_length, block_next, block_out,
+	block_previous, block_range, block_remove_from_graph, block_set_enabled, block_set_in,
+	block_set_length_and_media_in, block_set_length_and_media_out, block_set_range, block_track,
 	clip_media_in, clip_set_media_in, same_block, track_append_block, track_create,
 	track_insert_block_after, track_insert_block_before, track_replace_block,
 	track_ripple_remove_block, tracklist_append, tracklist_remove, tracklist_remove_last,
-	tracklist_track_at, tracklist_track_count, tracklist_type, BlockKind, NodeRef,
+	tracklist_track_at, tracklist_track_count, tracklist_type, transition_offsets,
+	transition_set_offsets, BlockKind, NodeRef,
 };
 
 // `oaknode/sequence.h` element input ids (the C++ automerge branch).
@@ -540,6 +543,104 @@ impl Command for TimelineRemoveTrackCommand {
 	}
 }
 
+/// `TransitionSetOffsetsCommand` — change a transition block's two wedge
+/// offsets, keeping the block pinned to the same seam.
+///
+/// The C++ timeline has no equivalent (a transition's offsets are written
+/// once, at creation); the Rust port needs one so the timeline's wedge drag
+/// (`TransitionChanged`) can resize a transition undoably. A transition spans
+/// `seam - in_offset .. seam + out_offset`, so the stored range follows the
+/// offsets — growing one wedge without moving the matching range end would
+/// drift the block off its seam.
+pub struct TransitionSetOffsetsCommand {
+	/// Transition block to change.
+	block: NodeRef,
+	/// New in offset (`None` keeps the current one).
+	new_in_offset: Option<Rational>,
+	/// New out offset (`None` keeps the current one).
+	new_out_offset: Option<Rational>,
+	/// Whether the pre-change state has been captured.
+	captured: bool,
+	/// In offset captured at `redo` time, restored by `undo`.
+	old_in_offset: Rational,
+	/// Out offset captured at `redo` time, restored by `undo`.
+	old_out_offset: Rational,
+	/// Stored range captured at `redo` time, restored by `undo`.
+	old_range: TimeRange,
+	/// The seam the offsets straddle, captured at `redo` time.
+	seam: Rational,
+}
+
+impl TransitionSetOffsetsCommand {
+	/// Construct from a transition block and the offsets to set (`None`
+	/// leaves that side as it is).
+	///
+	/// New signature (single-lib): `pub fn new(block: NodeRef, new_in_offset: Option<Rational>, new_out_offset: Option<Rational>) -> TransitionSetOffsetsCommand`
+	pub fn new(
+		block: NodeRef,
+		new_in_offset: Option<Rational>,
+		new_out_offset: Option<Rational>,
+	) -> Self {
+		Self {
+			block,
+			new_in_offset,
+			new_out_offset,
+			captured: false,
+			old_in_offset: Rational::new(0, 1),
+			old_out_offset: Rational::new(0, 1),
+			old_range: TimeRange::default(),
+			seam: Rational::new(0, 1),
+		}
+	}
+
+	/// `redo`: capture the offsets/range once, then write the new offsets
+	/// and move the range with them.
+	pub fn redo(&mut self) {
+		if !self.captured {
+			if let Some((in_offset, out_offset)) = transition_offsets(&self.block) {
+				self.old_in_offset = in_offset;
+				self.old_out_offset = out_offset;
+			}
+			if let Some(range) = block_range(&self.block) {
+				self.old_range = range;
+			}
+			self.seam = block_in(&self.block) + self.old_in_offset;
+			self.captured = true;
+		}
+
+		let in_offset = self.new_in_offset.unwrap_or(self.old_in_offset);
+		let out_offset = self.new_out_offset.unwrap_or(self.old_out_offset);
+		transition_set_offsets(&self.block, in_offset, out_offset);
+		block_set_range(
+			&self.block,
+			TimeRange::new(self.seam - in_offset, self.seam + out_offset),
+		);
+	}
+
+	/// `undo`: restore the captured offsets and range.
+	pub fn undo(&mut self) {
+		transition_set_offsets(&self.block, self.old_in_offset, self.old_out_offset);
+		block_set_range(&self.block, self.old_range);
+	}
+
+	/// Wrap as an oakundo command value.
+	pub fn to_command(self) -> UndoCommand {
+		box_command(self)
+	}
+}
+
+impl Command for TransitionSetOffsetsCommand {
+	/// `Command::redo` — the inherent method takes precedence.
+	fn redo(&mut self) {
+		self.redo();
+	}
+
+	/// `Command::undo` — the inherent method takes precedence.
+	fn undo(&mut self) {
+		self.undo();
+	}
+}
+
 /// `TransitionRemoveCommand` — detach a transition block from its neighbours,
 /// optionally removing it from the whole graph (timelineundogeneral.h).
 pub struct TransitionRemoveCommand {
@@ -553,6 +654,20 @@ pub struct TransitionRemoveCommand {
 	out_block: Option<NodeRef>,
 	/// Block before the transition.
 	in_block: Option<NodeRef>,
+	/// Whether the pre-removal state has been captured.
+	captured: bool,
+	/// In offset captured before removal, restored by `undo`.
+	in_offset: Rational,
+	/// Out offset captured before removal, restored by `undo`.
+	out_offset: Rational,
+	/// Stored range captured before removal, restored by `undo`.
+	range: TimeRange,
+	/// Node feeding the transition's `out_block_in` input (the outgoing
+	/// clip), captured before removal and restored by `undo`.
+	connected_out: Option<NodeRef>,
+	/// Node feeding the transition's `in_block_in` input (the incoming
+	/// clip), captured before removal and restored by `undo`.
+	connected_in: Option<NodeRef>,
 	/// Graph-removal command built at `redo` time.
 	remove_command: Option<UndoCommand>,
 }
@@ -568,19 +683,41 @@ impl TransitionRemoveCommand {
 			track: None,
 			out_block: None,
 			in_block: None,
+			captured: false,
+			in_offset: Rational::new(0, 1),
+			out_offset: Rational::new(0, 1),
+			range: TimeRange::default(),
+			connected_out: None,
+			connected_in: None,
 			remove_command: None,
 		}
 	}
 
-	/// `redo`: relink neighbours around the transition and remove it.
+	/// `redo`: unlink the transition from its neighbours and remove it.
 	pub fn redo(&mut self) {
 		self.track = block_track(&self.block);
 		self.out_block = block_next(&self.block);
 		self.in_block = block_previous(&self.block);
-		// NOTE: the C++ extends the neighbouring blocks by the transition
-		// offsets and disconnects the transition's inputs; the Rust block
-		// model has no transition edges, so only the ripple-remove is
-		// performed.
+		// The C++ also extends the neighbouring blocks by the transition's
+		// offsets; the Rust blocks already meet at the seam once the
+		// transition leaves the track (a transition's span overlaps its
+		// neighbours instead of filling a gap of its own), so only the detach
+		// is performed. The offsets, the stored range and both input edges
+		// are captured first so `undo` can put the block back as it was.
+		if !self.captured {
+			if let Some((in_offset, out_offset)) = transition_offsets(&self.block) {
+				self.in_offset = in_offset;
+				self.out_offset = out_offset;
+			}
+			if let Some(range) = block_range(&self.block) {
+				self.range = range;
+			}
+			self.connected_out = block_connected_input(&self.block, OUT_BLOCK);
+			self.connected_in = block_connected_input(&self.block, IN_BLOCK);
+			self.captured = true;
+		}
+		block_disconnect_input(&self.block, OUT_BLOCK);
+		block_disconnect_input(&self.block, IN_BLOCK);
 		if let Some(track) = &self.track {
 			track_ripple_remove_block(track, &self.block);
 		}
@@ -594,23 +731,27 @@ impl TransitionRemoveCommand {
 		}
 	}
 
-	/// `undo`: restore the transition between its neighbours.
+	/// `undo`: restore the transition, its offsets and its two edges.
 	pub fn undo(&mut self) {
 		if self.remove_from_graph {
 			if let Some(c) = self.remove_command.as_mut() {
 				c.undo_now();
 			}
 		}
-		// NOTE: the C++ re-inserts the transition between its former
-		// neighbours, reconnects them and restores the offsets; the
-		// re-insert is real, the offset/connection restoration is not
-		// modelled (no transition edges in the Rust block model).
 		if let Some(track) = &self.track {
 			match &self.in_block {
 				Some(before) => track_insert_block_after(track, &self.block, Some(before)),
 				None => track_insert_block_after(track, &self.block, None),
 			}
 		}
+		transition_set_offsets(&self.block, self.in_offset, self.out_offset);
+		block_set_range(&self.block, self.range);
+		if let Some(source) = &self.connected_out {
+			block_connect(source, &self.block, OUT_BLOCK);
+		}
+		if let Some(source) = &self.connected_in {
+			block_connect(source, &self.block, IN_BLOCK);
+		}
 	}
 
 	/// Wrap as an oakundo command value.
@@ -831,9 +972,10 @@ impl TrackReplaceBlockWithGapCommand {
 		}
 		// NOTE: the C++ checks whether the transition is connected only to this
 		// block (via `oaknode_transition_get_connected_out_block`/`_in_block` +
-		// `same_block`) before adding a `TransitionRemoveCommand`; the Rust
-		// block model has no transition edges, so a command removing the
-		// transition outright is produced.
+		// `same_block`) before adding a `TransitionRemoveCommand`; that guard is
+		// still not ported (the inputs are reachable through
+		// `util::block_connected_input`), so a command removing the transition
+		// outright is produced.
 		self.transition_remove_commands
 			.push(TransitionRemoveCommand::new(relevant, true));
 	}
diff --git a/crates/oak-timeline/src/util.rs b/crates/oak-timeline/src/util.rs
index 59473ce8e..766244e7c 100644
--- a/crates/oak-timeline/src/util.rs
+++ b/crates/oak-timeline/src/util.rs
@@ -33,7 +33,10 @@
 use std::sync::{Arc, Mutex, MutexGuard};
 
 use oak_core::Rational;
-use oak_node::block::{BlockCore, ClipBlockBehavior, GapBlockBehavior, TransitionBlockBehavior};
+use oak_node::block::{
+	AdjustmentBlockBehavior, BlockCore, ClipBlockBehavior, GapBlockBehavior,
+	TransitionBlockBehavior,
+};
 use oak_node::graph::{Graph, NodeEntry};
 use oak_node::id::NodeId;
 use oak_node::node::NodeCore;
@@ -122,6 +125,8 @@ fn block_core_of(project: &Project, id: NodeId) -> Option<&BlockCore> {
 			Some(&gap.core)
 		} else if let Some(transition) = a.downcast_ref::<TransitionBlockBehavior>() {
 			Some(&transition.core)
+		} else if let Some(adjustment) = a.downcast_ref::<AdjustmentBlockBehavior>() {
+			Some(&adjustment.core)
 		} else {
 			None
 		}
@@ -143,6 +148,13 @@ fn block_core_of_mut(project: &mut Project, id: NodeId) -> Option<&mut BlockCore
 				.expect("checked above")
 				.core,
 		)
+	} else if a.is::<AdjustmentBlockBehavior>() {
+		Some(
+			&mut a
+				.downcast_mut::<AdjustmentBlockBehavior>()
+				.expect("checked above")
+				.core,
+		)
 	} else {
 		None
 	}
@@ -213,6 +225,8 @@ impl oak_node::track::BlockRange for GraphBlockRange<'_> {
 					a.downcast_ref::<GapBlockBehavior>().map(|b| &b.core),
 					a.downcast_ref::<TransitionBlockBehavior>()
 						.map(|b| &b.core),
+					a.downcast_ref::<AdjustmentBlockBehavior>()
+						.map(|b| &b.core),
 				]
 				.into_iter()
 				.flatten()
@@ -233,6 +247,8 @@ impl oak_node::track::BlockRange for GraphBlockRange<'_> {
 					a.downcast_ref::<GapBlockBehavior>().map(|b| &b.core),
 					a.downcast_ref::<TransitionBlockBehavior>()
 						.map(|b| &b.core),
+					a.downcast_ref::<AdjustmentBlockBehavior>()
+						.map(|b| &b.core),
 				]
 				.into_iter()
 				.flatten()
@@ -339,6 +355,76 @@ pub fn clip_set_media_in(b: &NodeRef, media_in: Rational) {
 	}
 }
 
+/// The block's stored timeline range (`None` for a stale/non-block node).
+pub fn block_range(b: &NodeRef) -> Option<oak_core::TimeRange> {
+	let p = b.lock();
+	block_core_of(&p, b.id).map(|c| c.range)
+}
+
+/// Set the block's stored timeline range (a no-op for a stale/non-block
+/// node). The range is written whole, so the caller can move the in/out
+/// pair together without a media-in side effect (unlike
+/// [`block_set_length_and_media_out`] / [`block_set_length_and_media_in`]).
+pub fn block_set_range(b: &NodeRef, range: oak_core::TimeRange) {
+	let mut p = b.lock();
+	if let Some(core) = block_core_of_mut(&mut p, b.id) {
+		core.range = range;
+	}
+}
+
+// ---------------------------------------------------------------------------
+// Transition blocks
+// ---------------------------------------------------------------------------
+
+/// The transition block's `(in_offset, out_offset)` — the two wedges that
+/// reach into its neighbours — or `None` for a non-transition node.
+pub fn transition_offsets(t: &NodeRef) -> Option<(Rational, Rational)> {
+	let p = t.lock();
+	let a = p.graph.get(t.id)?.behavior.as_any()?;
+	let transition = a.downcast_ref::<TransitionBlockBehavior>()?;
+	Some((transition.in_offset, transition.out_offset))
+}
+
+/// Set the transition block's offsets (a no-op for a non-transition node).
+pub fn transition_set_offsets(t: &NodeRef, in_offset: Rational, out_offset: Rational) {
+	let mut p = t.lock();
+	if let Some(a) = p.graph.get_mut(t.id).and_then(|e| e.behavior.as_any_mut()) {
+		if let Some(transition) = a.downcast_mut::<TransitionBlockBehavior>() {
+			transition.in_offset = in_offset;
+			transition.out_offset = out_offset;
+		}
+	}
+}
+
+// ---------------------------------------------------------------------------
+// Graph edges (the transition block's two sides)
+// ---------------------------------------------------------------------------
+
+/// The node feeding `b`'s `input` (the input's connected source), or `None`
+/// when the input is unconnected or the node is stale.
+pub fn block_connected_input(b: &NodeRef, input: &str) -> Option<NodeRef> {
+	let p = b.lock();
+	p.graph
+		.connected_output(b.id, input, -1)
+		.map(|id| NodeRef::new(b.project.clone(), id))
+}
+
+/// Connect `from`'s output into `to`'s `input`. No undo row of its own: the
+/// calling command owns its undo/redo semantics.
+pub fn block_connect(from: &NodeRef, to: &NodeRef, input: &str) {
+	let mut p = from.lock();
+	let _ = p.graph.connect(from.id, to.id, input, -1);
+}
+
+/// Disconnect `to`'s `input`, returning the node that fed it (`None` when
+/// nothing was connected).
+pub fn block_disconnect_input(to: &NodeRef, input: &str) -> Option<NodeRef> {
+	let mut p = to.lock();
+	p.graph
+		.disconnect_input(to.id, input, -1)
+		.map(|id| NodeRef::new(to.project.clone(), id))
+}
+
 // ---------------------------------------------------------------------------
 // Block navigation (track membership)
 // ---------------------------------------------------------------------------
diff --git a/crates/oak-timeline/tests/domain_test.rs b/crates/oak-timeline/tests/domain_test.rs
index a93f4bbec..7a3f571cf 100644
--- a/crates/oak-timeline/tests/domain_test.rs
+++ b/crates/oak-timeline/tests/domain_test.rs
@@ -25,22 +25,26 @@
 use std::sync::{Arc, Mutex};
 
 use oak_core::{Rational, TimeRange};
-use oak_node::block::{ClipBlockBehavior, GapBlockBehavior};
+use oak_node::block::{
+	transition_input::{IN_BLOCK, OUT_BLOCK},
+	ClipBlockBehavior, GapBlockBehavior, TransitionBlockBehavior,
+};
 use oak_node::project::Project;
 use oak_node::sequence::SequenceBehavior;
 use oak_node::track::{TrackBehavior, TrackListBehavior, TrackType};
 use oak_timeline::undocommon::Command;
 use oak_timeline::undogeneral::{
 	BlockResizeCommand, TimelineAddTrackCommand, TimelineRemoveTrackCommand,
-	TrackReplaceBlockWithGapCommand,
+	TrackReplaceBlockWithGapCommand, TransitionRemoveCommand, TransitionSetOffsetsCommand,
 };
 use oak_timeline::undopointer::TrackMoveBlockCommand;
 use oak_timeline::undoripple::TrackRippleRemoveAreaCommand;
 use oak_timeline::undosplit::BlockSplitCommand;
 use oak_timeline::util::{
-	block_clip_create, block_gap_create, block_in, block_length, block_out, block_track,
-	track_append_block, track_block_at, track_block_count, track_length, track_ripple_remove_block,
-	tracklist_track_count, NodeRef,
+	block_clip_create, block_connect, block_connected_input, block_gap_create, block_in,
+	block_length, block_out, block_range, block_track, track_append_block, track_block_at,
+	track_block_count, track_length, track_ripple_remove_block, tracklist_track_count,
+	transition_offsets, transition_set_offsets, NodeRef,
 };
 
 /// A project with one sequence owning one (video) track list.
@@ -134,6 +138,32 @@ fn add_gap(track: &NodeRef, in_: Rational, out: Rational) -> NodeRef {
 	gap
 }
 
+/// Add a transition block spanning `[in, out)` with the given wedge
+/// offsets to `track`.
+fn add_transition(
+	track: &NodeRef,
+	in_: Rational,
+	out: Rational,
+	in_offset: Rational,
+	out_offset: Rational,
+) -> NodeRef {
+	let (core, behavior) = oak_node::block::transition_create();
+	let transition = {
+		let mut p = track.project.lock().unwrap();
+		let id = p.graph.add_node(core, behavior);
+		if let Some(a) = p.graph.get_mut(id).and_then(|e| e.behavior.as_any_mut()) {
+			if let Some(t) = a.downcast_mut::<TransitionBlockBehavior>() {
+				t.core.range = TimeRange::new(in_, out);
+				t.in_offset = in_offset;
+				t.out_offset = out_offset;
+			}
+		}
+		NodeRef::new(track.project.clone(), id)
+	};
+	track_append_block(track, &transition);
+	transition
+}
+
 /// The block span of `block` (`None` for a stale node).
 fn span_of(block: &NodeRef) -> Option<(Rational, Rational)> {
 	let p = block.project.lock().unwrap();
@@ -480,3 +510,193 @@ fn insert_gaps_round_trip() {
 		Some((Rational::new(0, 1), Rational::new(100, 1)))
 	);
 }
+
+/// `TransitionSetOffsetsCommand` rewrites the wedge offsets and moves the
+/// stored range with them so the block stays pinned to its seam; undo
+/// restores both.
+#[test]
+fn transition_set_offsets_round_trip() {
+	let project = make_project();
+	let (_seq, list) = sequence_and_list(&project);
+	let track = TimelineAddTrackCommand::run_immediately(list);
+	add_clip(&track, Rational::new(0, 1), Rational::new(50, 1));
+	// The transition overlaps both neighbours around the seam at 50.
+	let transition = add_transition(
+		&track,
+		Rational::new(40, 1),
+		Rational::new(60, 1),
+		Rational::new(10, 1),
+		Rational::new(10, 1),
+	);
+	add_clip(&track, Rational::new(50, 1), Rational::new(100, 1));
+
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(10, 1), Rational::new(10, 1)))
+	);
+
+	// The helper writes both offsets directly.
+	transition_set_offsets(&transition, Rational::new(12, 1), Rational::new(8, 1));
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(12, 1), Rational::new(8, 1)))
+	);
+	transition_set_offsets(&transition, Rational::new(10, 1), Rational::new(10, 1));
+
+	let mut cmd = TransitionSetOffsetsCommand::new(
+		transition.clone(),
+		Some(Rational::new(5, 1)),
+		Some(Rational::new(20, 1)),
+	);
+	cmd.redo();
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(5, 1), Rational::new(20, 1)))
+	);
+	// Range follows the offsets: [seam - in, seam + out), seam = 40 + 10.
+	assert_eq!(
+		block_range(&transition),
+		Some(TimeRange::new(Rational::new(45, 1), Rational::new(70, 1)))
+	);
+
+	cmd.undo();
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(10, 1), Rational::new(10, 1)))
+	);
+	assert_eq!(
+		block_range(&transition),
+		Some(TimeRange::new(Rational::new(40, 1), Rational::new(60, 1)))
+	);
+
+	// `None` keeps the offset on that side.
+	let mut partial =
+		TransitionSetOffsetsCommand::new(transition.clone(), Some(Rational::new(7, 1)), None);
+	partial.redo();
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(7, 1), Rational::new(10, 1)))
+	);
+	assert_eq!(
+		block_range(&transition),
+		Some(TimeRange::new(Rational::new(43, 1), Rational::new(60, 1)))
+	);
+
+	partial.undo();
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(10, 1), Rational::new(10, 1)))
+	);
+	assert_eq!(
+		block_range(&transition),
+		Some(TimeRange::new(Rational::new(40, 1), Rational::new(60, 1)))
+	);
+}
+
+/// `TransitionRemoveCommand` unlinks the transition from its neighbours and
+/// its two node inputs; undo puts it back between them with the offsets,
+/// the range and both edges restored.
+#[test]
+fn transition_remove_restores_offsets_and_edges() {
+	let project = make_project();
+	let (_seq, list) = sequence_and_list(&project);
+	let track = TimelineAddTrackCommand::run_immediately(list);
+	let a = add_clip(&track, Rational::new(0, 1), Rational::new(50, 1));
+	let transition = add_transition(
+		&track,
+		Rational::new(40, 1),
+		Rational::new(60, 1),
+		Rational::new(10, 1),
+		Rational::new(10, 1),
+	);
+	let b = add_clip(&track, Rational::new(50, 1), Rational::new(100, 1));
+	block_connect(&a, &transition, OUT_BLOCK);
+	block_connect(&b, &transition, IN_BLOCK);
+	assert_eq!(
+		block_connected_input(&transition, OUT_BLOCK).map(|n| n.id),
+		Some(a.id)
+	);
+	assert_eq!(track_block_count(&track), 3);
+
+	let mut cmd = TransitionRemoveCommand::new(transition.clone(), false);
+	cmd.redo();
+	// The transition left the track and both clips remain; the edges are
+	// gone too.
+	assert_eq!(track_block_count(&track), 2);
+	assert_eq!(track_block_at(&track, 0).unwrap().id, a.id);
+	assert_eq!(track_block_at(&track, 1).unwrap().id, b.id);
+	assert!(block_track(&transition).is_none());
+	assert!(block_connected_input(&transition, OUT_BLOCK).is_none());
+	assert!(block_connected_input(&transition, IN_BLOCK).is_none());
+
+	cmd.undo();
+	assert_eq!(track_block_count(&track), 3);
+	assert_eq!(track_block_at(&track, 0).unwrap().id, a.id);
+	assert_eq!(track_block_at(&track, 1).unwrap().id, transition.id);
+	assert_eq!(track_block_at(&track, 2).unwrap().id, b.id);
+	assert_eq!(block_track(&transition).map(|t| t.id), Some(track.id));
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(10, 1), Rational::new(10, 1)))
+	);
+	assert_eq!(
+		block_range(&transition),
+		Some(TimeRange::new(Rational::new(40, 1), Rational::new(60, 1)))
+	);
+	assert_eq!(
+		block_connected_input(&transition, OUT_BLOCK).map(|n| n.id),
+		Some(a.id)
+	);
+	assert_eq!(
+		block_connected_input(&transition, IN_BLOCK).map(|n| n.id),
+		Some(b.id)
+	);
+}
+
+/// `TransitionRemoveCommand` with `remove_from_graph` detaches the node
+/// from the project graph as well; undo re-attaches it with the same
+/// identity, track slot, offsets and edges.
+#[test]
+fn transition_remove_from_graph_round_trip() {
+	let project = make_project();
+	let (_seq, list) = sequence_and_list(&project);
+	let track = TimelineAddTrackCommand::run_immediately(list);
+	let a = add_clip(&track, Rational::new(0, 1), Rational::new(50, 1));
+	let transition = add_transition(
+		&track,
+		Rational::new(40, 1),
+		Rational::new(60, 1),
+		Rational::new(10, 1),
+		Rational::new(10, 1),
+	);
+	let b = add_clip(&track, Rational::new(50, 1), Rational::new(100, 1));
+	block_connect(&a, &transition, OUT_BLOCK);
+	block_connect(&b, &transition, IN_BLOCK);
+
+	let mut cmd = TransitionRemoveCommand::new(transition.clone(), true);
+	cmd.redo();
+	assert_eq!(track_block_count(&track), 2);
+	assert!(block_track(&transition).is_none());
+	assert!(project.lock().unwrap().graph.get(transition.id).is_none());
+
+	cmd.undo();
+	assert!(project.lock().unwrap().graph.get(transition.id).is_some());
+	assert_eq!(track_block_count(&track), 3);
+	assert_eq!(track_block_at(&track, 1).unwrap().id, transition.id);
+	assert_eq!(
+		transition_offsets(&transition),
+		Some((Rational::new(10, 1), Rational::new(10, 1)))
+	);
+	assert_eq!(
+		block_range(&transition),
+		Some(TimeRange::new(Rational::new(40, 1), Rational::new(60, 1)))
+	);
+	assert_eq!(
+		block_connected_input(&transition, OUT_BLOCK).map(|n| n.id),
+		Some(a.id)
+	);
+	assert_eq!(
+		block_connected_input(&transition, IN_BLOCK).map(|n| n.id),
+		Some(b.id)
+	);
+}
diff --git a/crates/oak-worker/src/worker.rs b/crates/oak-worker/src/worker.rs
index 7db4378aa..dff630191 100644
--- a/crates/oak-worker/src/worker.rs
+++ b/crates/oak-worker/src/worker.rs
@@ -57,7 +57,7 @@ use serde_json::{json, Value};
 use oak_core::backend::{BackendKind, DisplayRenderer};
 use oak_core::{PixelFormat, Rational};
 use oak_render::eval;
-use oak_render::ticket::{AudioTicketParams, MontageClip, VideoTicketParams};
+use oak_render::ticket::{AdjustmentSpan, AudioTicketParams, MontageClip, VideoTicketParams};
 
 use crate::framecache::FrameCache;
 use crate::ipc::{
@@ -1204,6 +1204,11 @@ impl WorkerSession {
 				effects: c.effects.iter().map(crate::ipc::montage_effect_from).collect(),
 			})
 			.collect();
+		let adjustments: Vec<AdjustmentSpan> = spec
+			.adjustments
+			.iter()
+			.map(oak_render::ipc::adjustment_from_wire)
+			.collect();
 		VideoTicketParams {
 			viewer: if spec.viewer_node != 0 {
 				spec.viewer_node
@@ -1223,6 +1228,7 @@ impl WorkerSession {
 			cache_timebase: None,
 			footage,
 			montage,
+			adjustments,
 		}
 	}
 }
diff --git a/crates/oak-worker/tests/procpool_integration.rs b/crates/oak-worker/tests/procpool_integration.rs
index 08c6d3a7e..a432f26fd 100644
--- a/crates/oak-worker/tests/procpool_integration.rs
+++ b/crates/oak-worker/tests/procpool_integration.rs
@@ -89,6 +89,7 @@ fn params(time: Rational, footage: Option<(String, i32)>) -> Arc<VideoTicketPara
 		cache_timebase: None,
 		footage,
 		montage: Vec::new(),
+		adjustments: Vec::new(),
 	})
 }
 
@@ -537,6 +538,7 @@ fn submit_audio(
 			cache_timebase: None,
 			footage: None,
 			montage: Vec::new(),
+			adjustments: Vec::new(),
 		}),
 		audio: Some(audio),
 		// Never invoked on the process backend (the worker renders audio
@@ -755,6 +757,7 @@ fn oversized_audio_ticket_is_refused_by_process_backend() {
 			cache_timebase: None,
 			footage: None,
 			montage: Vec::new(),
+			adjustments: Vec::new(),
 		}),
 		audio: Some(audio),
 		produce: Arc::new(|_, _| {
@@ -817,6 +820,7 @@ fn f32_ticket_gets_f32_slot_and_bgra8_stays_bgra8() {
 				cache_timebase: None,
 				footage: None,
 				montage: Vec::new(),
+				adjustments: Vec::new(),
 			}),
 			audio: None,
 			produce: Arc::new(|_, _| {
@@ -961,6 +965,7 @@ fn post_graph_job(
 			cache_timebase: None,
 			footage: None,
 			montage: Vec::new(),
+			adjustments: Vec::new(),
 		}),
 		audio: None,
 		produce: Arc::new(|_, _| {
diff --git a/docs/zh/plans/adjustment-layers-and-transitions.md b/docs/zh/plans/adjustment-layers-and-transitions.md
new file mode 100644
index 000000000..62a34546e
--- /dev/null
+++ b/docs/zh/plans/adjustment-layers-and-transitions.md
@@ -0,0 +1,253 @@
+# 调整图层与转场特效改造计划
+
+> 面向实现者的任务书（2026-09-10）。本文只描述方案与工作项，不含已执行的代码修改。
+> 用户要求："增加一项新功能：调整图层。可以将特效应用到多个 clip 那种。增加交叉叠化
+> 等几种常见的转场特效。我觉得转场特效可以做成应用在调整图层上的节点。注意转场开始
+> 和结束点，不能覆盖整个 clip，这要求调整图层也能针对 clip 的一段设置。"
+>
+> 配套调研：调整图层/转场基建审计（2026-09-10 会话，结论已并入本文各"现状"节）。
+
+## 1. 需求分解
+
+1. **调整图层（Adjustment Layer）**：时间轴上的一种块，位于某条视频轨、占一段时间
+   范围，其效果链作用于**它下方所有视频轨在该时刻的合成结果**。一段范围可覆盖
+   多个 clip（"将特效应用到多个 clip"），也可只覆盖一个 clip 的一段（"针对 clip
+   的一段设置"）。
+2. **转场特效**：交叉叠化（Cross Dissolve）等若干常见转场，有明确的**开始和结束
+   点**（不能覆盖整个 clip）。
+3. 用户倾向的实现形态：转场做成**应用在调整图层上的节点**（"我觉得转场特效可以
+   做成应用在调整图层上的节点"）——本文 §4 给出该形态的落地设计，同时指出其与
+   一等转场块的关系。
+
+## 2. 现状（关键事实，均带坐标）
+
+### 2.1 转场块模型已存在，渲染/创建/UI 全空
+
+- `crates/oak-node/src/block.rs:129-137`：`TransitionBlockBehavior { core: BlockCore,
+  in_offset, out_offset }`，与普通块一样挂在 `TrackBehavior.blocks`。
+  构造 `transition_create()`（block.rs:587-608）声明 `out_block_in`/`in_block_in`
+  两个 NodeRef 输入（接前后块）。
+- `value()` 是空实现（block.rs:501-510，注释明确"渲染成一个洞"）；
+  `render_graph_frame`（`crates/oak-render/src/eval.rs:1466-1472`）与
+  `video_montage`（`renderops.rs:270`）都只认 `ClipBlockBehavior`，转场块被跳过。
+- 创建路径只有 OTIO 导入（`loadotio.rs:249-333`，含连线与 offset 设置）；
+  OTIO/FCPXML 导出已支持（saveotio.rs:250-259、fcpxml.rs:596-600）。
+- gpui 时间轴有转场楔形绘制与 `TransitionChanged` 事件（timeline/data.rs:122-135、
+  clip.rs:346-377、timeline_view.rs:304-316），但 `RealClip` 不喂数据
+  （`real.rs:873-897` 用默认 None）、`real.rs:4683` 明确丢弃该事件。
+- `oak-timeline` 有 `TransitionRemoveCommand`（undogeneral.rs:543-632，edge/offset
+  恢复未建模）。行动注册表有 `Tool::Transition`/`DefaultTransition`(Ctrl+Shift+D)
+  但无后端（actions.rs:131/229/733/778-782）。
+
+### 2.2 调整图层：零代码，但模型/渲染/撤销的接入点都已确认
+
+- 块基类 `BlockCore`（block.rs:28-113）：range/media_in/speed/links/enabled/track。
+  clip 的效果链挂载点 = `tex_in` 输入 + `core.effect_input`（block.rs:527-534），
+  effectchain 的 insert/remove/move/set_enabled **与 clip 无关，任何有 effect
+  input 的节点都能挂**。
+- 渲染主路径 `render_graph_frame`（eval.rs:1409-1546）：逐 clip traverser 求值 →
+  `composite_tracks` 自底向上 alpha-over。**调整图层的挂钩点 = eval.rs:1451-1478
+  的轨道循环：遇到调整块时先把已收集的下层帧合成，跑调整块效果链，再往上。**
+- 备选渲染路径 montage：`VideoTicketParams.montage`（ticket.rs:67-83 的
+  `MontageClip{effects: Vec<MontageEffect>}`），由 `renderops.rs:249-309
+  video_montage` 构建。**两条渲染路径语义必须一致**（worker 快照图渲染优先、
+  montage 兜底：worker.rs:1253-1299、manager.rs:119-151），调整图层两边都要实现。
+- 撤销：`graphops::place_footage_clip`（graphops.rs:1445-1506）=
+  非撤销建节点 + 一条 undo 推 `TrackPlaceBlockCommand` + 连线；
+  `TrackPlaceBlockCommand`（undopointer.rs:541-710）与块类型无关；
+  graphops 的 clip 访问器（clip_ids/clip_range/trim_command 等）目前 downcast 到
+  ClipBlockBehavior，需要推广到"任何带 BlockCore 的块"（参照
+  `nodeops::block_core_of/set_block_core`，nodeops.rs:99-136 的模式）。
+- "给效果链喂一张外来纹理"目前**没有**现成 API：链首效果的 `tex_in` 未连接时
+  取不到值（traverser.rs:247-251）。需要 §4.2 的"纹理源"机制。
+- UI：块绘制按 `ClipData.color/label` 即可区分外观（如需斜纹装饰才动
+  gpui/crates/gpui/src/timeline/clip.rs）；移动/修剪/分割/删除对画成 RealClip 的
+  新块**免费**（只要 §2.2 的 graphops 访问器推广到位）。空区右键菜单在
+  panels/timeline.rs:1268（`empty_area_menu`，知道轨道+帧）。
+
+## 3. 设计 A：调整图层
+
+### 3.1 节点模型
+
+- 新行为 `AdjustmentBlockBehavior { core: BlockCore }`（oak-node/src/block.rs）+
+  构造 `adjustment_create()`：只声明一个可连接、不可打关键帧的 `tex_in` 纹理输入，
+  `core.effect_input = "tex_in"`；不需要 footage 链接与 media/speed/reverse 输入。
+  在 `nodes/mod.rs` 注册表注册（type id `org.olivevideoeditor.Olive.adjustment`）。
+- 范围 = `core.range`（与其他块完全同构）——**天然满足"针对 clip 的一段"**：
+  范围想拉多长拉多长，可跨多个 clip，可只盖一个 clip 的中段。
+- 效果链：与 clip 完全同构（effectchain 不变）。检查器选中调整块时，
+  `selected_clip_node` 等 graphops 访问器推广到"任何带 BlockCore 的块"
+  （详见 §3.4），效果栈/参数 UI 原样工作。
+- 序列化：`save_custom`/`load_custom` 走 block.rs:165-186 的扩展机制
+  （旧读者跳过新元素，docs/project-storage.md 已为此预留）；OTIO/FCPXML 导出
+  第一版跳过调整块（记 TODO；OTIO 无调整图层概念，需要时映射为带效果的 Gap）。
+
+### 3.2 渲染：图路径（render_graph_frame）
+
+在 eval.rs 的轨道循环（1451-1478）中：
+
+```
+对每条视频轨（顶层→底层顺序）:
+    若该轨在 time 有启用的调整块:
+        1. 把已收集的下层 frames 先经 composite_tracks 合成出"下层合成图"
+        2. 用 §3.3 的机制对下层合成图跑调整块的效果链 → 一张纹理
+        3. 把该纹理作为唯一一帧压入 frames，继续向上收集
+    否则: 照常收集该轨覆盖 time 的 clip
+最后: composite_tracks(frames)
+```
+
+- 多个调整图层自然嵌套（每条含调整块的轨道触发一次上述 flush）。
+- 中间合成用现有 GPU 合成 helper（composite_tracks 的 GPU 变体），避免回读；
+  调整链的输出再作为后续合成的输入纹理。
+- 调整块自己的 `value()`：直通 `tex_in`（与 clip 的直通同形）；
+  真正喂纹理由渲染驱动完成（§3.3）。
+
+### 3.3 渲染：给效果链喂"下层合成图"（纹理源）
+
+新增轻量"纹理源"节点（oak-node，`org.olivevideoeditor.Olive.composite_source`）：
+`value()` 把渲染驱动预置的一张纹理推进输出表（形状仿 footage 节点的
+FootageJobPayload，但零开销直接装箱 `NodeValue::Texture`）。渲染驱动在每次
+flush 时：
+
+1. 图内临时建（或复用一个每帧更新的）纹理源节点，预置下层合成图；
+2. 把它连到调整链首效果的 `tex_in`（临时边，求值后拆除，或每帧重连）；
+3. `traverser.evaluate(EvalRequest::new(调整块, time))` —— 效果链、嵌套
+   shader job、OFX 插件全部照常工作（无需任何特殊分支）。
+
+（调研的备选方案——在 hooks 里替换链首 payload 的 tex_in 参数——魔法且脆，
+不采用；montage 机制（§3.4）覆盖不了内置 shader 特效，也不够。）
+
+### 3.4 渲染：montage 路径（必须与图路径语义一致）
+
+- `VideoTicketParams` 增加
+  `adjustments: Vec<AdjustmentSpan { in_time, out_time, track_index, effects:
+  Vec<MontageEffect> }>`；`video_montage` 识别调整块并用现成的 `clip_effects()`
+  （renderops.rs:222-241，对任意宿主块可用）构建其效果快照。
+- `render_montage_frame_into` 在按轨道累积合成时，越过 `track_index` 边界就把
+  对应 span 的 effects 用现有 `apply_clip_effects/apply_montage_effect`
+  （eval.rs:1901-1999）应用到累积图上。
+- **已知局限（写进 TODO）**：montage 的效果执行目前只真支持 Opacity + OFX 插件，
+  内置 shader 特效会 pass-through（既有问题，非本计划新增）。图路径（worker
+  快照优先）始终正确；montage 兜底路径的保真度随 montage 效果执行器的
+  补齐而提升——把"montage 效果执行器支持内置 shader 特效"列为后续独立工作项，
+  不在本计划范围。
+
+### 3.5 创建/编辑/UI/撤销
+
+- 创建：时间轴空区右键菜单（panels/timeline.rs:1268 `empty_area_menu`）加
+  "添加调整图层"：以命中轨道+帧为锚，`adjustment_create()` 建节点 + 一条 undo 推
+  `TrackPlaceBlockCommand`（"Add Adjustment Layer"，默认 5 秒或贴近吸附）。
+  i18n 8 语言加 `timeline.context.add_adjustment_layer`。
+- 外观：RealClip 用区别于普通 clip 的颜色（引擎 `ClipData::color` 覆盖即可，
+  不动 gpui）；label 默认"调整图层"。
+- 交互：移动/修剪/分割/删除在 graphops 访问器推广后免费获得
+  （clip_ids/clip_range/clip_track/trim_command/move_clip_to_track_commands/
+  clip_block/delete_clip 全部改为认 BlockCore，而非只认 ClipBlockBehavior——
+  参照 nodeops::block_core_of/set_block_core 的多 downcast 模式）。
+- 撤销：放置一条 undo（§2.2 模板）；效果编辑走 effectchain 既有 undo；
+  删除用 delete_clip 同款（TrackReplaceBlockWithGapCommand + remove_node_command）。
+
+## 4. 设计 B：转场特效
+
+### 4.1 转场块一等化（基础件）
+
+转场块（TransitionBlockBehavior，§2.1）满足"开始和结束点不覆盖整个 clip"——
+它自身就是时间轴上一段独立范围的块，骑在两个 clip 的接缝上（in_offset 伸入
+前块尾部、out_offset 伸入后块头部）。
+
+1. **创建**：两种方式都实现——
+   a. 菜单/快捷键：`DefaultTransition`(Ctrl+Shift+D) 对选中 clip 的接缝插入
+      默认转场（交叉叠化，默认 1 秒，可配置）；clip 右键菜单加"添加转场"。
+      创建 = transition_create() + 连线前后块 + 设置 offsets + 放置，一条 undo
+      （连线模式照 loadotio.rs:307-333）。
+   b.（后续）拖 clip 边缘产生重叠自动建转场——**不做**，本计划只做显式创建。
+2. **渲染（图路径）**：`render_graph_frame` 在逐 clip 求值阶段识别转场块覆盖
+   的 time：对转场区间，同时求值前块与后块两张纹理，按转场类型混合后作为
+   该区间输出（两输入混合的 shader 模式参照 merge.rs base/blend；混合 progress
+   = (time − transition_in) / length）。**`TransitionBlockBehavior::value()`
+   保持不被求值（渲染在合成驱动层完成）**，或按调研备选：把 value() 做成双输入
+   shader job 的发出者（前后块纹理经 out_block_in/in_block_in 求值后注入）——
+   实现时在两者中选改动更小者（倾向合成驱动层：与调整图层 flush 同层、共享
+   中间合成设施）。
+3. **渲染（montage 路径）**：montage clip 列表按时间天然把前后 clip 都列出；
+   在 `render_montage_frame_into` 对转场区间解码两张并按类型混合
+   （与 §3.4 同一累积器阶段）。
+4. **转场类型（本批 4 种，均为内置 shader 节点实现，见 §4.2）**：
+   交叉叠化（mix 渐变）、淡入淡出（对黑/对白，单边 dissolve）、
+   划像（wipe，左/右/上/下方向+柔边）、推移（slide，推出/推入方向）。
+5. **UI**：`RealClip` 喂 `ClipData::in_transition/out_transition`
+   （查块的前后 TransitionBlockBehavior 邻居，换算楔形长度）；楔形拖拽
+   `TransitionChanged` 在 real.rs:4683 落地（undoable 改 offsets，
+   复用 transition_set_offsets_and_length 的 undo 模式）。
+   转场块自身的移动/删除随 §3.5 的 graphops 推广免费获得；
+   `TransitionRemoveCommand` 的 edge/offset 恢复缺陷（undogeneral.rs:580-583）
+   顺手修掉（删除时记录并在 undo 恢复 offsets 与连线）。
+
+### 4.2 转场 = 调整图层上的节点（用户提议的形态）
+
+**输入问题的解法**：转场节点需要"另一幅画面"作第二输入。落地设计——
+
+1. 转场节点（内置，`org.olivevideoeditor.Olive.transition.crossdissolve` 等，
+   类别 `Category::Transition`）声明两个纹理输入：`tex_in`（链输入）与
+   `blend_in`（第二输入，merge.rs 双输入绑定模式）。
+2. 叠加关系：调整图层盖在接缝上（范围=转场区间），其下层合成图在区间内已经
+   包含后块（底层轨上有后块时）——节点混合的是：
+   - `tex_in` = 下层合成图（当前时刻）；
+   - `blend_in` = 由图连接指定的参考源（默认：接缝前块的输出——创建转场时
+     自动连好，节点编辑器里可见可改）。
+   实际语义 = "下层画面 ⇄ 前块画面按 progress 混合"，足以表达交叉叠化/
+   划像/推移；淡入淡出不需要 blend_in（直接对黑场混合）。
+3. **progress 自动喂值**：渲染驱动在跑调整链时知道（layer_in, layer_out, time），
+   把 `progress_in = (time − layer_in)/length` 像 `resolution_in` 一样自动填充
+   （在 §3.3 的纹理源节点上顺带预置，或 hooks 增加 `layer_progress` 字段由
+   process_shader_job 注入——实现时选改动小者，倾向后者：hooks 加
+   `pub layer_progress: Option<f64>`，shader 声明 `uniform float progress_in`
+   且 job 未显式提供时自动填入）。
+4. 这样用户获得两种用法：
+   - 简单接缝转场 → §4.1 转场块（一键、楔形可见）；
+   - 自定义/复合转场 → 调整图层 + 转场节点（区间即调整块范围，
+     "开始/结束点不覆盖整个 clip"由调整块 range 直接保证，节点编辑器里还能
+     级联别的效果）。**两种用法渲染同一套 shader。**
+
+### 4.3 验收语义（写进测试）
+
+- 交叉叠化中点：两纯色输入输出各 50% 混合（GPU 像素测试）。
+- 转场区间外：画面与无转场完全一致（前后各采一帧断言）。
+- 调整图层跨 clip：下层两个不同色 clip，盖一个 Opacity=0.5 的调整层，
+  区间两端的帧分别断言；区间外一帧断言不受影响。
+- 淡入淡出首尾：progress=0 全黑/全白，progress=1 原图。
+
+## 5. 工作项（可分配给子代理的最小单元）
+
+1. **W1 调整块模型+graphops 推广**：AdjustmentBlockBehavior/adjustment_create/
+   注册/序列化；graphops 与 real.rs 的 clip 访问器推广到 BlockCore；
+   selected_clip_node 命中调整块。单测：创建/修剪/删除 undo。
+2. **W2 调整图层渲染（图路径）**：eval.rs 轨道循环 flush + 纹理源节点 +
+   hooks.layer_progress；GPU 测试（§4.3 的跨 clip 用例）。
+3. **W3 调整图层渲染（montage 路径）**：VideoTicketParams.adjustments +
+   video_montage 构建 + render_montage_frame_into 应用 + 测试。
+4. **W4 调整图层 UI**：空区菜单创建、外观、i18n、检查器选中、app 层测试
+   （mock+real）。
+5. **W5 转场块一等化**：创建（DefaultTransition+右键菜单）、图路径+montage
+   渲染（4 类型 shader）、楔形喂数据+TransitionChanged 落地、
+   TransitionRemoveCommand 缺陷修复、GPU/集成测试。
+6. **W6 转场节点（调整图层形态）**：transition.* 节点族（双输入+progress_in），
+   自动连线（创建时把 blend_in 接到接缝前块）、节点编辑器可见性检查、
+   GPU 测试（§4.3 中点/边界断言）。
+
+依赖：W1→W2/W3→W4；W5 与 W1-W4 并行可做（只共享 §3.5 的 graphops 推广，
+排在 W1 后）；W6 依赖 W2（layer_progress）与 W5（shader 共享）。
+建议 W1+W2 一个子代理、W3 一个、W4 一个、W5 一个、W6 一个，主代理统一审查。
+**冲突根禁令**：eval.rs 的 render_graph_frame 轨道循环只能由 W2 所有者改，
+W3/W5 需要同区改动时通过评审串行合入。
+
+## 6. 验收标准
+
+1. 空区右键能加调整图层；上面加 Blur/Opacity 后，其下方所有视频轨在层覆盖
+   范围内的画面整体受影响；范围外不受影响；范围可只盖一个 clip 的一段。
+2. 选中接缝 Ctrl+Shift+D 插入 1 秒交叉叠化；播放/暂停/导出三种路径画面一致；
+   楔形可见且可拖改长度；转场不覆盖整个 clip（区间外画面逐像素一致）。
+3. 调整图层上加交叉叠化节点，blend_in 接到前块后呈现同样叠化；
+   progress 沿区间线性推进。
+4. 撤销/重做覆盖全部创建/编辑/删除路径。
+5. `cargo test --workspace` 全绿；新增 GPU 测试在无 GPU 环境跳过。