* Start adding a corner pin node
* Start gizmo support
* Correctly setup QVector2d inputs
* Missing breaks were causing incorect data to be set
* Add ability to override vertex coords
* Get perspective pin working
* Cleanup
* Add none perspective corner pin
* Small optimisation
* Improve comments, clean up and fix bug in job optimisation
* [skip ci] cornerpin: Improve description
* Move GL extension from shader to standard preamble.
* Use #version 130 for mobile GPU compatibility
* Cache half resolution
* Use built in NodeInputDragger::GetStartValue
* Pass cnst references to avoid deep copies
Co-authored-by: itsmattkc <itsmattkc@gmail.com>
If you run into new OpenGL issues, this is likely the commit that caused it.
I've tested this on as many systems as I have access to (including AMD, Apple,
Intel, and NVIDIA GPUs with a mixture of Windows, Linux, and macOS) and it
appears to work fine, however we all know how unpredictable GPUs can be.
This should help increase GPU performance and compatibility.
Squashed commit of the following:
commit fa3c0ac7c469aebe2941a803736bdc06cdd5a2d8
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 13:17:44 2021 +1000
viewer: updated pixel sampling for shared contexts
commit 9334261da235f09cb67b193c5bb6169abd7a95f7
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 13:12:39 2021 +1000
pixelsampler: show color preview box
commit dacd9be6971c5c0312ae45c8db349d47910a80dd
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 12:43:32 2021 +1000
renderer: interlace on GPU
commit 074cea06cb87e9fdc4e8ea6b09587be6df296057
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 12:43:14 2021 +1000
renderer: added texture clear option
commit 5d4f86b4cf1b1dcd3fb8e8c854ffa7f6fb8ca346
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 12:42:12 2021 +1000
renderer: updated scopes for shared textures
commit 6fad96358d189874cf45f7887b7e13cc2b1603c4
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 11:42:05 2021 +1000
opengl: fixed edge case
commit 1a19ffb53703ac6f30ed210f6d8eb5fc4b7b0f2d
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 10:56:13 2021 +1000
opengl: fix deinterlace shader
commit 085961aa39cca90ebf51b919a78ef07a4882f0e1
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 10:08:40 2021 +1000
opengl: downgrade to ES 2.0
With half-float/float/3D extensions, this should be doable
commit ea1ab45cbcf5e017dfd0d8f28b231ff3255dfe92
Author: itsmattkc <itsmattkc@gmail.com>
Date: Thu May 27 10:06:59 2021 +1000
renderer: fixed mutex deadlock
commit ecc2f08b16f03421870c5b61625ecd26f2f96582
Author: itsmattkc <itsmattkc@gmail.com>
Date: Wed May 26 21:28:13 2021 +1000
opengl: fixed shared contexts
commit 261c689ea8089463155cf11b9ab15aa3f879f439
Merge: bfd4338a 91c88b1c
Author: itsmattkc <itsmattkc@gmail.com>
Date: Wed May 26 19:52:00 2021 +1000
Merge branch 'master' into shared-contexts
commit bfd4338aaa3b09b6fbb7717c637face8d5490721
Author: itsmattkc <itsmattkc@gmail.com>
Date: Wed May 26 18:52:38 2021 +1000
viewer: remove debug line
commit 45868d8a396181bf81c24b92a1ac5ce9ffed3426
Merge: 642f8e9d d0bd5e21
Author: itsmattkc <itsmattkc@gmail.com>
Date: Wed May 26 17:21:48 2021 +1000
Merge branch 'master' into shared-contexts
commit 642f8e9df7d7d1a9ed4202d428f4d968ff668552
Merge: 861e5b89 4de0a727
Author: itsmattkc <itsmattkc@gmail.com>
Date: Wed May 26 17:01:38 2021 +1000
Merge branch 'master' of https://github.com/olive-editor/olive into shared-contexts
commit 861e5b89a38c281897dc44b20e3587d39c4f808c
Merge: 472598a7 ef31c3fb
Author: itsmattkc <itsmattkc@gmail.com>
Date: Wed May 26 14:28:21 2021 +1000
Merge branch 'master' into shared-contexts
commit 472598a740a095aa43a4aad5680dd5abcf453a4c
Author: itsmattkc <itsmattkc@gmail.com>
Date: Fri May 14 16:29:35 2021 +1000
attempt at sharing contexts
Since sequence resolution is provided to the nodes now, it no longer needs to be a global in the shaders
Removed some unused shaders
Added default params to ShaderCode to allow skipping types that aren't used
- Abstracts all OpenGL functionality including UI objects
- Cleans up several rendering codepaths
- Improves threading functionality
- Removes old code and problematic functions
General code and functionality improvements. Moved more code out of the
OpenGL backend for portability. Implemented audio transitions.
Implemented basic transition animation curve settings.
The nodes now have more control over how their accelerated shaders/sample
functions are run, as well as how items are popped off the value tables.
This allows for various optimizations that we didn't have access to before.
Fixes:
* Values exceeding the encoded range.
Future may account for a certain range
above and below current values.
* Minor calculation problem that prevented
the above fix from working properly.
* Streamlined the shader code.
* Add a vertex shader to handle the scaling.
* Use the viewport resolution as the scaled
version to save performance.
Basic pass at histogram. Uses a power based
scale compared against the total pixel count.
Probably needs a UI toggle to flip lines on
and off as the text and lines in OpenGL are
a large cycles sucker. Could be likely much
faster when migrated to a geometry shader
approach. Uses the scaled viewport resolution
to increase performance as much as possible.
Remove the sum, add some commments, and
make it scale to the full window.
TODO:
* Hook up CM.
* Add UI for intensity
* Add lines at 20% increments, plus under /
overshoot?
* Add UI for increments.
Rather than plugging a matrix into the video input node, the matrix is now
multiplied by the video input using a math node. This is probably more
sensible from a user perspective.
This also means the renderer is tolerant of texture sizes that are not equal
to the sequence size, however most nodes will downsample the texture to the
sequence size (and if not, it will be downsampled once it is cached). Textures
will still *always* be in reference space and the sequence's format. This seems
like the best compromise between backend and frontend congruity.
Results are identical (tested with Olive 0.1.x). Using a straight multiply is faster if the compiler doesn't optimize pow(). It's also safer because with pow(), [the result is undefined if x<0 or if x=0 and y≤0](https://www.khronos.org/registry/OpenGL-Refpages/gl4/html/pow.xhtml).
Thomas Wilshaw