- open_hw_accel marks the device unavailable when the decoder OPEN fails
(cuvidCreateDecoder OOM at 4K) too, not just device-context creation:
without it every subsequent decoder session retried CUDA and flooded
the log per open.
- Decoder gains hardware_decoding(); the oak-render decode-session LRU
evicts hardware sessions first (each pins a GPU surface pool — ~100 MB
at 4K), so a full cache cannot exhaust video memory before the next
open.
- Worker pool count now factors GPU vram: per-worker budget = 1 GiB
(1080p peak) scaled by pixel ratio + 256 MiB idle floor, 10% reserve
of free vram; applied when hardware decoding is on (nvidia-smi query,
None otherwise falls back to the RAM/CPU policy).
- Dynamic pool resize: ProcessDispatcher::set_target_workers grows or
retires workers; retiring ones stop claiming, drain their in-flight
batch (future playback frames included), then exit naturally on the
shutdown signal — no mid-work kill (30 s deadline only as a hung-
decoder last resort). A retiring worker that dies re-queues its frames
to surviving workers. Resizes are throttled to 2 s (a resolution burst
merges; only the latest target applies) so 1080p<->4K flaps cannot
thrash process spawns.
- RenderManager::set_workspace_size announces the sequence resolution;
RealEngine calls it from refresh_sequence_info.
- Integration test: shrink 3->1 mid-wave (all frames complete, retired
workers exit naturally) then regrow 1->3 and render a fresh wave.
Preview now follows the project output colorspace end to end: the
display chain derives its content space from the project's OutputColorSpec
instead of a hardcoded sRGB name, self-managed ICC transforms go through
an XYZ D65 interchange stage (OCIO cie_xyz_d65_interchange) for non-sRGB
targets, and the platform layer declares the content colorspace (gpui
submodule bump). macOS defaults to OS-managed (fixes wide-gamut UI
oversaturation); Windows ACM warns once on non-sRGB targets.
Multi-monitor: the display ICC is looked up per the window's current
screen (macOS display id, Windows per-monitor DC, X11 RandR output
profile) with a throttled poll that invalidates frame caches on moves.
Pipeline precision: 10-bit+ sources fall back to YUV444P16LE + a Rust
matrix conversion when swscale lacks F32 output (no more 8-bit
truncation); BT.709/2020 SDR decodes with BT.1886 gamma 2.4 instead of
the sRGB EOTF; working-space compositing no longer clamps RGB to [0,1]
(alpha still clamped); the output node clamps to the target gamut;
frames without colorimetry metadata convert with BT.709 defaults
(warned once) instead of passing through; scopes read the
output-colorspace signal on both F32 paths.
Also: only emit rerun-if-changed for .env when it exists (a missing file
made every build fully dirty).
Adding an effect to a clip did nothing: the sequence render is
flattened into a montage (decode + composite), and MontageClip carried
no effect data at all.
- MontageClip gains an ordered effect stack (type id / enabled /
effect input / parameter values); protocol v2 carries it as an
additive wire field (older peers default to an empty stack).
- renderops::video_montage fills the stack from the effect chain
(the footage source node — the chain end without an effect input —
is dropped; the montage decodes the footage itself). Export
(oak-task) and the multicam single-track montage fill it too.
- The worker applies the stack between decode and composite: built-in
Opacity gets a CPU evaluator (C++ opacity.frag parity — whole vec4,
alpha included, unity pass-through); everything else dispatches as an
OFX plugin job through a new instance-factory slot (oak-plugin
lazily creates + caches one instance per identifier per render
process) with the montage's parameters injected. Disabled effects
bypass (the C++ traverser pushes the effect input through). Unknown
types warn once per type id and pass through — no silent no-ops.
Not covered (explicitly): Transform/Crop and the other ~30 built-in
effects have no CPU evaluator in oak-render (they pass through with a
warning), keyframed parameter animation, audio effect chains, and the
CLI's simplified montage.
Acceptance: a real 50% Opacity on real media quarters the rendered
pixels both in-process (renderops test) and through a real worker
process over IPC + shared memory (procpool_integration test);
disabling restores the plain render byte-for-byte.
convert_bgra8 applied packed u8 pixels to the default (F32-finalized)
OCIO CPU processor, which rejects them with a bit-depth mismatch; the
callers swallow the error, so the display-ICC transform was silently
inert on the shm preview path. ocio-rs exposes no Uint8-finalized CPU
processor, so the conversion now detours through F32. Verified against
the machine's actual display profile with the new
display_icc_bgra8_never_outputs_black test (OAK_DISPLAY_ICC).
Also:
- procpool_integration: audio tickets are Seek priority and claimable
by any worker now, so the shard-spread assertion goes (rendering on a
live worker is what matters).
- OAK_DEBUG_VIEWER=1: the program viewer logs frame pushes and dumps
the displayed frame to /tmp/oak_viewer_frame.ppm (the black-screen
investigation tooling).
ci (Windows): the runner's msys2 shell starts as the base MSYS
environment (MSYSTEM=MSYS), which install-deps.sh rejects. Set the
job-level MSYSTEM=UCRT64 env and prepend /ucrt64/bin to PATH in every
Windows step, so pacman installs and the toolchain resolve against the
mingw-w64-ucrt-x86_64 packages.
fixtures: the restructured workspace moved the media fixtures into
crates/oak-app/tests; the remaining references pointed at the old
repo-root tests/ — oak-codec realmedia_tests + hwdecode, oak-node
serializer golden, oak-worker procpool integration. All point at
../oak-app/tests now (oak-app's own tests/demo.mp4 references were
already correct after the move).
All crates take the oak-* kebab-case naming (oak-audio, oak-codec,
oak-common, oak-core, oak-ffmpeg-link, oak-node, oak-otio, oak-plugin,
oak-render, oak-storage, oak-task, oak-timeline, oak-undo), with the
lib identifiers rewritten (oakrender:: -> oak_render::, oakcore_rs:: ->
oak_core::, ...) across all 226 referencing files.
The GUI application moves from the workspace root into
crates/oak-app/: src/, build.rs (paths fixed for the new location) and
tests/ travel with it, the root Cargo.toml becomes workspace-only
([workspace] + workspace.package + profiles), and the app package
inherits the workspace version. The screenshots example becomes a
standalone crate examples/simple_player/ with its own Cargo.toml.
Every crate now inherits the single workspace version
(version.workspace = true), and the workflows' crate paths and the
build docs follow the renames.
Validated with a clean cargo check --workspace.
A claim mixing audio and video tickets is delivered as the video
message first and the audio message second, and the worker pops one
free-ring slot per ticket in that message order, checking each pop
against the assignment. The dispatcher however assigned slots in the
scheduler's interleaved frame order, so every audio ticket inside a
mixed batch mismatched, and each mismatch consumed a worker slot
without recycling it — cascading into the 'slot assignment mismatch'
flood and failed frames during playback.
Slot assignment now partitions the claim: video tickets first, then
audio. The mixed_audio_video integration test forces mixed claims
(queue depth > slot count with immediate releases) and fails with the
exact production signature when the fix is reverted.
- Audio tickets join the process backend: render_audio_batch wire
message, workers mix straight into shm slots (SLOT_FORMAT_AUDIO_F32),
ShmAudio payload with release semantics, crash isolation covers audio
renders; playback audio uses an async 4-chunk prefetch drained on the
UI tick (also fixes the sub-60fps chunk truncation bug); oversized
ranges and dispatcher outages fall back to in-process inline.
- Per-ticket slot formats: force_format is honored (exports request
F32 slots, dropping the BGRA8 round-trip and its 8-bit quantization);
segments grow on demand via worker-idle rebuild with generation
handoff; the scheduler filters over-capacity tickets.
- Adaptive defaults: 128-256MB/worker segment budgets drive slots per
worker, batch size follows workers/slots; bench_process example
measures throughput and adjacent-frame completion deltas
(e.g. 4 workers: 841 fps, 4.6ms mean delta).
- WorkerPool thread pool deleted; RenderManager defaults to the
Processes backend (oak-worker children), Threads kept as a test-only
inline dispatcher; audio tickets stay in-process until S3.
- Onscreen path reads worker shm slots directly: BGRA8 slot format,
RenderedFrame::Shm wrapped into the display buffer (single disclosed
GPU-staging memcpy), scopes analyze BGRA8; the long-lived full-res /
thumbnail paths take the counted slot_to_vec copy and release.
- Playback pre-render window: forward 120 frames (configurable) fed to
the PreviewScheduler at Playback priority, interleaved across
workers, cached in shm slots until the playhead consumes them;
generation-based invalidation cancels and releases on edits.
- oaktask export and oak-cli run on private ProcessDispatchers (fixed
a pump-while-locked self-deadlock in the export loop); facade
get_frame handles ShmFrame payloads.
- Acceptance: preview path main_heap_frame_copies == 0 with spawned
workers, CLI transcode/render verified end to end.
Mike-Solar