* EXR export: write the output space's chromaticities attribute (#3037)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* EXR chromaticities: tests for unmanaged and PQ ProEXR, every space once
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* EXR export: note why the comp background is added before decoding (review point)
CodeRabbit asked to convert Comp.background to the output space before
compositing. Every output path (Image::to_rgba8_over) adds the background to
the output-encoded frame as it is, so the EXR now matches the other outputs;
converting would make the EXR colour differ from the PNG of the same comp.
Comment only.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* Format the merged code with rustfmt
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
* Plug-ins: API 2 (PluginHost, padding, layer_span)
Split out of the OpenFX work so it can be reviewed on its own. Additive; the WebAssembly ABI
stays at 1.
- point3 / layer / text / hidden parameters, integer sliders, parameter groups.
- EffectPlugin::padding (transparent pixels around the buffer, at most 4096 per side).
- EffectPlugin::render_v2 with a PluginHost: the input at other times, other layers, parameters
at other times, frame rate, per-instance key, the layer's source range.
- EffectEnv::layer_span from the renderer; MAX_PLUGINS 1024.
- plugin_error: a plug-in effect's last render error, for the UI.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KxacwTpUwqyfpcLgKW8yjS
* Plug-ins: external effect helpers (crates/helper-host)
Native plug-in formats run in separate helper programs, never in EffectCraft's process
(docs/adr/0001-external-effect-helpers.md). New safe-Rust crate effectcraft-helper-host (L3,
native only, unsafe forbidden): it finds helper programs, starts each as a child process and
registers its effects through plug-in API 2, so the UI and command surface need nothing new.
- Protocol v1 (docs/external-helpers.md): JSON lines over stdin / stdout (helper replies are
prefixed so plug-in chatter is ignored); pixels go through a frame exchange file (/dev/shm on
Linux), never through the pipe. Render-time callbacks for the input at other times, other layers
and parameters at other times, with nested renders through the same helper.
- A helper that exits, hangs past its timeout or breaks the protocol is stopped. The render fails
with a message, the frame is left as it was, and the helper restarts for the next render. An
effect that stops its helper 3 times is turned off (helper_blocklist.txt).
- Opt-in: nothing starts helpers unless the user asks (Effect > Scan for Effect Helpers), or opts
in to starting them at launch with EFFECTCRAFT_HELPERS=1, an environment variable, so writing
to the settings folder alone can't run code. effect.helpers.scan / load are refused for
confined agent sessions (confine.rs), like file.executeFile.
- Discovery: helpers/ next to the app (Contents/Helpers on macOS), helpers/ in the settings
folder, EFFECTCRAFT_HELPER_PATH.
- Commands effect.helpers.list / scan / load; Effect menu entries (translated); the last render
error of an effect shows as a warning sign after its name in Effect Controls.
- serve(): the helper side for Rust helpers. tests/it/main.rs runs the whole protocol (render,
padding, callbacks, nesting, threads, crash, restart, block list, hang) against a fake helper,
with nothing to install.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KxacwTpUwqyfpcLgKW8yjS
* Effect helpers: address review (start failures, frame file mode, nextest listing)
- A helper that can't start (not installed any more, restarting too often, no frame exchange
file) is reported but no longer charged to the effect being rendered: it is Failure::Refused,
not Failure::Broken, so it never counts toward turning that effect off. New check in
tests/it: once the helper is out of restarts, renders of an effect fail without blocking it
(fails without the fix: the effect was turned off after 3 such renders).
- The frame exchange file is created owner-only (0o600) on Unix: it holds the user's frames and
/dev/shm is shared. Asserted in the frames unit test.
- tests/it (harness = false) speaks the libtest listing protocol cargo-nextest uses:
`--list --format terse` prints `helper_protocol: test` (nothing with `--ignored`), and
`helper_protocol --exact` runs the checks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KxacwTpUwqyfpcLgKW8yjS
* Effect helpers: bound writes to a helper, drain stderr, keep callbacks in step
- Writes to a helper go through a writer thread with the request timeout: a
helper that stops reading its stdin (hung with a large callback answer or
render request pending) no longer blocks the render thread forever while
holding the helper.
- stderr is drained as bytes (lossy, in 64 KiB pieces) to the end: output that
isn't UTF-8 stopped the reader, closing the pipe under the helper (EPIPE /
SIGPIPE on its next write); lines were unbounded.
- Callback images that can't be written are answered with `null` instead of
failing the render while the helper waits (which left the protocol out of
step); a frame that can't be written before the request is sent is not
charged to the effect.
- Mid-render calls use the render's own process: no respawn or heartbeat ping
while the helper is waiting for an answer.
- Waiting for a busy helper is bounded (load timeout) instead of forever;
status() no longer blocks on a running render.
- Reap the child when start-up fails after spawning; cap the block list read.
- Scripts start helpers only with Allow Scripts to Write Files and Access
Network on (they run native code), as for File.execute().
- helper-host declares the workspace's no-panic crate lints; the desktop
main() keeps its doc comment.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Effect helpers: rustfmt the fake provider, build the test session in its initializer (clippy)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Plug-in Library: offer WebAssembly API 1 releases only; a helper already stopped mid-render isn't charged again
- WasmPlugin loads only modules at the WebAssembly ABI version, which stays 1 while
PLUGIN_API_VERSION is 2: share it as effects::plugin::WASM_API_VERSION and have the
library's compatibility check use it, so it never installs an API 2 release that
can't load.
- with_running: a process that is already gone was stopped by what broke it (a nested
render, charged there), so this render fails without charging its own effect.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* frames: plug-in effects count as reading other frames (plug-in API 2)
With main's narrowed frame invalidation (#2795), an edit inside a precomp under an
effect that reads its layer at other times must reach the layer's whole span.
Plug-in API 2 gives every plug-in effect the input at other times, so any effect
outside ec. counts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
* frames: keep cached frames an edit can't reach
Nudging, trimming or splitting a layer re-rendered the whole comp. The RAM
preview now carries every frame whose stack of active layers is unchanged
to the new comp identity; only the segments an edit touches re-render.
* disk cache: never drop frame writes; save frames kept after edits
Frames get their own write lane (bounded by bytes, written before layer
buffers), so a burst of layer writes no longer drops frames and leaves gaps
in the blue bar after reopening. Frames an edit kept in RAM are written to
disk under the comp's new content key when idle.
* frames: music and hidden layers keep the caches
Layers that can't change pixels (audio-only footage, AV layers with Video
off, when nothing reads them and there are no expressions) no longer count
for the RAM preview's dirty segments or the disk cache's comp and footage
keys. Adding, trimming or moving music keeps both caches.
RENDER_VERSION bumps to effectcraft-render-3 for the new key formula.
* ui-egui: background caching benchmark
cache_bench fills a comp's RAM preview the way idle caching does and
prints frames/s with decoder, footage-cache and disk-cache counters.
* frames: soloed music and hidden layers aren't silent
Solo hides every other layer, so soloing a music track or a layer with
Video off changes the picture. silent_layers no longer drops soloed
layers, so the RAM preview and the disk cache see Solo toggles.
* disk cache: audio files a layer reads stay in the layer salt
Audio-only footage left the disk layer salt so importing music keeps the
keys, but Audio Spectrum and Audio Waveform draw from an audio layer.
Footage another layer reads (or any, when a read layer is a precomp)
counts again, so overwriting that file doesn't reuse stale buffers.
* disk cache: retry frame writes the queue refused
put_frame reports whether the frame is on disk or queued. persist_missing
forgets the key of a refused write, so a full frame lane no longer leaves
RAM frames that are never written.
* ui-egui: cache_bench measures idle caching; checks --comp and --slots
Frames go through request_background with one slot per frame thread, as
Cache Frames When Idle does. An unknown --comp is an error (the active
comp is used only without --comp), --slots is at least 1, and the disk
report flushes queued writes instead of sleeping.
* disk cache: RENDER_VERSION effectcraft-render-6
This branch changes the cache-key formula (silent layers, Solo, read
audio in the footage salt), and upstream #2759 already took render-5
for the clip-clock change, so entries from either must not be reused.
* frames: rewrite RAM frames the disk cache evicted
persist_missing kept every disk key it ever queued, so a frame the disk
cache evicted later was never written back. The disk cache now tracks
frame writes still queued or in progress (frame_pending), and
persist_missing only remembers keys while it converts them.
* frames: idle passes continue while RAM frames wait for disk
persist_missing reports whether there may be more to write (a content key
being computed, frames being converted). Cache Frames When Idle then
schedules another pass, so a refused write is retried without waiting for
a UI event, and the passes stop once every frame is on disk or queued.
* frames: rewrite an evicted frame once, not on every idle pass
A disk cache smaller than the RAM preview, or a frame too large for it,
made idle passes write, evict and rewrite the same frames every 500 ms.
Each disk key is now written at most twice under one disk limit (the
write and one rewrite after an eviction); a new limit resets the count.
The persist test waits for the conversion to forget its keys before
checking that no work is left.
* frames: the work area doesn't change pixels
Setting the work area (B, N) cleared every cached frame, in RAM and on
disk: it was compared with the comp's pixel settings and hashed into the
disk key. It is a preview and render range only, like the Resolution, so
it no longer counts.
* frames: audio edits keep frames unless something draws the audio
Changing a clip's Audio Levels, its audio effects or its Audio switch
cleared the clip's cached frames, and the comp's disk cache, although
nothing drew that audio. The audio now counts only for the layers whose
audio is heard (Project::heard_layers): those a layer picks in an effect
(an Audio Waveform or Spectrum, even on itself), the layers of a picked
precomp, and every layer while an expression (which can read Audio
Levels) exists. Music in a precomp that a waveform in the outer comp
picks is no longer treated as silent either.
* frames: an expression only reaches the frames of its own layer
One expression anywhere in the project made every edit clear the
comp's whole cache, and so did editing a layer another layer reads
(its parent, matte, or a layer it picks). Expressions can read
anything, but they only draw in their own layer: an edit now re-renders
the frames where it changed something, plus those of the layers with an
expression and of the layers that read a reached layer. With
expressions, an edit elsewhere (another comp, an item's name, a marker,
music) reaches only the expression layers' frames.
* frames: an edit inside a precomp re-renders only where the precomp shows it
Editing anything inside a precomp cleared every cached frame of the
comps that nest it, even a music level in it. Project::dirty_spans now
follows precomps: it takes the spans their comp changed in, shifts and
scales them by the precomp layer's start and stretch (its whole span
under a time remap or Responsive Design timing), and keeps the outer
comp's other frames. Layers that read a changed precomp (a matte, a
waveform) re-render where they are active.
* disk cache: keep the frames an edit can't reach
The disk cache keys frames on a hash of the whole comp, so every edit
lost the frames that were only on disk (the blue bar), even a rename;
only those still in RAM were written again under the new key. When an
edit keeps RAM frames, it now also queues a disk carry: on the frame
pool, each kept frame of every disk series in use for that comp is hard
linked (or copied) to its new key with DiskCache::link_frame, oldest
edit first. The old entries stay, so undo finds them too.
* frames: a keyframe edit re-renders only around the keys it changed
Changing one keyframe (its value, interpolation or ease, in the
Timeline or the Graph Editor) re-rendered the layer's whole length, so
a key on a full-length layer cleared the whole cache. A layer that
differs only in keyframes of its Transform properties now re-renders
from two keys before the first changed key to two after the last
(auto-Bezier and continuous tangents reach the neighbouring segments),
to its in or out point past the first or last key. Static values,
expressions, roving keys, time remap and time effects still reach the
whole layer; effect keyframes do too for now (simulations and time
effects read other frames).
* frames: music volume keeps frames while no expression reads audio
Any expression in the project (a wiggle on another layer, even a
disabled one) made every layer's audio count as read, so changing a
music layer's Audio Levels re-rendered every expression layer and
changed the comp's disk key. Audio now counts for expressions only when
one may read it (Heard::audio_exprs): its text names audio or walks
properties by index. An audio-only change nothing reads no longer counts
as a change elsewhere in dirty_spans either.
* frames: a keyframe edit with few keys or beside a self-contained expression keeps the other frames
Changing a Scale key of a layer with two or three keys re-rendered the
layer's whole length: with fewer than two keys before (or after) the
changed one, the span fell back to the in (or out) point instead of the
first (or last) key, past which the value holds. And any layer with an
expression re-rendered on every edit, so a wiggle on a full-length layer
cleared the whole cache. An expression layer is now reached by edits
elsewhere only when its text may read outside its layer (it names a
comp, layer, footage, the project or its source); its parent and the
layers it picks are still read through referenced_layers.
* viewer: the caching chip stays up between uncached frames, with a spinner
The "Caching frames…" chip showed only while playback waited for a
frame, so during caching it flickered on and off with every frame that
arrived. It now stays up 0.6 s after the last wait (and hides at once
when playback stops). It is also smaller and quieter: a spinner, then
"Caching frames" without the ellipsis in dimmed smaller text, on a
fainter chip sized to its text.
* preview: the Preview panel's cache bar counts the preview's Resolution
With a preview Resolution set (Half, Third…), the frames a preview
plays are cached at that scale, but the Preview panel's bar counted the
series the viewer shows. When playback stopped, the viewer went back to
the comp's Resolution and the bar dropped from full to the frames cached
there. It now counts the preview's own series (Auto still follows the
viewer). The viewer and the Timeline's cache bar are unchanged.
* Contributors: add nazar-gavrylyk's name
* frames: an expression reading its own layer's markers or audio re-renders that layer
With expressions in the project, a layer's markers count when comparing it, and its audio counts
when an expression may read audio, so a marker or volume edit re-renders the edited layer over
its own span. Comp markers still count only through layers whose expressions name thisComp.
* frames: a keyframe edit on a layer with an expression re-renders the layer
An expression on the keyed layer can read the keyed property at other times (Rotation reading
Position with valueAtTime), so the frames past the changed keys may change too.
* cache_bench: read decoder counters through effectcraft-media
Only crates/media, crates/export and crates/interchange may depend on the FilmCraft crates.
* cache_bench: clips end where the next one starts
* frames: write-back uses the worker's disk keys while a Roto Brush model is active
persist_missing keyed disk frames on FrameKey::opts, which also carries the Roto Brush model
stamp; the worker and disk_frames key on the render opts' hash alone. With a model active every
RAM frame looked missing, was written again under keys nobody reads, and recorded a second disk
series.
* frames: an edit two precomps down reaches the outer comp
A precomp layer whose comp Arc didn't change was skipped, but the comps nested in it may have
(an edit in Inner leaves Mid's Arc as it was), so Outer kept every frame. Every nested comp is
walked now, each once per edit.
* cache_bench: stop when the disk cache can't be cleared
* frames: one disk carry job at a time
carry_disk runs on every timeline paint and idle pass, and spawned a pool job whenever carries
were queued; during a drag the extra jobs waited on the carry lock and held frame-pool threads
from the viewer. A scheduled flag keeps one job, which drains the queue in order until it's
empty.
* frames: a panicking disk carry is skipped, not left holding the job
The carry job clears carry_scheduled only on its normal exit; a panic in a carry would have left
it set for the session, so no later carry ran. Each carry is caught like the content key's
hashing, so the job always reaches its exit.
* frames: an expression reading its own in point or index re-renders its layer
A trim or rename kept the layer's frames outside the cut when the layer
drew the same apart from its span, but an expression on it can read its
own inPoint, outPoint or name (a fade from inPoint), so every frame of
the layer changes. Such a layer is no longer treated as only respanned.
An expression reading `index` reads the layers above it too: a layer
added or removed there shifts it, so `index` now counts as reading
outside the layer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* frames: an edit of a comp reaches the expressions of the precomps it nests
An expression in a precomp can read the comp that nests it
(`comp("Main").layer("Controls")`), but the walk into nested comps only
counted changes outside the edited top comp. Editing a hidden Controls
layer there (or any layer, when the precomp is time-shifted or reads it
at other times) kept every precomp frame in RAM and linked them all on
disk under the new key. Nested comps now see an edit of the comp above
them as an edit elsewhere.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* frames: an edit in a precomp under a time effect reaches its whole layer
A precomp layer running Echo, Posterize Time, Time Displacement or
another effect that reads the layer at other times shows a nested change
far from where it maps, but only the mapped span (a nested frame either
side) re-rendered. Such a layer now gives its whole span, like a time
remap. `reads_other_frames` lists those effects (every `ec.time.` and
`ec.sim.` effect and the others that call `self_at`); keyframe spans use
it too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* frames: a keyframe edit beside muted keys re-renders the whole layer
Muted keys are skipped when a property is evaluated, so the keys that
play around an edited key can be any distance away, but the span counted
two stored keys either side. A segment from an earlier playing key kept
its stale frames. Muted keys now give the layer's whole span, like
roving keys.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* frames: one echo in the precomp time-effect test (three saturate alike), rustfmt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* gpu: run Content-Aware Fill's PatchMatch on the device
Content-Aware Fill's synthesis is the stage that makes the tool slow: at
1920×1080 with a 300×200 hole, PatchMatch costs ≈ 1.1 s per frame on the CPU,
and it is what Object, Surface and every refining frame spend their time in.
Add a `FillBackend` trait to `effectcraft-raster` whose stages (the membrane
solve, the flow fields, PatchMatch) default to the CPU reference, and
`Accelerator::inpaint()` on the compositor, which is how the engine asks for a
device backend. `effectcraft_gpu::inpaint` implements PatchMatch: the
multi-scale pyramid, the EM rounds, the propagation (a two-pass checkerboard
instead of the CPU's scan order), the random search, the SSD, the
histogram-based σ² and the weighted vote, as compute kernels in
`shaders/inpaint.wgsl`.
The fill's frames are read back to the CPU before it runs (analyses composite
on the CPU by design), so the backend is about the fill's own work; the
orchestration, the flow-guided propagation and the per-frame flow control stay
on the CPU. Only the hole's bounds cross back to the CPU, which is what makes
the offload pay.
Measured on an Intel Arc (MTL) iGPU, 1920×1080, 300×200 hole:
CPU 1087 ms GPU 397 ms (2.7×, 41.0 dB PSNR against the CPU's synthesis)
The membrane (harmonic) solve deliberately stays on the CPU: the same
measurement has it at ≈ 90 ms on the device against ≈ 55 ms on the CPU,
because a fill's full-frame data movement (upload, the known pyramid, a copy)
does not shrink with the hole, while the relaxation does. The device takes only
the stage where its parallel search pays; `membrane_tex` still runs on the
device inside PatchMatch, where it is the coarsest level's start and never
crosses back.
Closes#1839.
* gpu: fix the fill's vote histogram and run its flow on the device
Review follow-ups on the Content-Aware Fill GPU port, plus the fill's flow.
Fixes:
* imp_hist kept about one sample in 256: each of the 256 threads scanned a
strided subset of the pixels and retained only those whose bin selected it,
so the vote's sigma² came from a sample correlated with the pixel index (and
a small hole usually fell back to 1.0, making the weights near-uniform). One
workgroup now accumulates the whole population through shared atomic bins,
with a test that pins the sample count.
* imp_pm_prop and imp_pm_rand were dispatched over grow(near, 1) but with
near's origin, so the ring's left column and top row were never written
(NaN under test, stale pooled data in production) while a strip past it was.
`over` now derives the shader's origin from the box it sizes the dispatch
from, so a dispatch's region and its origin can no longer disagree.
* Gpu::inpaint() requires its kernels to have built. Otherwise every stage
paid an upload and a failed dispatch to fall back to the CPU, while the fill
still reported the GPU path.
* imp_rnd's XOR chain parenthesises its multiplicative terms: WGSL's grammar
makes each XOR operand a unary expression, so `a * b ^ c * d` is rejected
by spec-compliant parsers (naga accepts it).
* FillBackend documents that its stages may be called concurrently.
* The coarsest membrane's sweep count (64, against the CPU's 200) is now
justified: matching 200 measured ~6% slower over the fill with no change in
the result.
And the flow: Gpu::flow runs block_flow's coarse-to-fine search, sub-pixel fit
and vector median on the device. The luma is the CPU's (flow::luma_pair), so
the field matches block_flow's to a float rounding and the propagation sees
the same motion; the hole completion stays on the CPU with the membrane. A
four-frame 1920x1080 fill goes from 1.27x to 1.42x against the CPU, and the
device path is covered for holes from 16x16 to 300x200.
* gpu: keep the fill vote's histogram range on HDR input
imp_hist binned every distance at a fixed 256 bins per unit and clamped to
1023, so the vote's sigma2 topped out at 3.998. A 32-bpc frame's channels can
exceed 1, and masked_frame forwards them unclamped: a patch distance of 1600
then reported sigma2 ~ 4, exp(-1600/(2*sigma2)) rounded to zero in f32, and
imp_pm_vote kept the prior image instead of synthesising.
The histogram now reports the largest finite distance it saw (atomicMax on the
float bits, carried in bin 0's unused .g channel). SDR input keeps the fixed
0..4 range with identical bins; HDR input is re-binned over the observed range
and percentile() maps the bin centre back through it. Nothing clamps the input.
Also group over/sweeps/pm_level's arguments into small structs: the workspace
denies too_many_arguments and the lint policy (#2496) does not allow new
#[allow]s. No behaviour change on those paths.
* engine: don't report Edge Blend fills as device fills
Edge Blend calls only the membrane, which the GPU backend leaves to the CPU
reference, so run_plan's "Filling (GPU)" status was wrong for it whenever a
backend was present. Object and Surface still run the flow and PatchMatch on
the device and keep the label.
CodeRabbit's #1856 finding.
* gpu: run the fill's device stages inside error scopes
An allocation or validation failure inside patch_image/flow_image left an
invalid texture whose later use raised an uncaptured error, retiring the
shared compositor instead of falling back to the CPU for that frame. Wrap both
stages in Gpu::within_memory, whose scopes are native thread-local, so the
fill's rayon workers do not share them.
---------
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
* GPU: free effect working textures after each call instead of at frame end
A GPU frame keeps its encoder open while the CPU renderer runs layers with GPU
effects (Gpu::effects -> run_chain). Each call's Enc handed its textures back
to the shared pool, where they waited for every open encoder, so all effect
intermediates of a frame stayed allocated until the frame ended (17.7 GB by
layer 17 of a heavy 4K comp, then OOM). Effect calls now borrow a pool owner
(Enc::lent) that is given back once the call's work is submitted or
discarded; its kept textures are trimmed to the pool budget, so the rest are
freed after each call.
* GPU: one pool budget for idle effect calls, trimmed at the end of a render
Review of #1523: each idle lent owner kept up to POOL_BUDGET of textures of
its own, outside the shared accounting, and there were as many idle owners as
effect calls ever ran at once (16 frame threads: ~12 GB after a 4K render).
Giving a lent owner back now moves its kept textures to the shared free list,
so one POOL_BUDGET bounds what the free list and every finished effect call
keep. They skip Pool::release's wait for open encoders: only the lent Enc
recorded work on them, and that work is submitted or discarded. Each lend is
a fresh owner id, so a lease that outlives its call (an image handed to
another encoder) is released through Pool::release; this does not rely on
the result being read back.
Accelerator::render_end (called when an export ends) drops the kept textures.
* GPU: evict the pool's oldest free textures, guard escaped leases
Review of #1523 (follow-up to 0a884fb6):
- Pool::make_free rejected a texture when the free list was full and never
evicted, so once sizes that no longer occur (per-frame blur bounds, padded
layer sizes) filled the budget, every later free was a drop and every take
an allocation. It now evicts the least recently freed textures until the
new one fits.
- A lease read by another Enc (dispatch, copy, readback, display) or handed
to a cache is marked escaped and always released through the guarded
Pool::release, so a lent call can no longer reuse a texture that a still
open encoder reads.
- Docs: POOL_BUDGET caps the free list only; textures waiting for open
encoders and those of live Encs are outside it.
- export_async calls render_end from a drop guard, so the pool is also
trimmed when an export panics or its future is dropped.
- Tests use a private context with its own pool budget instead of a global
override; new tests for stale sizes and an image read by an open encoder.
* GPU: evict only stale pool sizes, count evicted textures until the GPU frees them
Evicting the least recently freed texture for every newcomer thrashed when a frame
frees more than the budget in a repeating order (blur/glow intermediates): each free
evicted the size the next call takes first, so misses came back to the old level and
evict + reallocate raised the peak.
The free list now evicts only sizes nobody asked for in the last 2 output frames
(Accelerator::frame_end, called by the export per frame); otherwise the incoming
texture is dropped and the kept set keeps hitting. Evicted bytes count against the
budget until the work submitted so far is done, so evictions run at most one budget
ahead of the GPU.
* GPU: lower the shared pool budget to 256 MB
On 4K RTX 3090 runs 256 MB matches the old code's peak reserved on d_blurglow
(9204 MB) and is 256-512 MB above it on c_effects (6592 vs 6080 MB), while
staying faster.
* GPU: end frames on the multi-frame render paths, route recording through Enc::record
Only the export pipeline called frame_end, so off export (preview, remote
frames) the pool's clock stood still, no size went stale and a full free
list dropped every newcomer. passes::comp_frame (exports), remote frames
and the preview workers now end each frame; test without pool_frame_end.
Docs: the frame clock is coarse with many frames in flight; dying only
throttles eviction, approximately; no cache path escapes images. Enc::record
is the single way to get the encoder outside context.rs and marks the
images it names (Enc::uses).
* GPU: cap in-flight frame memory on Vulkan and DX12 (#1550)
wgpu destroys a dropped texture or buffer only once the submissions
that used it have completed and the device is maintained. With several
4K frames in flight on an export or effect-heavy render, frames can
record and drop working textures faster than the GPU retires them,
accumulating freed-but-not-destroyed allocations until a 4-8 GB adapter
runs out of memory (#879).
Add a memory gate (MemCap) checked every 8 effect steps or layers:
- Default cap of 3 GiB on discrete GPUs (low enough for 4-8 GB cards
while leaving headroom above the 1.5 GiB upload cache and 256 MB
shared pool) and 2 GiB on integrated/software adapters, overridable
via EFFECTCRAFT_GPU_MEMCAP_MB (0 disables).
- Compare total_allocated_bytes against non-reclaimable bytes (upload
cache, pool free list, live leases, staging, zeros) with hysteresis
and a post-wait floor so the gate only drains when a wait can free
dropped allocations and does not spin when caches or other open
encoders hold memory over the cap.
- Submit and wait on the current Enc's own SubmissionIndex via a
dedicated poll_wait helper (serialized by poll_lock) so a timeout
simply carries on without retiring the context or its readback pool.
* GPU: warn on invalid EFFECTCRAFT_GPU_MEMCAP_MB, register crop sources, require GPU in the mem-cap child test
- memcap: an EFFECTCRAFT_GPU_MEMCAP_MB value that is not a whole number
of MiB or overflows 64 bits now logs a warning and uses the default
cap (it used to fall back silently, or wrap on overflow). Unset still
means the default and 0 still turns the gate off.
- Enc::crop records through Enc::record, so a foreign source lease is
marked escaped like in copy_into.
- The mem-cap child test starts its GPU through require_gpu, so
EFFECTCRAFT_REQUIRE_GPU=1 fails when the GPU does not start, and
prints a skip line (with the adapter name when the backend has no
allocator report), which the parent passes on.
* GPU: do not repeat the pool frame-end wait until evicted textures drain
A wait that frees none of the evicted bytes (timeout, or a WebGPU poll that
does not block) marks the pool stalled; later frames only poll until dying
drops. Test covers the no-repeat and partial-progress cases; dying decrement
saturates.
* Remote frames: end a pending frame that a new render replaces
A render arriving while a multi-pass frame waits for readbacks dropped that frame without
`frame_end`, so the GPU pool's clock skipped it. Errors return before `frame_begin`.
* Render: end the frame even when a comp frame panics
`passes::comp_frame` called `frame_end` only after a normal return, so a frame that
panicked (or whose future was dropped between passes) never aged the GPU pool.
* GPU memcap: warn once when kept memory alone is over the cap; test the edge frames
A frame whose non-reclaimable memory exceeds the cap never waits when reclaimable
bytes stay below the hysteresis (waiting cannot free the kept set) and now logs
that once. Tests cover a frame over the cap, two frames that fit one at a time, and a
dropped frame giving its memory back.
---------
Co-authored-by: Matej Kozlovský <MatejKozlovsky@users.noreply.github.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
* GPU: free effect working textures after each call instead of at frame end
A GPU frame keeps its encoder open while the CPU renderer runs layers with GPU
effects (Gpu::effects -> run_chain). Each call's Enc handed its textures back
to the shared pool, where they waited for every open encoder, so all effect
intermediates of a frame stayed allocated until the frame ended (17.7 GB by
layer 17 of a heavy 4K comp, then OOM). Effect calls now borrow a pool owner
(Enc::lent) that is given back once the call's work is submitted or
discarded; its kept textures are trimmed to the pool budget, so the rest are
freed after each call.
* GPU: one pool budget for idle effect calls, trimmed at the end of a render
Review of #1523: each idle lent owner kept up to POOL_BUDGET of textures of
its own, outside the shared accounting, and there were as many idle owners as
effect calls ever ran at once (16 frame threads: ~12 GB after a 4K render).
Giving a lent owner back now moves its kept textures to the shared free list,
so one POOL_BUDGET bounds what the free list and every finished effect call
keep. They skip Pool::release's wait for open encoders: only the lent Enc
recorded work on them, and that work is submitted or discarded. Each lend is
a fresh owner id, so a lease that outlives its call (an image handed to
another encoder) is released through Pool::release; this does not rely on
the result being read back.
Accelerator::render_end (called when an export ends) drops the kept textures.
* GPU: evict the pool's oldest free textures, guard escaped leases
Review of #1523 (follow-up to 0a884fb6):
- Pool::make_free rejected a texture when the free list was full and never
evicted, so once sizes that no longer occur (per-frame blur bounds, padded
layer sizes) filled the budget, every later free was a drop and every take
an allocation. It now evicts the least recently freed textures until the
new one fits.
- A lease read by another Enc (dispatch, copy, readback, display) or handed
to a cache is marked escaped and always released through the guarded
Pool::release, so a lent call can no longer reuse a texture that a still
open encoder reads.
- Docs: POOL_BUDGET caps the free list only; textures waiting for open
encoders and those of live Encs are outside it.
- export_async calls render_end from a drop guard, so the pool is also
trimmed when an export panics or its future is dropped.
- Tests use a private context with its own pool budget instead of a global
override; new tests for stale sizes and an image read by an open encoder.
* GPU: evict only stale pool sizes, count evicted textures until the GPU frees them
Evicting the least recently freed texture for every newcomer thrashed when a frame
frees more than the budget in a repeating order (blur/glow intermediates): each free
evicted the size the next call takes first, so misses came back to the old level and
evict + reallocate raised the peak.
The free list now evicts only sizes nobody asked for in the last 2 output frames
(Accelerator::frame_end, called by the export per frame); otherwise the incoming
texture is dropped and the kept set keeps hitting. Evicted bytes count against the
budget until the work submitted so far is done, so evictions run at most one budget
ahead of the GPU.
* GPU: lower the shared pool budget to 256 MB
On 4K RTX 3090 runs 256 MB matches the old code's peak reserved on d_blurglow
(9204 MB) and is 256-512 MB above it on c_effects (6592 vs 6080 MB), while
staying faster.
* GPU: end frames on the multi-frame render paths, route recording through Enc::record
Only the export pipeline called frame_end, so off export (preview, remote
frames) the pool's clock stood still, no size went stale and a full free
list dropped every newcomer. passes::comp_frame (exports), remote frames
and the preview workers now end each frame; test without pool_frame_end.
Docs: the frame clock is coarse with many frames in flight; dying only
throttles eviction, approximately; no cache path escapes images. Enc::record
is the single way to get the encoder outside context.rs and marks the
images it names (Enc::uses).
* GPU: wait for the empty submission while evicted pool textures still count
stale_sizes_do_not_starve_a_steady_size failed with 0 hits and 20 misses on
hardware adapters: pool_frame_end submitted an empty batch and then only did a
non-blocking Poll, and the queue's work-done callback did not run before the
next frame, so Pool::dying never drained and make_free stopped evicting stale
sizes.
While dying is non-zero, pool_frame_end now waits on that empty submission for
up to the readback timeout (the pool lock is already released). A wait timeout
just moves on and does not retire the context or its readbacks; other poll
errors still retire them. Without dying textures it keeps the cheap Poll.
Enc::crop records through Enc::record(&[src, &dst]) so both pooled images are
registered like on every other recording path (review).
* GPU: do not repeat the pool frame-end wait until evicted textures drain
A wait that frees none of the evicted bytes (timeout, or a WebGPU poll that
does not block) marks the pool stalled; later frames only poll until dying
drops. Test covers the no-repeat and partial-progress cases; dying decrement
saturates.
* Remote frames: end a pending frame that a new render replaces
A render arriving while a multi-pass frame waits for readbacks dropped that frame without
`frame_end`, so the GPU pool's clock skipped it. Errors return before `frame_begin`.
* Render: end the frame even when a comp frame panics
`passes::comp_frame` called `frame_end` only after a normal return, so a frame that
panicked (or whose future was dropped between passes) never aged the GPU pool.
* GPU pool: log the reuse rate at render end; test reuse across effect calls and frames
Log cumulative hits/misses (and reuse %) once per render before trimming. Tests
cover reuse within an open frame across effect calls, and across repeated frames.
---------
Co-authored-by: Matej Kozlovský <MatejKozlovsky@users.noreply.github.com>
Co-authored-by: bflatastic <bombayapirat@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
The pool learned that an effect reads a footage item's channels only at the
first aux() call, so the first frame was decoded twice (picture, then in
full), and the hint was never cleared after the effect was removed.
The session now gives the footage source the whole set of files read by
enabled 3D Channel effects (FootageSource::set_aux_footage) with every
project change (Session::bump: edits, undo, redo, history jumps, rollbacks,
project open), sending it only when it changed. The pool keeps that set and,
as a safety net, what aux() learns; a new set drops what was learned.
Refs #474
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Rename RenderOpts::scene_output to skip_hdr_step (Pipe::without_hdr_step):
the consumer decodes with Pipe::output_linear or applies the step itself.
- Alpha-only EXRs hold the alpha as it is in every output space (it is
coverage, not encoded colour; it used to be sRGB-decoded).
- Tests: a half-opacity layer over a coloured background (premultiplied and
RGB-over-background paths, golden hashes taken before the change), Rec.
2100 PQ and HLG EXR outputs, raw alpha, and a visible skip of the GPU test
without an adapter.
- docs/architecture.md: the Working Gamma curve for unmanaged projects, and
no HDR step for unmanaged projects or PQ / HLG outputs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
OpenEXR renders (Render Queue sequences and Save Frame As ProEXR) received
the frame after Project Settings > HDR, so Compand and Tone Map (the default
of every OCIO project) compressed highlights and changed values below 1.0
before the writer decoded them, always with the sRGB curve.
- RenderOpts::scene_output: the top-level comp converts to the output space
without the HDR step (CPU and GPU walks, via Renderer::output_pipe).
- Pipe::output_linear: the writers decode with the output space's curve
(the Working Gamma's when unmanaged) instead of a fixed sRGB curve.
- Render Queue EXR jobs and ProEXR set the flag; PSD layers, 8-bit stills,
movies and the viewer keep the HDR step. The flag is part of the viewer
frame and disk cache keys (opts_hash).
Golden hashes, taken before the change, pin PNG/TIFF/Targa bytes for
8/16/32 bpc x Clip/Compand/ToneMap, Clip-mode EXR samples with an sRGB
output, and the layered PSD bytes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Animator Opacity stacked by multiplying, so once one animator set 0 %
no animator below it could raise it again. Each animator now blends the
opacity left by the ones above it toward its own value. A single Opacity
animator gives the same result as before.
Fixes#2838.
A project file can carry footage with a frame rate of den 0; frame_position
divided by zero while rendering it. Treat non-positive rates as having no
frame to land between.
#2653 (#2618) changed what an adjustment layer in a mode other than Normal
renders, but its cache keys are unchanged, so frames cached before it keep
showing the Normal result. Bumping the render version keeps old entries
from being reused.
An adjustment layer replaced the comp below with its effects' result whatever its blend
mode. As in After Effects, a mode other than Normal now blends the adjusted comp onto the
comp below in that mode, inside the footprint, track matte and opacity (an adjustment
layer with no effects in Screen screens the comp with itself). CPU and GPU paths.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Noise, paint, the shutter, and the comp scene were still reaching the footprint key, so a finishing adjustment re-uploaded a full frame every frame. The key now uses EffectScope::Footprint and ignores that stack, while a moving mask still re-keys it.
Add Per-Corner Roundness to Rectangle Path: a checkbox (off by default) and four Pixels properties (top-left, top-right, bottom-right, bottom-left). New rectangles initialise the four values from their Roundness, so ticking the box keeps the shape; the radii are animatable and expression-driven.\n\nrect_radii() builds the path clockwise from the top-left with kappa quarter-circle arcs. Overlapping radii scale down like CSS border-radius; negative and non-finite radii are square; rect() delegates to it, so uniform roundness clamps to half the shorter side as before.\n\nLottie rc has one r: four equal static radii export as that value; mixed or animated radii export roundness 0 with a warning. Import puts r into the four per-corner values so ticking the box on an imported rectangle keeps the shape.\n\nTests: path geometry and clamping, rendered per-corner pixels, Lottie equal/mixed/checkbox-off export.
#2410's dependency walk for a soloed 3D model was written before
reads_other_times took the effect count; it asks about all effects (None),
as the other walks do.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Text strokes were always drawn with round joins. The document gets a
Line Join (`lineJoin: miter|round|bevel` in layer.setText / newText),
Miter for new text as in After Effects, Round for documents saved before
it so their look is kept. The 2D and Advanced 3D (extruded) text strokes
use it, expressions' style copies carry it, and the Character panel has
Miter / Round / Bevel buttons under the stroke width.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Valentin KleinandClaude Sonnet 5.5
ash
FredApps
gmrmad1andFredApps
JonRC
tylerjenningsw
Brandon Thomas
jayzed80
becauseno7
Vikram Bhaskaran
Camille Moine