* 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>
Effects that grow their buffer (CC Radial Fast Blur, CC Lens, Magnify,
the distorts and blurs) could pad a 20 x 20 layer by 4096 px a side, a
~8200 x 8200 f32 buffer plus its clone. Buf::pad (CPU and GPU) now cuts
the pad so the grown buffer stays within 16 Mpx or four times its own
area, whichever is larger: small layers are bounded at 256 MiB, big
layers still grow as before in practice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Resolve the conflict in crates/gpu/src/fx_stylize.rs: main moved ops::warp to
the ops::WarpOptions parameter struct; the branch's per-sample transform loop
(motion-blur samples, now passing the sample by reference) uses the new
signature with the branch's reference argument kept.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
After Effects' Curves effect interpolates its points with a natural cubic
spline (zero curvature at the ends), clamped to 0..1: AE 2026 renders of an
RGB curve (0,0) (125,199) (255,255) and a Red curve (0,0) (128,44)
(255,255) (cases 962 / 963, ROUND4) give exactly the natural spline's values
on every grey step (0, 30, 59, 115, 164, 202, 227, 242, 250, 255 and 0, 0, 1,
6, 19, 44, 83, 135, 194, 255). EffectCraft used a monotone (Fritsch-Carlson)
spline: 34 / 67 / 125 / 171 ... on the RGB curve, up to 10/255 off (ROUND4
measured up to 6 levels on real projects); now within 1/255, CPU and GPU (the
same Hermite tables carry the spline's knot slopes). Lumetri's curves and
Glow's Arbitrary Map keep the monotone spline (not measured).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
After Effects draws the Transform effect as position + K * R * S * (p - anchor)
with the skew K = rot(-axis) * shearY(tan skew) * rot(axis) applied after the
rotation, and keeps what lands outside the layer's bounds. EffectCraft skewed
before rotating (shearX about the axis) and clipped to the layer.
Measured in AE 2026 on a grey 960x540 solid (cases 723, 819, 820): 723 (anchor,
position, scale 120/80, skew 10 axis 20, rotation 30, opacity 70) differed in
11.25 % of the frame and now in 0.065 % (edges, max 16/255), CPU and GPU alike;
819 / 820 (skew 15 axis 60, rotation -25, scale 130/70, anchor) match to the
edge. The buffer grows on each side by that side's overshoot (also for the
motion-blur samples), capped at 16 384 px a side and 16 Mpx in all so a layer
moved far away costs a bounded allocation (regression test); adjustment layers
keep their size.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
After Effects 2026 draws Fractal Noise's Block noise at Scale 100 and
Complexity 1 in cells of 66.67 px (the value changes every 66.67 px along a
row), not 100 px; Soft Linear noise has the matching correlation length.
EffectCraft's lattice is now Scale x 2/3 px (NOISE_CELL), CPU and GPU. Case
904 E03 (Soft Linear, Complexity 4, Scale 100): the noise's autocorrelation
at 32 px was 0.68 (AE 0.41), now 0.54; pixel values cannot match (other
seeds). Still open: AE's noise spreads about 20 % wider around its mean (std
48 vs 40/255 in E03). Existing projects and the Backgrounds presets with
Fractal Noise get finer noise (two thirds of the former cell size).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With Glow Based On Color Channels, After Effects gives the glow the layer's whole
alpha (blurred), not the thresholded part, and Intensity scales only its colour:
a layer below the threshold darkens what lies under its edge. Fitted to AE 2026
renders of a grey solid over red (cases 524, 807-809): the halo is the same at
Intensity 1 and 3 and with Composite Original Behind or On Top, and its edge is
an erfc step of sigma = 0.384 x Glow Radius at radius 10 and 30 (was radius / 2).
524 (Glow defaults on a 50 % grey solid): EffectCraft had no halo (max 116/255
off at the edge); now red 248/234/208/165/140 at 7.5/5.5/3.5/1.5/0.5 px outside
vs AE 249/235/206/164/140, max 2/255 over the frame, CPU and GPU. The blur
width applies to every glow; alpha-based glows keep their coverage rule
(thresholded alpha, scaled by Intensity). Not matched yet: the brightness of glows above the
threshold (AE about 3.7 x brighter for a white solid, cases 806, 810, 811).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Since #659 every text or shape layer with an enabled video effect grew its
effect buffer to the comp-sized layer rectangle before the stack ran, so
glow and blur — which only need their own margins — allocated a 1920x1080
RGBA32F texture per layer per frame in flight (~5 GB peak VRAM in parallel
export, measured on an RTX 3090).
Effects that actually place pixels outside the painted content already
grow the buffer themselves by exactly what they need (CC Tiler, CC Smear,
CC Mosaic sharp via layer_cover_pad / Buf::pad), on CPU and GPU. The
blanket target.cover(layer_size) in Renderer::run_effects_on, the
FxTarget::cover trait method with its CPU default, and the GPU
GpuTarget::cover texture allocation are removed; Buf::cover/cover_plan
stay as the effects-crate helpers they were designed as.
The #659 regression tests (tiling across the layer region, centred-shape
edge pinning) and the render/effects/gpu suites pass unchanged.
An authored layer's effect buffer spanned only its painted content — the glyph
bounding box for text, the drawn bounds for shapes — while the size the stack
measures from (effect_bounds) is the whole comp-sized layer rectangle. Tiling
and displacement effects computed correct output for pixels outside the painted
rectangle and then wrote them into the buffer that was too small, so CC Tiler
repeated only within the text box and CC Smear lost its liquid trail at the
glyph edge (#658). Buf::cover now grows the stack's input to the layer's
rectangle with transparent margins — what a sized source's buffer already is —
applied by run_effect_stack whenever an enabled video effect will run, and
natively on the GPU by reallocating and copying the texture. The placement and
the sized-source paths are unchanged; two regression tests (CC Tiler spans the
region, CC Smear trails past the text bounds) fail without the growth.
GpuContext::new built every compute pipeline up front and failed as a
whole when one didn't build. On DirectX 12, FXC (wgpu's default D3D shader
compiler) aborts on the `pointwise` effect kernel, so the whole app fell
back to CPU compositing.
The compositor's own kernels are still required, but the effect kernels
(`pointwise`, the blur/glow/shadow passes and every fx_* kernel) are now
optional: one that fails to build is logged and listed
(GpuContext::unavailable_kernels), and a GPU effect chain that dispatches a
missing kernel reports it (Enc::take_missing) so those effects render on
the CPU while everything else stays on the GPU.
This does not make FXC accept `pointwise` (that needs a Windows/FXC box to
bisect); it limits the damage to the effects using it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QiAJeFttZp83Fh7YuzAR5f
The geometry stays on the CPU as per-frame plans shared with the CPU effects (bolts, mask
strokes, scribbles, contour segments, audio marks, glyph strokes, waves, flood-fill region,
eyedropper sample); segments and convex polygons are tile-binned and rasterised on the GPU
with the CPU's coverage functions. CPU output unchanged (gen_golden). CC Threads' exact ties
at round parameter values are a documented exception; Fractal stays on the CPU (needs f64).
Regenerated docs/effects.md, docs/parity.md EFF-5.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GPU render passes for CC Particle World / Systems II, Particle Playground,
CC Ball Action, CC Pixel Polly, CC Scatterize and Curl Noise (fx_particles,
chunked multi-pass raster); Shatter wireframe views, Foam extras, Timewarp
with a Matte Layer, strong Warp Fisheye/Twist (CPU f64 inverse map) and
custom .ocio configs no longer fall back to the CPU. CPU refactors pinned by
particle_golden / timewarp_golden / sim_golden hashes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Inner/Outer Key's trimap and Cleanup strokes, Color Link's sample and CC Glass/Plastic's
light are shared with the CPU effects (CPU output unchanged: gen_golden). CC Plastic now
reads its Bump Layer (it read the nonexistent bumpMap id).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CC Rainfall, Snowfall, Star Burst, Bubbles, Drizzle, Hair, Mr. Mercury,
Caustics, Wave World, Foam, Shatter, Card Dance and Card Wipe render on the GPU
(fx_sim.rs / fx_sim.wgsl). The per-frame simulation / layout stays on the CPU
and is shared with the CPU effect through plans (effects::sprite_plan,
piece_plan, bubble_list, mercury_blobs, drizzle_drops, wave_world_plan,
caustics_setup); a tiled item rasteriser composites sprites, bubbles, blob
fields and textured pieces in the CPU's order and shading. Layer-dependent
sprite colours (refracting rain, star-burst and hair roots) are sampled per
pixel; piece plans read the frame back once.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- GPU in the browser: each frame worker opens its own WebGPU device (Gpu::request_deferred) and
renders the GPU compositor and GPU effects with deferred readbacks (effectcraft-gpu deferred):
readbacks keyed by content start with map_async and miss, the pass returns placeholders, the
layer cache's gate drops inserts, and FrameServer::resume renders again once the device
delivered. Backend Auto times whole frames per comp in the workers. Frames with effects go to
GPU workers instead of the page. CPU fallback without WebGPU in workers (?nogpuworkers).
- Fix: a WGSL constant (3.40282347e38) that Chrome rejects invalidated every GPU kernel in Chrome
(the page's compositor too); a test checks every float literal fits f32.
- Disk cache in the Origin Private File System: workers write sealed frame entries with sync
access handles (temp file + move), the page keeps the DiskIndex (LRU under the settings' limit,
at most half the quota) and reads hits instead of rendering; survives reloads; blue cache bar.
- Storage manager: Settings ▸ Disk ▸ Browser Storage (usage/quota, persistent storage, Clear
buttons, automation ids) and storage.info / storage.persist / storage.clear (StorageHost).
- Docs: web.md (GPU workers, deferred passes, disk cache, storage manager, threading design:
one engine instance per worker), architecture, preferences, parity (WEB-1), roadmap.
- Tests: deferred passes vs CPU on a real device, placeholder gating, frame server passes and
Auto, GPU worker routing + disk cache across a reload, storage commands and widgets, store
usage by kind; smoke test: GPU effect frames in workers with pixel parity, disk hits after a
reload, storage manager.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Timewarp without a Matte Layer is now planned once (time_fx::timewarp_plan: frames, shutter
instants, motion fields) and built by the CPU or the GPU from the plan; with a Matte Layer it
renders on the CPU (catalog::gpu_supported).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
VR Blur, Chromatic Aberrations, Color Gradients, Converter (all projections and
cube layouts), De-Noise, Digital Glitch, Fractal Noise, Glow, Plane to Sphere,
Rotate Sphere, Sharpen and Sphere to Plane run on the GPU (fx_vr.rs /
fx_vr.wgsl) with the CPU's equirectangular maths: seam-wrapping, pole-continuing
sampling and latitude-widened box blurs, per eye for stereo layouts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CC Environment now fades nearly transparent environment texels to black continuously
(colour / max(alpha, 1e-3)) instead of a hard 1e-6 un-premultiply cut-off, which turned the
filtered map's float noise into colour.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Apply Color LUT, OCIO CDL / Color Space / Display / File / Look Transform and
Color Profile Converter compile to colour programs (effects::color_program,
built next to the CPU effects) that fx_lut.wgsl interprets per pixel; LUTs
(1D, 3D, cineSpace shapers, inverses) are read from the storage buffer with the
CPU's nearest / trilinear / tetrahedral interpolation. Lumetri's Input LUT and
Look no longer force the CPU. Custom .ocio configs stay on the CPU.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
3D Channel Extract, Cryptomatte, Depth Matte, Depth of Field, EXtractoR, Fog 3D,
ID Matte and IDentifier run on the GPU (fx_depth.rs / fx_depth.wgsl): aux planes
upload as a texture and are resampled onto the buffer grid with the CPU's index
arithmetic; tests compare against the CPU at 1x / 2x aux resolution, padded
buffers, half resolution and 8 / 32 bpc composites.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- gpu: pooled working textures (released once every encoder that could read
them is submitted), shared zero images, reused readback staging buffers,
direct readback into Image; NaN-poisoned scratch images under test and
NaN-aware diffs; validation errors panic under test
- render: AutoPick (per comp / scale / path EWMA of CPU vs GPU frame times,
warm-up and periodic re-probe) behind Backend::Auto; viewer uses it too;
cached adjustment footprints (uploaded once)
- gpu fx_warp: Warp, Bezier Warp, CC Bend It, CC Page Turn, Smear, Reshape,
Color Emboss, Cartoon; sliding-histogram median for large radii; CC Power
Pin perspective < 100
- cli: bench --gpu reports Auto and per-frame upload / readback MB
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Levels (Individual Controls), Gamma/Pedestal/Gain, Photo Filter, Change Color,
Change to Color, Leave Color, Broadcast Colors, Color Balance (HLS), Video
Limiter, PS Arbitrary Map, CC Toner, CC Color Offset, CC Kernel, Auto Levels /
Contrast / Color and Equalize (histograms measured on the CPU, applied on the
GPU) on the GPU; Hue/Saturation colour ranges and Levels' channel controls no
longer fall back to the CPU.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Transform: Use Composition's Shutter Angle / Shutter Angle, sampling the transform across
the shutter via EffectHost::params_at (EffectEnv::shutter carries the comp shutter when
motion blur is on); the GPU path defers to the CPU when blurring.
- Glow: Glow Operation (blend modes) and Glow Colors > Arbitrary Map.
- Cartoon: Advanced > Edge Enhancement (shock-filter sharpening / softening).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Fill: Fill Mask, All Masks, Horizontal / Vertical Feather, Invert outside the mask
(GPU falls back to the CPU when masks are used).
- Refine Soft / Hard Matte: Reduce Chatter (temporal steadying of oscillating edges) and
Use Motion Blur (samples, shutter angle, higher quality) from neighbouring frames.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Lumetri: HSL Secondary key/refine/correction with Show Mask, Hue vs Sat/Hue/Luma,
Luma vs Sat, Sat vs Sat and per-channel RGB curves, High Dynamic Range (HDR White /
Specular), Input LUT and Look (custom .cube + built-in looks).
- Hue/Saturation: Channel Control with Reds..Magentas ranges (range + fall-offs).
- Levels: Channel popup and Red/Green/Blue/Alpha controls; Curves: Channel parameter.
- catalog::param_shown: Effect Controls shows only what a popup selects.
- Timewarp Pixel Motion: block-matching flow (raster::flow::block_flow_window) with
Vector Detail, Smoothing, Error Threshold, Block Size, Weighting, Correct Luminance,
Filtering, Build From One Image; Matte Layer/Channel, Warp Layer, Show, Source Crops.
- GPU falls back to CPU when channel/range controls are used.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New L3 crate effectcraft-gpu: a wgpu compute backend for the per-frame
compositing walk (layer transforms with the CPU's nearest/bilinear/bicubic
sampling and minification pre-filter, motion-blur accumulation, track
mattes, Preserve Transparency, layer style passes and knockout, all 38
blend modes as a WGSL port of color::blend, 8/16 bpc clamp+quantise,
colour-space conversions). 3D runs, adjustment and wireframe layers run on
the CPU between GPU steps. GPU effects with the CPU semantics: Gaussian,
Fast Box and Directional Blur, Glow, Levels, Curves, Hue/Saturation, Tint,
Fill, Gradient Ramp, Fractal Noise, Drop Shadow, Brightness & Contrast,
Exposure, Invert, Transform (registry gpu badge from effects::GPU_EFFECTS).
render: Backend (Cpu/Gpu/Auto) in RenderOpts, the Accelerator trait and
Renderer::accel, GPU effect chains in the effect stack, hooks for the GPU
walk. Project setting gpu_acceleration (Mercury GPU Acceleration / Software
Only), render.backend command, file.projectSettings renderer. Session and
Render Queue exports carry the accelerator. CLI --gpu and bench --gpu.
Tests compare GPU vs CPU renders (blend modes x bit depths, transforms,
masks, mattes, motion blur, colour management, styles, collapse,
adjustment fallback, every GPU effect) and skip without an adapter.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
