Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 2232354b902238827b2e80a29ef266c2e5bde272)
(cherry picked from commit 95da7943454ca5b2575993d11f9ba046822ce255)
As in After Effects, the Delete key and the Project panel's trash button ask before deleting
items that layers in other compositions use, saying how many layers in how many compositions go
with them (Delete / Cancel; Escape cancels). Unused items are deleted at once; agents running
project.delete directly are not asked. project.usage {items?} reports the count, sharing the
deletion's item and usage helpers. Follow-up to #72.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Japanese (and Chinese, Korean) text in a family that lacks the glyphs, such as the default Inter,
rendered as missing-glyph boxes: `fallback_for` only searched the bundled faces and faces some
other layer had already loaded, so the result depended on render order. Now, after those cheap
candidates, it tries the platform's usual family for the character's writing system (kana and
hangul pick their language; Han ideographs follow the locale, Japanese first when it is unknown,
`EFFECTCRAFT_CJK_LOCALE` overrides) in the preferred face's style, and finally any installed face
that covers the character, reading files without keeping them. Results are cached per 256-character
block and preferred face.
`resolve` also accepts a face's full name (`Yu Gothic Bold`) or PostScript name (`YuGothic-Bold`),
which is how After Effects scripts set `textDocument.font`; the sfnt reader now records name id 6.
Tests: script classification, PostScript / full-name resolution on the bundled Inter, and (native,
skipped when no Japanese family is installed) that kana, ideographs and CJK punctuation in Inter
Bold fall back to an installed Japanese family's bold cut, with the cache returning the same face.
`cargo xtask ci` runs clippy and the tests with --release, where egui has no Style::debug (it
exists only with debug_assertions), so ui_agent_audit.rs failed to compile since M13.14. The
debug option is set only in debug builds, and the test, which needs the widget types egui records
with it, is ignored in release builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
As in After Effects, a double-click below the items in the Project panel opens File ▸ Import ▸
File... (rows and the scroll bar keep their own clicks). Dropping files on the window already
imported them; the test now covers both. (On Wayland, winit delivers no file drops at all, so
dragging files in works on X11, macOS and Windows.)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The details header (item name, size, duration, codec) and the "Select an item" hint were painted
with no width limit, so in a narrow Project panel they ran past its right edge. They now stop
10 px short of it and end in "…" when they don't fit (widgets::text_fit, a shared helper), and
are registered for automation (project.details.*).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Covers what issue #46 reports (v0.1.1 on Linux): a panel dragged by its hamburger onto another
group's edge docks there as its own group, and dragging the gap between panels resizes them.
Both work on main (tab dragging was reworked in M3.15 for v0.2.0); this keeps them working.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
While paused, the app prefetched the frames around the current one as soon as the viewer's
frame finished. During a drag every step changes the project, so those frames were stale at the
next step, and they held the CPU and GPU the next viewer frame needed: on the demo project a
drag step took 102 ms on average to show (up to 413 ms) though a frame renders in 20 ms. Paused
prefetch now waits until edits pause for 0.3 s: 16 ms a step (11 ms render).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Two implementations of Show 3D Reference Axes had landed (M13.5 and M14.5) and both drew,
offset by a few pixels: lines, and arrows. Keep the arrowed one (perspective-correct, shown
while a 3D layer is active) and test that only one is drawn.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The handle drag rescaled incrementally from the current scale every frame: errors built up,
edge handles scaled both axes, and once a factor reached ~0 the drag could never bring it
back. It also tested the handle at the point where egui recognised the drag (past the drag
threshold), so a quick grab often missed the handle and moved the layer instead.
Now, as in After Effects, the drag scales about the anchor point relative to the layer as the
drag began, so the grabbed point stays under the pointer: corners scale both axes, edges one,
Shift keeps the proportions, and dragging past the anchor flips. Selection-tool gestures start
at the press point.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
sysctl hw.physmem / hw.pagesize / vm free+inactive page counts, hw.model,
and freebsd-version, with a unit test for the sysctl parser. The FreeBSD
CI workflow itself landed with #69.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
egui-wgpu creates its device on GL with WebGL2's limits (no storage buffers), so the desktop
app on FreeBSD built the compositor's pipelines on it: validation errors (only logged in
release builds) and wrong frames. GpuContext::new now checks the limits its pipelines bind
(storage buffers and textures per stage, bind groups, workgroup sizes) and returns Err, which
keeps the app on the CPU. ui_gpu skips its GPU-vs-CPU comparison when that happens on GL.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
FreeBSD's CI VM (Mesa llvmpipe over GL) failed 22 effectcraft-gpu tests:
- Advanced 3D's camera-depth target was R32Float, which GLES can't render to without
EXT_color_buffer_float: create_render_pipeline failed validation (release builds only log
that and draw nothing). The target is now R32Uint holding the f32 bits (renderable on every
backend, lossless), and GpuContext checks the adapter's format features (render attachment,
blendable colour) and independent blending up front; when they're missing the render
pipelines are never built and Advanced 3D renders on the CPU.
- Signed `%` in the shaders: naga emits GLSL `%`, undefined for negative operands, which broke
the VR effects' horizontal wrap and Cell Pattern's tiling. A shared `imod` never takes `%` of
a negative number.
- asin/acos: Mesa's are coarse approximations (~1e-4 rad), visible in every lens/sphere/cylinder
mapping. `asin_p`/`acos_p` (Cephes' asinf polynomial) are f32-accurate on every backend.
- "Could not lock adapter context": the GL backend has one context lock per adapter with a
timeout, so tests driving devices concurrently panic. Every gpu test helper that hands out a
device holds one process-wide lock for the calling test; ui_web_job_gpu's two device tests
share a lock too.
- On GL, up to 1 % of pixels may flip across a decision boundary (division through a
reciprocal, unfused fma); the GPU-vs-CPU comparisons allow that there only.
Also adds the FreeBSD workflow (from the FreeBSD CI PR) so this branch is checked there.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- effectcraft-segment: a FaceModel interface (find a face in a region, follow it; a Topology
maps the model's points to the tracker's landmarks and outline), a TensorFlow Lite reader and
interpreter (float ops, float16 weights folded; matches ai-edge-litert to ~1e-6), and
MediaPipe Face Landmarker (Google, Apache-2.0: BlazeFace + Face Mesh V2, 3.8 MB, pinned
SHA-256). The registry gains a task and authors per model; installed weights get a NOTICE.
- effectcraft-track: FaceTracker::new_with uses the model, falls back to the classical fitter
when it finds no face; chin and jaw come from the outline either way so measurements agree.
- Engine: roto_models becomes models (both tasks); face.models / face.model.* commands,
Settings ▸ Face Tracking, `wait` on select/install/download for scripts; track.mask reports
faceModel.
- Credits: ATTRIBUTION.md lists the optional models and the Apache-2.0 specs the code follows.
- Docs: README, ROADMAP (G7), gaps, parity (TRK-3 score unchanged pending G1), agents,
preferences, architecture, segment README.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- New L1 crate effectcraft-segment: the MaskModel interface (prompts → foreground probability),
a registry of open-licensed models (licence must be MIT/Apache/BSD/…; URL, size, SHA-256),
a PyTorch checkpoint reader (zip + pickle subset), dependency-free SHA-256, and MobileSAM
(TinyViT-5M + SAM prompt encoder / two-way transformer / mask decoder) on rayon and ndarray's
safe GEMM. Loads the official mobile_sam.pt as published; checked against candle
(embedding 0.3% RMS, 99.9% equal mask pixels); ~0.7 s/frame encoder at 32 threads, cached
per frame, ~0.1 s decoder.
- track::roto: segment_with / propagate_with prompt a model with the strokes or the flow-warped
matte (box, core points, mask); its probability is a strong, unblurred prior for the same
graph cut (strokes stay hard constraints); a failing model or one disagreeing with the flow
(IoU < 0.5) is ignored. Prior gets a weight.
- effects::roto: Version 2.0 / 3.0 use the model chosen in Settings (set_model / model_for);
the model id is part of the chain seed.
- engine: roto_models (roto.models, roto.model.select / install / download / remove), weights in
the models folder (host::config_dir()/models, shared by the app, CLI and MCP), background
loading, downloads through the system curl, verified; Settings ▸ Roto Brush page (prefs
roto.model). host::config_dir moved from the desktop app.
- Tests: segment (kernels vs direct sums, checkpoint reader, SHA-256 vectors, registry), track
(fake model: prior shapes the cut, failing / disagreeing models ignored; real weights with
EFFECTCRAFT_MOBILESAM: base IoU 0.989, propagated ≥ 0.980 vs classic 0.973), engine (registry,
verification, selection; real install + load), MCP.
- Docs: README, ROADMAP, parity (MSK-4 done: 90/2/0), preferences, agents, architecture,
crates/segment/README.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- audio::AudioScrub: an open output that plays a faded snippet (one frame, 30–100 ms) of the
comp's mix at each new frame, replacing what is still queued; holding still repeats nothing.
The mixing is shared with the preview feeder (mix_block); AudioFeed::clear.
- The Timeline ruler: Ctrl/Cmd-drag scrubs audio (EffectcraftApp::scrub_audio opens the output
through the audio device hook on first use). The Audio panel meters follow the snippets; an
idle second closes the output.
- playback.scrubAudio {time?} for agents.
- Tests: ui_audio_scrub (snippet length, fades, same-frame no-op, replace not pile up, silent
comps, Cmd-drag in the ruler, idle close) with a fake device. Docs: parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- commands::markers::nested_markers: a precomp layer's comp markers at their times in the
containing comp (start and stretch in closed form; with time remapping or a stretch over
protected regions, the first frame whose source time reaches them), within In–Out.
- markers.nested {layer}: the same list for agents.
- Timeline: drawn on the precomp bar as outlined read-only markers (positions cached per
revision), hover names them, double-click opens the nested comp at the marker; automation id
timeline.layer.<id>.nestedMarker.<i>.
- Tests: engine (start, stretch, trim, time remap, non-precomp error), UI (draw, label, open).
Docs: parity, control protocol.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- PanelKind::Viewer(n): more Composition viewers (panels::viewers). One is active (the full
interactive viewer, showing the active comp); the others show their own comp's frame at that
comp's current time (CPU or GPU frames, their own textures) and become active on a click or
their tab, their comp becoming the active comp.
- view.newViewer (View menu, the Composition panel menu, Alt+Shift+N): a new viewer of the
active comp next to the one in use, which is locked. Opening a comp while the active viewer
is locked shows it in an unlocked viewer, or a new one when every viewer is locked. Closing
the active viewer hands over to another with its comp. Tabs are named after their comp, with
their own lock and swatch; Preview full screen maximizes the active viewer.
- UI state: viewers (comp per viewer) and active_viewer (serde).
- Tests: ui_viewers (lock on New Viewer, routing, click to activate, all locked → new viewer,
close hand-over, tab titles; an ignored snapshot). The bounded-drag frames test asserts what
holds when freed addresses are reused. Docs: parity, control protocol.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- FrameKey's identity is the comp's content (frames::comp_content), not the project revision:
an edit keeps the cached frames of comps it doesn't touch and undo finds the frames of the
state it returns to. The identity hashes the addresses of the comps a comp draws (~0.1 ms on
the 1500-layer test project; the full content hash took ~117 ms), by value the footage, solids,
proxies and project settings; with expressions, every item. Frames::content_of keeps a
project snapshot per identity with frames (or in flight), so Arc::make_mut always copies an
edited comp and its address can't be reused meanwhile; past 64 identities the least recently
used go with their frames (long drags).
- drop_stale is gone (frames of other contents are evicted least-recently-shown first).
- disk_cache::comp_content_key covers items' proxies and Use Proxy (it served stale frames).
- Tests: frames (content keys, edits elsewhere and undo, bounded drags, in-place edit safety),
ui_ram_preview (end to end), the disk cache proxy key. Docs: roadmap (M4 done), parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Home covers the whole workspace below the Tools bar (the panels wait behind it) instead of
the Composition panel.
- Left rail: New Project (primary) and Open Project, the Home / Templates / Learn pages with
icons, and the community links and ArtCraft apps at its foot.
- Home page: "Welcome to EffectCraft", quick-start tiles (New Composition, Open Demo Project,
Import Footage, New from Template) and Recent as a table — thumbnail, name and folder,
Opened (relative), Size, Kind — with a filter field, Clear list and an empty state.
- Theme tokens throughout (light themes follow); web_time for "now" (works on the web).
- Tests: formatters (opened / size / kind) and the new automation ids. Docs: parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- paint_logo draws docs/brand/artcraft-mark.png (the README's / getartcraft.com's ArtCraft
mark) as supplied, decoded once into a mipmapped texture, instead of a drawn placeholder.
- The "More ArtCraft apps" sibling and help.sibling named DrawCraft (storytold/drawcraft, which
doesn't exist): it is VectorCraft (storytold/vectorcraft); provenance comments in the path
crate follow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Collapsed precomps: the renderer records the precomp layers a nested comp is drawn through;
motion-blur sub-samples re-evaluate their transforms (when their motion blur is on) and map
times down through each one, so the precomp layer's own motion blurs its nested layers. The
containing (outermost) comp's shutter applies to collapsed layers. The GPU compositor gets
this through Renderer::placement.
- Shape and text layers whose content changes within the shutter (compared by its cache key)
draw their source at Samples Per Frame times and average them; the cache key covers every
sub-sample's content.
- Tests: tests_motion_blur_depth (collapsed vs flat, the containing shutter, shape content vs
the same move by Position); each fails without the change.
- Docs: roadmap, parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- The Timeline shows one tab per open comp (dock DocTab). Clicking a tab shows that comp; ×
or middle-click closes that comp's Timeline (comp.close), not the panel. Opening a comp brings
its viewer and Timeline tab to the front.
- Project panel: a double-click on an item's name opens it (a comp gets its own tab) instead of
renaming; Enter and the context menu rename.
- Tab drags: over a tab strip an insertion mark shows where the tab lands, and it lands there
(Layout::insert_tab), which also reorders a group's own tabs; side zones are unchanged.
- Tests: ui_tabs (double-click, tab switch / close, drag reorder and cross-group insert; an
ignored mid-drag snapshot); the Timeline close test follows the new behaviour.
- Docs: parity, control protocol.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- effect.pickColor {effect, param|prop, x, y (layer px), average?}: the colour of the effect's
input at a point, set on the parameter (one undo step). Keying effects' Effect Controls
eyedroppers use it, since the shown frame is already keyed; Ctrl/Cmd+click averages 5x5.
- effects::aliases / name_matches: Effects & Presets and effect.list (MCP list_effects) find
Key Light by After Effects' name ("Keylight (1.2)").
- End-to-end test: a green-screen plate with an uneven screen, spill and soft edges keyed via
effect.apply "Keylight (1.2)", pick, Clip Black / White and a render; UI eyedropper test.
- Docs: parity, agents (green-screen recipe).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- J / K (Go to Previous / Next Keyframe) and Ctrl+Alt+A (Select All Keyframes) act on what the
Timeline shows, as in After Effects: the keys of its revealed properties, layer and comp
markers and the work area (they used every property of the selected or all layers, revealed
or not). The commands take `visible: [{layer, prop}]`, which the Timeline fills in; agents
calling them without it keep the previous behaviour.
- Graph Editor ▸ Auto-Select Graph Type (an "Auto" button, on by default as in After Effects):
the speed graph when every property shown is spatial (Position, Anchor Point…), else the
value graph; choosing Value or Speed turns it off.
- Docs: parity, agents.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Go to Next / Previous Keyframe (K / J) also stops at the work area's start and end, as in
After Effects.
- With one keyframe selected, the Info panel shows its property, time and value instead of
"1 keyframes selected".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Keys copied from several layers and pasted onto as many selected layers go first to first,
second to second, as in After Effects; every target used to receive every copied property
(the later ones overwriting the earlier).
- Edit ▸ Paste Reversed Keyframes mirrors the clipboard over its span and pastes it through
`keys.paste`, so it now follows the same rules as Paste: the layer's time stretch, the
selected target property, Hold-only properties, value conversion and roving keys (its own
simplified paste ignored them and kept existing keys' easing on a time collision).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Double-clicking a keyframe of a numeric property (or its context menu ▸ Edit Value…) opens
the property's value dialog for that key, as in After Effects (it only moved the current
time); other keys still move the current time to them.
- The keyframe context menu has Select Equal Keyframes, Select Previous Keyframes and Select
Following Keyframes (`keys.selectEqual` / `selectPrevious` / `selectFollowing`).
- Roving keys draw as small dots in the Timeline, like After Effects.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- `keys.transform {fromStart}`: each step of a merged drag gives the whole transform since the
drag started and applies to the keys as they were then. Keys squeezed onto one frame on the
way (Alt-drag scaling, the transform box) are no longer merged for good, and a slow
transform-box move, whose per-frame offsets rounded to nothing, now moves the keys.
- Dragging a key in the Graph Editor moves every selected key in time and value (it moved only
the grabbed one), by how far the pointer went from where the button went down (the key no
longer jumps under the pointer); Shift keeps the drag to the time or the value axis.
- Shift+click toggles a key in or out of the selection; Ctrl+click switches Linear ↔ Auto
Bezier and Ctrl+Alt+click toggles Hold, as in the Timeline.
- The transform box and the Timeline's Alt-drag scale send whole-gesture values.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- A drag (keys.move steps sharing a merge key) applies its total offset to the keys as they
were before it (`Session::key_move`, `prop::shift_keys`): dragging over another key of the
property no longer deletes it; only the drop position replaces a key, as in After Effects.
- Keys of time-stretched layers move with the pointer: offsets are applied in comp time and
snapped to comp frames (drags and Alt+Arrow nudges; they moved by the layer-time distance).
- Shift+click toggles a key in or out of the selection (`keys.select {toggle}`); Ctrl+click
switches Linear ↔ Auto Bezier and Ctrl+Alt+click toggles Hold.
- Keys of locked layers can't be selected (clicks, marquee, Select All Keyframes), so nothing
edits them.
- Toggle Hold off returns an eased key to Bezier (it always restored Linear).
- Shift-dragging a key snaps to everything the current time snaps to — other keys, layer in /
out points, layer and comp markers, the work area — and a tooltip shows the key's time and how
many frames it moved.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each reading starts PowerShell on Windows (in the background since the previous commit); the
cache budgets only need to follow memory pressure, not track it closely.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The windowing layer delivers Ctrl+C, Ctrl+X and Ctrl+V as clipboard events instead of key
presses, and the shortcut dispatcher only read key presses: in the app, copying and pasting
keyframes, layers and effects with the keyboard did nothing (tests passed because they inject
key presses). Clipboard events now map back to their keys with the modifiers held (so Ctrl+Alt+C
and the other variants work too), except while text is edited in the viewer, which handles them
itself. Ctrl+V sends no event while the system clipboard is empty, so copying now also puts a
line naming what was copied on it ("EffectCraft: 3 keyframes").
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Dragging a keyframe moved it a frame or two and then let go: the key's widget id includes its
time, so after the first move it was a new widget and egui ended the drag. The Timeline now
follows the whole gesture itself (`KeyDrag`: from where the button went down, by whole frames,
every selected key, one undo step), like After Effects. Holding Shift snaps the grabbed key to
the current time, the work area and composition markers.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every 10 seconds the UI thread re-read the system's memory for the cache budgets, and on
Windows that starts PowerShell (`Get-CimInstance Win32_OperatingSystem`) and waits for it:
about a second, sometimes two, during which the whole app froze. The reading now runs on a
background thread (`sysinfo::MemoryWatch`); `Session::memory_tick` takes the last reading and
starts the next one, never waiting. Console programs started for it no longer open a console
window (`CREATE_NO_WINDOW`).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`Project::same_content` ignored `next_id`, so an edit that only took an id was dropped as a
no-op and the id could be handed out again (an edit may keep it beyond the project).
`guard::tests::the_session_survives_a_panicking_edit`, merged after M3.9, failed on main for
this reason. Ids now count; setting a value to what it already is still records nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The web app reopens the last visit's project and, when it had unsaved changes, marked it
modified with `bump()`. Since M3.9 the modified mark follows the saved snapshot, and a bump on
the very project just opened counts as "back at the saved state", so restored changes looked
saved (no `*`, no unsaved-changes prompt). `Session::mark_unsaved` forgets the saved snapshot
instead: the project stays modified until saved, whatever is undone. Regression test included.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Following AGENTS.md's Never crash rules for the M3 / M4 additions:
- Cache Frames When Idle panicked (`clamp` with min > max) when the work area was inverted
(start set past the end); an empty work area now has nothing to cache.
- `file.newCompFromSelection`: the sequence overlap times the layer count overflowed for a huge
`overlapDuration` (wrapping to a wrong comp length without overflow checks); negative or
non-finite values fall back to 1 s and the product saturates. The lead item is read with
`get()` instead of indexing.
- Layer Settings' enablement reads the first selected layer with `first()`.
- Regression tests for both (each failed before the fix).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`Renderer::mb_on` (render options × the comp's switch × the layer's switch as the precomp
layers above limit it) now decides everywhere:
- Mask motion blur checked only the comp's switch and the layer's own switch, so Render
Settings ▸ Motion Blur ▸ Off for All Layers and a precomp layer with motion blur off still
blurred masks; it now takes the gate (`masks::MaskBlur`) and the comp's Samples Per Frame
(four in Draft) instead of a fixed eight.
- Effects that read the shutter (`EffectEnv::shutter`) get it only when the layer is motion
blurred in this render.
- Advanced 3D layers follow the precomp layers' Motion Blur switches too.
- The layer cache key of a layer with masks or effects includes whether it is blurred and the
comp's shutter and samples, so a buffer blurred for the viewer is never reused by a render
with motion blur off.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
These crates have no production panic sites: add the deny attribute so they stay clean
(geom, color, model, automation, host, export, plugin, and the desktop, CLI and web apps). The
plugin's wasm-only test module gets a separate cfg(test) attribute so clippy recognises it as
test code.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ui-egui without expect()/unreachable!(): the frame pool is optional (rayon's global pool if the
OS refuses threads) and gets a panic_handler, because a panic in a rayon spawn() job aborts the
whole process by default; render-queue progress, the text-edit anchor and the GPU viewer
texture are bound where they're checked; the settings panel's Bool arm no longer panics.
Denies the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A precomp layer trimmed out (or time-remapped) past its nested comp's end kept drawing the
nested layers whose bars ran on. Like After Effects, the nested comp now shows nothing outside
its own span (`Comp::covers`), flattened, through a proxy or with Collapse Transformations, on
the CPU and the GPU (both go through `Renderer::source` / `collapse_into`).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Composition Settings ▸ Advanced has both options (they were missing from the model, the
renderer and the dialog):
- Preserve frame rate: nested in another comp, the comp shows only its own frames
(`EvalCtx::nested_time` floors the precomp layer's source time to the nested comp's frame
rate; used for the nested frame, its proxy, aux channels, collapsed transformations and the
viewer's internal wireframes), and the Render Queue renders it at its own rate whatever the
Frame Rate override (`RenderSettings::rate`).
- Preserve resolution: nested in a comp rendered at Half / Third / Quarter, it renders at full
size and the buffer's scale places it.
- `comp.settings {preserveFrameRate, preserveResolution}`, `comp.info` reports them, scripts
read and set `CompItem.preserveNestedFrameRate` / `preserveNestedResolution`, Pre-compose
copies them, and the dialog registers `dialog.comp.preserveFrameRate` /
`dialog.comp.preserveResolution`.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
In EPS/PostScript, `0#7` asked i64::from_str_radix for radix 0, which panics (found by
fuzzing). Radixes outside 2..=36 are not numbers now.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
render: the debug-only quantisation check uses debug_assert! instead of panic!. gpu: the
test-only panic on a wgpu validation error is cfg(test) with a narrow allow, and release
builds only log. Test-only unreachable!() calls become panic!(). Denies the panic lints in both
crates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Tick arithmetic saturates, seconds-to-tick conversion clamps to Tick::MIN..=Tick::MAX (NaN is
zero), zero rates/timebases no longer divide by zero, and timecode / feet+frames parsing and
formatting use checked or saturating math. `layer.timing {"delta": -1e308}` over the control
channel overflowed FrameRate::snap_nearest (found by fuzzing every command), and a timecode
field holding 99999999999999:00:00:00 overflowed parse_timecode. Denies the panic lints.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Effects without expect(): nested parameter groups are looked up with ? (falling back to the
effect's top level), the OCIO effects use named constants for the built-in ACES spaces instead
of expect()ing a table lookup, curve fitting and Path Text no longer expect on just-pushed
vectors. Test-only unreachable!() becomes panic!(). Denies the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
roxmltree recurses per element and overflowed a 2 MiB stack at a few thousand nested
elements (aborting the app). parse() now scans the element depth first and returns
Error::Xml beyond 256 levels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
FCP 7 XML / FCPXML nested about 1 000 elements deep overflowed an 8 MiB stack in the XML
parser (aborting the app). Import now scans XML and OTIO JSON nesting first and returns
Error::TooDeep beyond 256 levels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The descriptor depth limit restarted at 0 inside every nested descriptor (Objc / GlbO /
ObAr), so a crafted file of 100 000 nested descriptors (2.4 MB) recursed until the stack
overflowed and aborted the app. The depth now carries through nested descriptors.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Every purge emptied the viewers' RAM preview; Purge ▸ All Disk Cache and Snapshot now leave
it (Image Cache Memory, All Memory, All Memory & Disk Cache, All and All 3D Cache still
clear it).
- The memory purges also drop the decoded footage frames (`FootageSource::purge`, the media
pool's `clear_frames`; decoders stay open), so footage reads decode again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Composition ▸ Preview ▸ Cache Frames When Idle was only a saved checkbox. After a second
without input or edits it now renders the work area's frames (from the current time on, then
from the start) into the RAM preview, a few at a time behind the viewer's own frame, until the
work area is cached or the RAM preview budget is full.
Docs: preferences.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Viewer Resolution Auto rendered at half its scale while a preview played, so frames cached
while scrubbing were not played and frames cached by the preview were re-rendered once it
stopped. Auto now follows the zoom alike playing and paused (Fast Previews still lowers it
while dragging).
- Cache Before Playback waited for every frame of the range: a range larger than the RAM
preview budget evicted its first frames and never started. The range is now cut to the
frames that fit (90 % of Settings ▸ Memory & CPU's budget) with a message, then cached and
played.
- The Info panel shows the frame rate a playing preview achieves and whether it is real time
(`Playback::achieved_fps` / `real_time`, automation id `info.previewFps`).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- `FrameKey` carries a hash of the render options (`frames::opts_hash`): switching Fast
Previews (Draft / Fast Draft) or Realtime Shadows no longer shows frames rendered the other
way.
- Eviction drops the least recently shown frame (`Frames::get` touches it), not the oldest
inserted: a looping preview a little over budget no longer evicts the frame it shows next.
- Frames of earlier project revisions are dropped on every change (`Frames::drop_stale`); they
could never be shown again but held memory (and GPU textures) until evicted.
- The green cache bars (Timeline, Preview panel) count the frames of exactly what the viewer
shows — scale, 3D view / region of interest and options (`EffectcraftApp::shown_series`) —
instead of any frame at that scale.
- Playback loops compute the key's fixed part once (`EffectcraftApp::frame_series`).
- Docs: architecture.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
raster: the inpainting pyramid stops at a missing level, and video inpainting falls back to a
plain patch fill if the previous frame is missing, instead of expect(). path: curve fitting of
an empty point list returns a degenerate cubic instead of indexing/unwrapping, and Merge Paths
no longer has an unreachable!() arm. Denies the panic lints in both crates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Removes the engine's production expect() calls: background-job polling (camera, mask, roto,
point and warp tracking, render queue) uses let-else / if-let instead of take().expect("job"),
command helpers bind what they checked instead of re-reading it with expect, the menu bar falls
back to an empty tree (logged) instead of expect()ing that a constant parses, and
autotrace/text animator group lookups no longer expect a just-inserted group. perf.rs, the
benchmark harness behind `effectcraft-cli bench --ops`, opts out with a module-level allow.
Test-only unreachable!() calls become panic!() (clippy.toml allows panics in tests). Denies the
panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Text layout and font resolution without expect(): an empty line list keeps the default bounds,
the line breaker stops instead of expect()ing a checked option, and font resolution falls back
to the first bundled face. The one remaining expect (the bundled-font database is never empty)
carries the documented narrow allow. Denies the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Session::execute runs every command (UI, CLI, control channel, MCP, scripts; file import,
export and save are commands too) through guard::guarded, which catch_unwinds a panic that
escapes the command and returns EngineError::Other("command `x` failed with an internal error
(...); it was stopped and your project is still open"). Edits work on a copy of the project, so
a command that panics mid-edit leaves the project untouched. The desktop app also installs a
panic hook that logs every panic through the app logger (then runs the default hook).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Scripts (Script Console, effectcraft-cli script, the MCP run_script tool) are parsed by Boa,
which recurses per nesting level and overflows the stack (aborting the app) at about 200
levels on an 8 MiB thread. run() now rejects a script nested deeper than MAX_NESTING (96)
brackets with a ScriptError before parsing; strings, template literals and comments are
skipped. Also replaces expect("console context") with an error, and denies the panic lints.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The camera solver, KLT pyramid and subspace (warp stabilizer) tracker expect()ed data-structure
invariants: a track point seen in a frame, a solved camera, a pyramid level. A missing value now
skips that point/track (or, for finish(), returns a SolveError: "no frame was solved") instead of
panicking. Denies the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Lottie stores line cap, line join, merge mode and some text-selector enums 1-based, and the
importer subtracted 1 from whatever number the file held: -1e308 (rounded to i64::MIN)
overflowed, a panic in debug builds (found by structural fuzzing). Use saturating_sub. Denies
the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
opusenc: the SILK state is used through if-let / let-else instead of expect("SILK state").
vp9enc: the motion-search SAD and inter prediction return early without a reference frame
instead of expect("reference"). Denies the panic lints in both crates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Replaces the encoder's 8 expect()/unwrap()/unreachable!() calls with handling that can't
panic: coding-unit decisions without a candidate report an infinite cost, inter helpers
without a reference picture return a neutral result, residual_coding of an all-zero block
writes nothing, the impossible Intra arms are no-ops, and the edge padding of an empty row uses
0 instead of unwrapping. Denies the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Replaces the encoder's 11 expect()/unreachable!() calls on internal invariants with handling
that can't panic: the rate-control state is passed in rather than re-borrowed, inter decisions
without a reference frame fall back to intra, per-plane results are bound directly, and the
writer's impossible branches become debug_asserts. Denies the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An expression nested about 60 brackets deep overflowed the stack of a 2 MiB thread in the
JavaScript parser (release and debug; ~200 levels on 8 MiB), which aborts the whole app: it
can't be caught. The array-maths rewrite also nests one call per arithmetic operator, so a long
sum went the same way, and the rewriter itself recursed per bracket. rewrite() now measures the
nesting (brackets plus chains of prefix operators) before and after rewriting and turns an
expression deeper than MAX_NESTING (32) into a RangeError; check_syntax reports the same error.
Also replaces the unreachable!() in the bracket-tree builder with a let-chain, and denies the
panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- `Project::to_file_json` writes `savedBy` (the EffectCraft version) ahead of the project and
checks that the text reads back before anything is written: JSON has no NaN or infinity, so
such a value made a file that saved but could never be opened. Save now fails with a message
and the file on disk is left as it was.
- Opening a project saved by a newer EffectCraft warns that settings this version doesn't know
are lost on saving; `file.open` reports `savedBy`. A file that isn't a project names itself in
the error.
- Increment and Save keeps the extension (`.ecprojx` stays XML); Save a Copy writes XML for
`.ecprojx` / `.xml` paths, like Save As.
- Docs: architecture (document model), parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Layer::motion_trackers_mut expect()ed the group it had just inserted; it returns an Option
now and track.motion reports an error if the group is somehow missing. Denies the panic
lints in the crate (the project model behind .ecproj files).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Group Shapes (Ctrl+G) / Ungroup Shapes (Ctrl+Shift+G) sit in the Layer menu itself, as in
After Effects (they were under Mask and Shape Path).
- Window ▸ Learn opens the Home screen's Learn tab.
- Composition ▸ Composition Flowchart is Ctrl+Shift+F11 and Window ▸ Flowchart Ctrl+F11 (the
flowchart had the panel's shortcut and the panel none).
- Layer ▸ Transform ▸ Center In View shows Ctrl+Home.
- "Reveal in Finder" reads "Reveal in Explorer" on Windows and "Reveal in File Manager" on
Linux (`menus::platform_label`).
- Docs: parity (View ▸ New Viewer noted as not yet).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Pre-compose makes the new composition with the original's pixel aspect, background, motion
blur and frame blending settings and 3D renderer (`Comp::nested_like`); it kept only size,
rate and background.
- File ▸ New Comp from Selection records one undo step (two per item before) and takes each
item's pixel aspect. Several items selected open the New Composition from Selection dialog:
Single / Multiple Compositions, Use Dimensions From, Still Duration (default from Settings ▸
Import ▸ Still Footage), Add to Render Queue, Sequence Layers with Overlap, Duration and
Transition. `file.newCompFromSelection {single, dimensionsFrom, duration, addToRenderQueue,
sequence, overlap, overlapDuration, transition}`; `layer.addItem {duration}` for stills.
- Docs: parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Locked layers can't be selected (Timeline click, Select All, Ctrl+Up/Down; shy layers while
hidden too) and are left alone by Clear/Delete, Arrange, Layer ▸ Transform and Center Anchor
Point, which fail when every target is locked (`commands::unlocked`, `commands::selectable`).
- Deleting a parent unparents its children where they are, in the same undo step: the
pick-whip compensation moved out of `layer.setParent` into `layer::parent_fixes` /
`ParentFix::apply`, which now also handles separated X/Y Position.
- Bring Forward / Send Backward move each selected layer one step (a non-contiguous selection
was squeezed into one block).
- Center In View / Fit to Comp keep a 3D layer's Z (and Fit keeps Scale/Anchor Z); Reset /
Center / Fit / Center Anchor Point work with separated dimensions (`layer::update_position`).
- Duplicate and paste keep parents and track mattes, following the copies when both were
copied; paste dropped them always.
- Docs: agents, parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Layer bounds come straight from the file as four i32s. Rect::width/height subtracted them in
i32 (overflow: a panic in debug builds, garbage sizes in release), and layer_pixels allocated
width x height pixels before reading any data, so a 100-byte file claiming a 2^31-square layer
hit "capacity overflow" (or an out-of-memory abort). Widths are now computed in i64 and
clamped, every decoded image is capped at 2^28 pixels (Error::Unsupported beyond that), zlib
data is inflated with a limit of the expected size, RLE count tables and UTF-16 strings
reserve no more than the remaining bytes, and the per-pixel copy only visits the part of the
layer that lands in the output. Denies the panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Layer ▸ Layer Settings on a solid or adjustment layer opens Solid Settings on the layer's
solid (it ran an empty command before): name (renames the layer and the solid), size, Pixel
Aspect Ratio, colour, and "Affect all layers that use this solid". It starts off, as in After
Effects, so changing a solid other layers share gives this layer its own copy.
`layer.settings {pixelAspect?, affectAll?}` → `{layer, solid?}`; `layer.newSolid {pixelAspect?}`.
- Nulls ask for their name; text and shape layers have no settings (the entry is disabled).
- Composition Settings / New Composition and Solid Settings register an automation id for
every control (`dialog.comp.*`, `dialog.solid.*`). Composition Settings takes any typed frame
rate besides the list, and New Composition starts from the last settings used.
- Colour buttons in both dialogs show the stored sRGB colour (`widgets::srgb_color_button`);
egui's linear picker showed it lighter.
- Docs: agents, parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The smart-object branch of the layered-file decoder bound the footage layer twice and
re-read it with expect("embedded implies a layer"); bind it once with a let-chain. Denies the
panic lints in the crate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The s, b and b* operators closed the path even when there was no current point, which kurbo
asserts against (a panic in debug builds; found by fuzzing). Close only when a subpath is open.
CFF charstrings that draw before their first rmoveto now start the subpath at the current
point, as Type 1 charstrings already did. Denies the panic lints in the crate (it had no panic
sites).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Path data like "M0 0 L10 0 Z9" repeated Z implicitly for the stray number; Z consumes no
numbers, so the parser looped forever appending close-paths until memory ran out (a 1-byte
change to a valid SVG used 24 GB and was killed). A number after Z is a syntax error, so
parsing now stops there and keeps what was read, as the spec asks. Also denies the panic lints
in the crate (it had no panic sites).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Removes the six expect() calls from timeline export/import and denies the panic lints in the
crate. A precomp layer whose source item is not a composition (a damaged project, or an item
that was deleted) made export panic in Exporter::sequence; it now leaves that layer out.
Exporter::sequence and track_item return Option, the Motion/Opacity/Volume effect defaults
come from effect_or_new (None if FilmCraft's registry lacks one), and import's motion() skips
a layer without a transform group.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Quit and the window's close button, File ▸ New Project, Open Project, Open Recent, Close
Project, the demo project and a Home template show Save / Don't Save / Cancel when the
project has unsaved changes (Save asks for a path when untitled; cancelling that keeps the
project open). File ▸ Revert asks to discard. Before, all of them dropped unsaved work
without asking.
- The window title shows the project's path and `*` while it is modified.
- The modified mark follows undo: the session keeps the snapshot it last opened or saved
(`Session::saved_project`), so undoing, jumping in the History panel or rolling back a batch
to the saved state clears it.
- An edit that changes nothing (a value set to what it already is) records no undo step and
doesn't mark the project modified (`Project::same_content`, comparing untouched comps by
pointer).
- Agents' `engine.execute` / MCP calls are never asked; `ui.menu.invoke` shows the prompt like a
click (automation ids `dialog.unsaved.save` / `dontSave` / `revert` / `cancel`), and the
control channel's `app.quit {force: true}` skips it.
- Docs: architecture, agents, control protocol, parity.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The getters were registered as keySpatialInTangent / keySpatialOutTangent,
so scripts using the documented After Effects names failed with
"TypeError: not a callable function" (setSpatialTangentsAtKey and the
other keySpatial* methods already use AE's names).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
segment_value treated any non-Linear in-side of the next key as Bezier,
so a key set to Hold (KeyframeInterpolationType.HOLD, add_keyframe
"interpolation": "hold", or the Keyframe Interpolation dialog, which set
both sides) turned the linear segment into it into an ease-in with the
default ease: 0 -> 100 over 2 s read 27.0 instead of 25 at 0.5 s. Hold
only affects what follows the key, which the out side already handles.
The Graph Editor and the Lottie exporter already treat a Hold in-side as
linear, so exports didn't match the preview either.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Off macOS `Cmd` maps to Ctrl, so Enter Full Screen's Ctrl+Cmd+F parsed
to the same keys as Find's Cmd+F and both commands were bound to Ctrl+F
(ui_and_menu_shortcuts_do_not_collide fails on Linux and Windows).
Shortcuts that use both Ctrl and Cmd can't be typed distinctly there,
so they are skipped when the key bindings are built off macOS.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
warp.status solved the stabilization plan (Subspace Warp eigen-decompositions and mesh fits)
on the page the first time it saw an analysis: ~4.9 s blocking the browser's event loop in
the smoke test. The analysis job now sends the plan summary with its result, and a settings
change has a WarpPlan job solve it (status reports stabilizing meanwhile). Desktop unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
contentFill.generate sends its plan (FillPlan) to a browser job worker:
the layer's frames render in passes on the worker's WebGPU device, the PNG
sequence goes back to browser storage (Services::store_file), and the page
adds the fill layer when the plan returns (WorkerReply::Fill). The smoke
test checks it; docs updated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- GPU in browser job workers: each job worker opens its own WebGPU device
(deferred readbacks); the Render Queue export and the analyses' frame loops
are futures (render::passes, offload::run_request_async) whose frames
render in passes, awaiting the device between them. GPU particles and
Advanced 3D read back under keys too. workers.jobs reports passes and
readbacks; the smoke test measures them.
- Layer buffers in the browser's OPFS disk cache: frame workers back their
layer caches with a PrefetchStore; the page plans what to fetch before a
request from recent misses (LayerPrefetch over the shared DiskIndex).
- Templates embed footage (default on, size-capped with a warning) and
creating from a template extracts it next to the new project / into
browser storage.
- ICC LUT profiles (lut8/lut16/lutAtoB/lutBtoA, from the public ICC spec) in
Simulate Output > My Custom RGB, baked into a 3D LUT for the viewer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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>
Pin exact CPU pixel hashes of the simulation effects refactored into shared
GPU plans in M13.23 (CC Rainfall/Snowfall/Star Burst/Bubbles/Drizzle/Hair/
Mr. Mercury, Caustics, Wave World, Foam, Shatter, Card Dance, Card Wipe):
404 frames over default and non-default settings, four times incl. a seek
backwards, 8 and 32 bpc, full and half resolution. Verified bit-identical
between 7a5a93b (before M13.23) and 808ca70 (after).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
FilmCraft's AV1 decoder now frame-threads by default, so a picture may
come out of a later decode call; the round-trip check wants each
packet's picture from its own call.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Bump the FilmCraft crates from 89cfbf2 to 2c5ba7a (118 commits; no API
changes needed here), whose Matroska reader now decodes the VP9 alpha
stream from BlockAdditions. The compositing QA scenario now checks that
WebM renders round-trip their alpha like ProRes 4444.
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>
Brandon Thomas
midasdf
Dexter
bflatastic
fortesnow
Jonas Neubauer
Jonas Neubauer
storybored379