The Composition viewer ignored drags from other panels. Now, as in After
Effects:
- A Project item dropped on the viewer becomes a layer centred where it
lands (above the selected layer, like any new layer); while dragging, the
viewer is outlined and the layer's frame follows the pointer.
- Media Browser files do the same; files dropped from the OS file manager
do too when the platform reports where the pointer is during the drag
(otherwise they import as before).
- An effect from Effects & Presets goes on the layer under the pointer,
which is outlined while dragging.
layer.addItem and layer.newModel take `position` ([x, y] comp px).
file.import takes `addToComp` with `time`, `index` and `position`;
mediaBrowser.import shares it (file::add_to_comp): layered files go in as
their comps, the comp they were dropped in stays open, and items that
can't be layers are reported in `errors`. viewer::layer_at is the hit
test pick() and the drop share.
Tests: ui_drag_drop.rs (project item and effect dropped on the viewer)
and imported_files_go_into_the_comp_where_they_were_dropped (engine).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A Project item (or Media Browser file) dropped in the Timeline always
went above the selected layer, starting where new layers start.
Now, as in After Effects, a line shows where it goes in the stack (above
the first layer row whose middle is below the pointer, else below the
last), and over the time graph a marker with a timecode tag shows its In
point, which follows the pointer (frame-snapped) or, with Shift, the
current time. The layer reorder drag shares the insertion logic.
layer.addItem and layer.newModel take `index` (1-based stack position);
mediaBrowser.import passes `time` and `index` on, one layer under the
other.
Tests: project_items_land_where_they_are_dropped_in_the_timeline
(ui_drag_drop.rs, real pointer drags: outline, time graph, Shift, below
the last layer) and add_item_at_a_stack_index_and_time (engine).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
egui takes i.modifiers from ModifiersChanged events, not from the modifiers
on a pointer event, so a control-channel Shift-click arrived as a plain click
(e.g. Shift-click in the Timeline replaced the selection instead of adding).
ui.click and ui.drag now queue ModifiersChanged before the gesture and clear
it after. The synthetic-input feeding moves into EffectcraftApp::feed_synthetic
so it can be tested; the new tests fail without the fix.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The desktop app shares egui-wgpu's device, which had no uncaptured-error
handler, so wgpu's default one panicked on an out-of-memory error on
every frame thread; the panicking frame stayed "in progress" forever and
never rendered again. The GPU frame budget was a fixed 1 GiB.
- GpuContext::new logs uncaptured errors once its pipelines are built
(errors while building them still panic there, which the app catches
to fall back to CPU compositing).
- Gpu::within_memory runs GPU work in an out-of-memory error scope; on an
error the result is discarded and uploaded layers and pooled textures
are freed.
- Frame workers render each frame within it; out of memory, the frame
renders on the CPU and the RAM preview keeps half as many GPU frames
(down to 64 MiB), since wgpu can't tell how much video memory there is.
- A frame whose render panics renders again on the CPU, or is released,
so the viewer asks for it again.
Tests: a_frame_whose_render_panics_is_released (failed before),
running_out_of_video_memory_halves_the_gpu_frame_budget and
within_memory_returns_the_work_when_there_is_room.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Commit c9d5e733 (M13.4, hevcenc deblocking) committed 207 files under
crates/hevcenc/target/agent-codec-hevc/: an agent's CARGO_TARGET_DIR with
rlibs, rmetas, build-script outputs and .rustc_info.json (~36 MB). They
were tracked because .gitignore only ignored the root `/target`.
- git rm --cached the directory; nothing references it.
- .gitignore: `/target` -> `target/`, so a per-crate or per-agent target
dir is ignored anywhere in the tree. No tracked file lives under a
directory named `target`, so nothing else becomes ignored.
The blobs stay in history; this keeps them out of every new checkout and
stops the next agent run from committing more.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Toggling a layer's Audio, Lock or Shy switch (or Hide Shy Layers) threw
away every cached frame: the RAM preview identity hashes each comp's
address, which any edit replaces, and the layer and disk cache keys
hashed every switch. None of these change pixels.
- Switches::pixels, Layer::same_pixels and Comp::same_pixels (exhaustive
field lists, so a new field must be considered) and Comp::pixel_form.
- Layer cache keys hash Switches::pixels; the disk cache's comp content
key hashes Comp::pixel_form.
- Frames::identity: when a new revision's comps draw the same pixels as
those of the comp's last new identity, that identity (and the project
holding its comps' addresses) is kept, so the frames stay, also when the
switch is toggled back.
With audio on, playback skipped frames that weren't rendered yet. Now the
sound waits: until the frames ahead (to the end of the preview, or what
fits in the RAM preview) are cached, every frame shows as it renders,
silently, like a preview without audio; then the sound starts in sync.
Reaching an uncached frame stops the sound instead of skipping frames.
Tests: audio_lock_and_shy_switches_keep_the_frames (frames.rs, nested
comps and toggling back), audio_lock_and_shy_keep_cached_layers (engine),
the disk key in tests_disk_cache, and
preview_with_audio_shows_every_frame_and_sounds_once_cached.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An angle's revolutions were painted text: only the degrees could be
dragged or typed, so full turns needed the dial or typing past 360.
fx_widgets::angle_field draws `0x+45.0°` as two hot numbers: the
revolutions (widgets::hot_int_at, ten pixels a turn, fractions adding up
between frames so slow drags still step; a typed value is truncated) and
the degrees, each keeping the other. Effect Controls, the Timeline and
the Properties panel use it, so they read and edit angles alike (the
Timeline showed `-1x-40.0°` and Properties had its own split).
Automation ids: timeline.prop.<uid>.revolutions,
effectControls.prop.<uid>.revolutions, properties.prop.<uid>.revolutions.
Regression test: angle_revolutions_scrub_and_take_typing drags and types
the revolutions in all three panels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Choosing Draft Settings (or any entry) in a Render Queue item's Render
Settings menu did nothing when the menu had been moved up to fit the
window. egui hit-tests a fixed-position Area where it was asked to go,
not where it was moved, so the popup's "pressed outside" check was true
over its own entries and the menu closed on the press, before the click.
widgets::pressed_outside tests the press against the rect the popup was
drawn at. Every popup with the same check uses it: popup_menu (Render
Queue, Effect Controls, dropdowns), the colour popup, the Preview and
viewer popups, the panel menu, the Flowchart, OpenType and font menus.
Regression test: ui_render_queue.rs picks Draft Settings from a menu
moved up to fit a 1000 px window; it failed before the fix.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Use checked hex slices so prefs.set follows the existing invalid-colour fallback when a component splits a UTF-8 character. Cover invalid Unicode placements and valid mixed-case colours through the command.
Co-Authored-By: OpenAI Codex <noreply@openai.com>
effect.apply / effect.applyLast run through the menu dispatcher (Effect
menu, Last Effect, an effect dropped on a Timeline layer or on Effect
Controls) now bring up Effect Controls, opened if closed and brought to
the front without taking the focus, on the layer the effect went to
(selecting it if it was dropped on an unselected layer, keeping the new
effect selected). Double-clicking an effect's name in the Timeline opens
it in Effect Controls (layer.select, prop.select, window.panel) instead
of twirling it, as in After Effects.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 04bd305b9dccbb34d8c70fe870513208cdae0b7b)
Opening a comp (Project double-click, Open Composition, a Timeline tab)
now reopens the Composition panel if it was closed and brings it (and
the Timeline) to the front, as in After Effects. A closed Composition
panel also reopens in the centre (with the Layer panel or other viewers,
else above the Timeline) instead of beside Effects & Presets, which also
fixes Window > Composition.
App::raise_panel opens/fronts a panel without moving the focus;
show_panel builds on it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit adf2c6dac78ce426ea0bf6b4199b7238d8ad99dd)
A Windows user reported that the Properties panel's font menu offered only the bundled Inter,
JetBrains Mono and Noto Serif. The system font folders were scanned only when a project asked for
a family the bundled fonts lack, so until then every font menu (Properties, Character, Essential
Graphics) and `file.findMissing {what: fonts}` saw the bundled families alone, on every
platform. Listing the families now scans the system fonts first, and the desktop app starts that
scan on a background thread at launch so the menus are complete when first opened.
The style menus in the Properties and Character panels list the chosen family's own styles
instead of a fixed Regular / Medium / SemiBold / Bold / Italic. Both menus build their lists only
while open.
New query for agents: `text.fonts {query?, rescan?}` (MCP tool `list_fonts`) lists every family
with its styles, origin (bundled / system / user) and own-language name; `rescan` picks up fonts
installed since launch.
Tests: a fresh-process test that the family list includes every family in the platform's font
folders without a prior scan (failed before), and `text.fonts` filtering, rescan and per-family
styles, and the `list_fonts` MCP tool.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019EjTYdt76hTG4Kh2CJFnCr
A selected pin was drawn only 1 px bigger and slightly brighter. Now, like
selected and unselected mask vertices in the viewer, a selected pin is
filled with its kind colour and ringed in white, and an unselected pin is
hollow (dark centre, coloured outline). Drawing only; hit-testing is
unchanged.
Regression test: selected_puppet_pins_are_filled_and_unselected_hollow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 103b5c41e9e8ea84008593e05af50b01a2800b3e)
Clicking an effect's name selected it (prop.select with the effect's group
uid) but group rows never painted the selection, so there was no feedback.
Group rows (effects, masks, shape groups...) now fill with row_selected when
selected, exactly like selected property rows.
Regression test: timeline_selected_effect_row_highlights.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 8b140428b39a24c3e1207ffde0430e759e316112)
- Project panel Clear (#95's focus routing) goes through invoke, so deleting
items that compositions use still asks first (M3.16).
- ui_project_delete: let a dialog finish egui::Modal's sizing frame before
clicking its buttons (#95 moved dialogs to egui::Modal).
- #71's i18n / theme tests clear the frame's texture delta; egui asserts on
unhandled deltas in debug builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rust 1.99 renamed AtomicU32::fetch_update to try_update; try_update is not
available on the workspace's minimum Rust (1.95).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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>
Dexter