run_tracker follows the object with the model's video tracker on the same
synthetic clips (--sam2 FILE or the installed weights). 12 frames, dev build:
re-entry clip mean IoU 0.632 classical vs 0.975 SAM 2 (0.000 vs 0.967 after
the object returns); the other clips 0.98 vs 0.99.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
When the chosen model has a tracker, roto.propagate runs one tracking pass over
the span: prompts from the strokes (a handful of points, which SAM 2 trusts),
the clip tracked both ways, each frame's probability map refined by the same
graph cut with the strokes as hard constraints. Results are cached under their
own keys with a per-frame confidence and an occluded flag; roto.status reports
them (source: tracker|classical). Without a tracker nothing changes.
Real weights: IoU 0.98 on all 21 frames, confidence 0.99.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
ObjectTracker interface in the segment crate (prompts on some frames, the
clip tracked both ways, each frame streamed with confidence and an occluded
flag), implemented by SAM 2. MaskModel::tracker() is None for other models.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Sam2Mask wraps the image path behind the Roto Brush's MaskModel interface
(cached frame encoding, bilinear mask resize). The engine loads the verified
weights natively (not on wasm). Checked with the real file: install, notice,
load, use by Roto Brush 3.0, deselect.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Apache-2.0 entry pinned to a Hugging Face commit with its SHA-256 and size,
listed in Settings > Roto Brush and installable with the existing verified
install flow. No weights are bundled.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Click on frame 2 of a 7-frame clip with an occlusion, tracked forwards and
backwards: IoU 1.0000 on all 9 frame results, worst mask-logit error 0.0038,
object scores equal to 3 decimals.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Per-object memory banks, prompts on any frame, forward and backward
propagation with streaming results, cancellation, object pointers with
temporal encoding, pruning of memories no later frame can use, and checked
input handling. Hostile-input and cancel tests.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Self and cross attention with axial 2-D RoPE (keys tiled over memory frames,
object-pointer tokens left unrotated). Max abs error 2.6e-5.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Mask down-sampler, feature fuser and projection, sine position encoding,
occlusion embedding and bfloat16 rounding of stored memories. Max abs error
3.5e-6 on the memory features, 6e-8 on the position encoding.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Encode 5.0 s -> 3.4 s per 1024x1024 frame in release on 16 threads (fused
softmax, queries scaled before the matmul, flat 2-D linear layers); a click
on an encoded frame is about 120 ms. Reference tests unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A click on the synthetic image: mask logits within 1.1e-4 (range 11.6), IoU
head 1e-6, object score 2e-5, prompt embeddings 1e-7.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Weights loader with shape/finite validation (S2.10), patch embed, bicubic
positional embedding, window/global attention, query pooling. Reference test
(skips without weights): max abs error 2.6e-5 over all 12 blocks.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Synthetic Roto Brush clips with known mattes (movement, lighting change,
occlusion, leaving and re-entering) and a runner that compares the classical
engine with MobileSAM when installed. Needs no weights; SAM 2 plugs in later.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Selecting a mask group now clears the previous key selection, so Copy/Cut act
on the mask. Pasting masks clears the source mask's vertex selection, so Clear
removes the pasted mask rather than points of the source.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Session::write_output (and refuse_export) is the one place that checks; Save Frame As
(PNG/PSD/EXR), Export Lottie, Save a Copy (+XML), Collect Files, Export Timeline (and its
pre-rendered movies), Export Template, Save Animation Preset, Export Guides and Export
Shortcuts write through it. The control channel's render.frame/ui.screenshot, MCP
render_frame and CLI render-frame --out check first (new query file.checkOutput for the
bridged cases). Save/Save As/auto-save are untouched.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Two queue items that resolve to one output (the same comp added twice, a duplicated item) replaced
each other's file inside one job. Nothing is on disk when the job starts, so no check said anything.
- A default output name (renderQueue.add, render.preRender, comp.saveFrameAs {queue},
file.createProxy) that another item of the queue writes, in any output module or segment, is
numbered like a footage collision: Main.mp4, Main_1.mp4, Main_2.mp4. Sequences are compared by
the frames they write, so two stills of one comp at different times keep their names. Names a
caller gave and names already stored in items are never changed.
- Outputs of one job that still name the same file (explicit names, renderQueue.duplicate, output
modules, segments) refuse the whole render before anything is written, naming both items.
Sequences clash only on frames both write; a later item with Skip Existing Files leaves the
earlier item's frames alone; items refused for footage take no part.
Refusing the whole render, not just the later item: which of two items should keep the file is the
user's to say, and failing only the later one would leave a half-rendered queue found after they
left. renderQueue.duplicate stays a verbatim copy so the clash is reported, not hidden.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RDvxCS8xcR28VArjzn1x7U
#1339 was written before ui-egui's integration tests became modules of one binary
(`tests/it/main.rs`), so its six test files landed at the top of `tests/`, each its own crate
and link. They move into `tests/it/` as modules; the a11y module's doc names the new path.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
edit.copy {"layer": id} copied the selected mask instead of the layer:
the mask branch only checked for `layers`, but `layer` is an explicit
target too.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Edit > Copy with masks selected copied the layer, so Paste duplicated it
instead of adding the mask. Masks now go to their own clipboard and Paste
appends them, with unique names, to the selected unlocked layers.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
#2653 (#2618) changed what an adjustment layer in a mode other than Normal
renders, but its cache keys are unchanged, so frames cached before it keep
showing the Normal result. Bumping the render version keeps old entries
from being reused.
File ▸ Export ▸ Adobe Premiere Pro Project… pre-rendered every layer Premiere can't show to
ProRes 4444 with the window waiting: minutes on a large comp, which Windows reports as not
responding. The export now runs as a job (Window ▸ Progress, cancellable, with per-layer
progress) when started from its dialog (`background: true`); without it, the command still
waits and returns the same result.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
UPDATE_APP gains our accent (#e0368f, our colour in ArtCraft Launcher). The prompt draws it
on the UPDATE chip, the buttons and the What's new underline, and says "by ArtCraft"
(craft-update's default developer); #2646 gave it our icon.
craft-update is pinned to storytold/craft-libs main at 1065653e7 (#434).
A slip on a layer with no source handles succeeded with `delta: 0` and nothing else, which
read like a no-op bug. The result now carries `requested` (seconds) and, per layer, `clamped`
when the source could not move that far.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Timeline and Effect Controls value editors sent every write with the property's scrub merge
key, so two separate picks from a drop-down (or two check box toggles) on the same property
folded into one undo step. Drop-down, check box, layer and mask picks now record their own
step; scrubs and typed numbers still merge as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
