[jrusso1020]

What

Two fixes to page motion blur (captureMotionBlur), both of which made elements flicker during playback and in full renders:

1. A blurred element is now copied with everything it paints, not just its layout box.
2. SVG resources (defs, clipPath, mask, pattern, marker, symbol) stay visible while the moving elements are captured alone.

Why

With data-motion-blur="on", an endcard alternated between the full "HyperFrames Studio" lockup and a cut "HyperFra" with no logo. The composition's state was correct at every frame; only the blurred frames were drawn wrong.

• Cut text. readMoverFrames sized each copy as offsetWidth/offsetHeight plus the glow pad. #brand is 540px wide but its one-line wordmark is 1528px, so blurred frames kept only "HyperFra". Sharp frames were fine, hence the alternation. The filter fallback crops to the same box, so it is affected too.
• Vanishing logo. To capture the moving elements alone, the isolation rule hides every other element, including the <clipPath> a logo uses through clip-path: url(#...). A hidden clip path clips everything, so the logo captured as nothing.

On v0.9.7, a full render of that endcard had the wordmark cut in 43 of 91 frames and the logo missing in 31 of 44. With --page-motion-blur=off both were 0.

Related work

Refs #7060 (a different motion-blur fix, for WebGL canvases).

How

• frames.ts: contentBox() grows the mover's box to every descendant's client rect, mapped back into the mover's own px through the inverse of its pose. Elements that clip on both axes are left alone, and at most 256 descendants are read per mover. The frame now carries an origin, so the box can start before the element's own origin.
• filterPlan.ts: the filter region is offset by that origin.
• index.ts: the hide rule excludes SVG resources and their contents, and a second rule sets them visible, so they do not inherit hidden from their <svg>.

Test plan

• Unit tests added/updated
  • frames.test.ts: content wider than its box is included (fails on v0.9.7: expected 540 ≥ 1528); content that stays inside keeps the box.
  • index.test.ts: the isolation rules leave clip paths and their shapes unhidden (fails on v0.9.7).
  • All 192 motion-blur tests pass; typecheck, oxlint and fallow are clean.
• Manual testing performed on the affected project in headless Chrome with html-in-canvas:
  • Playback, capture tier: 0 cut frames, and the logo is present in blurred frames.
  • Filter tier forced: 0 of 86 bad frames (65 of 107 before).
  • Full render with blur on: 0 bad frames for both the wordmark and the logo.
• Documentation updated (if applicable)
• Comments follow CONTRIBUTING.md "Comments"

🤖 Generated with Claude Code

[github-actions]

Edit accuracy: accurate 267 (base branch 267), smooth 119 of those

The gate passes.
Smoothness is reported in the artifact, not gated. A case fails only if it fails 2 of 3 runs.

Quarantined, measured but not gated (8)

• seqplainkeys-none-center-r0-nested-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): pass, passes
• seqplainkeys-none-center-r0-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): fail / pass / pass, passes
• seqplainkeys-none-center-r30-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run
• seqplainkeys-none-pct-r0-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): fail / pass / pass, passes
• seqplainkeys-none-pct-r30-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run
• seqplainkeys-none-px-r0-nested-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): pass, passes
• seqplainkeys-none-px-r30-nested-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run
• seqplainkeys-none-px-r30-root-z100 (fixed by studio: undo/redo on a keyframed plain element sometimes leaves the box 230-400px off (edit-accuracy seqplainkeys-none-*) #6989): not run

Unstable (1)

• seqplainkeys-none-px-r0-root-z100: tracking 0.06, pressJump 0, drop 0, reload 0, render 0.03, renderKey -, undo true, teleport true / tracking 0.06, pressJump 0, drop 0, reload 0, render 0.03, renderKey -, undo true, teleport true / tracking 0.06, pressJump 0, drop 0, reload 0, render 0.03, renderKey -, undo true, teleport true

[heygengenesis]

Verdict: APPROVE · 0 blocker · 0 important · 0 nits · f755566

I reproduced the containing-block case. The swap now adds will-change: transform, so an absolute child still resolves against the mover while it is measured. A mover that is turned only by an ancestor skips the swap and is measured with the upright box mapped back, which covers more than it needs to but never cuts anything off. Both new tests fail on 88b2de1 and pass here. All 197 motion-blur tests pass locally, and every workflow is green at this head.

Scope

I read all of frames.ts, filterPlan.ts and index.ts (the SVG-resource isolation rules), and the filterTier.ts/composite.ts wiring. rectsAtStyle has the same pre-existing exposure; the author has explicitly left it out of scope, and it is fine as a follow-up.

— Genesis