feat: automatic tier, 240°/32 defaults, check finding, and Studio stops opting out by vanceingalls · Pull Request #6108 · heygen-com/hyperframes

vanceingalls · 2026-10-23T03:14:09Z

What

hyperframes check reports motion_blur_filter_fallback (info) when the filter tier draws because capture could not run, with the cause and the way back. The engine's per-session log line gains tier=.
Studio stops sending previewMotionBlur. Every Chrome can preview blur now (the fallback covers browsers without html-in-canvas), so a render no longer has to be switched off to match a preview that could not blur. The server still accepts false from an older Studio.
Automatic tier. The scan notes whether any mover turns or scales (the linear part of its transform chain changes) and whether it already forms its own stacking context while it moves. When every mover only translates, the filter tier draws as well as capture and costs about what no blur costs. Anything that turns or scales, or moves without a transform, stays on capture. data-motion-blur-tier="capture" or "filter" forces a tier. check stays quiet for these two choices.
Defaults match the shutter blur Jake built: 240° (was 180°) and at least 32 samples (more for a long smear).
Docs and skill note describe the two tiers, the attribute, the measured cost and the findings.

Evidence

Cost, 8 s bench (four whips), wall time including startup, two runs each: no blur 5.3–5.9 s, Jake's filter 5.4 s, ours automatic 5.2–5.4 s, capture forced 9.8 s.

Distance from Jake's own render at his settings (PSNR, higher is closer), default settings: whip 51.0 dB (was 34.6), glow 51.2, waterfall 55.7 (was 39.9), WebGL 50.6. Spin 28.2 and crash zoom 29.4 stay apart on purpose: capture streaks along the arc or zoom where his softens the whole element.

The real-Chrome pixel tests run the shared assertions (trail shape, centroid, ink conservation, backdrop untouched, glow carried, centred box) through both tiers.

Behaviour changes to be aware of

New public attribute data-motion-blur-tier.
The automatic choice changes which tier most translation-only compositions use. It has been checked on the eight shots only; the corpus run is still to do. Blur stays opt-in (data-motion-blur="on").
A preview in a browser without html-in-canvas uses the filter tier, so a spin looks softer there than in the render. Force a tier for an exact match.

Rebased onto main after #6106 and #6107 merged.

No visible change

The only Studio edit removes the 8 lines in useRenderQueue.ts that sent previewMotionBlur with a render, and its test. Blur now picks its tier from the page, so nothing on screen moves.

🤖 Generated with Claude Code

heygengenesis

Verdict: APPROVE · 0 blocker · 1 important · 2 nits · 393e017e

chooseTier does what its doc comment says: renderer-off gives off; forced or unavailable capture gives filter when allowed. check stays quiet for the two chosen cases. The optional Comments check is red because the comment-share ratchet fails on tier.ts (8/71, base 5/51). Trim 3 comment lines before this reaches main.

— Genesis (pr-review)

heygengenesis · 2026-10-23T04:38:44Z

+    const sig = linearSignature(el, scene);
+    const first = signatures.get(el);
+    if (first === undefined) signatures.set(el, sig);
+    else if (first !== sig) notPlain.add(el);


important: the linear signature is only compared on a mover's first fast step and every 10th scan step. A turn or zoom that falls between those samples is classified as translation-only, and auto picks the filter tier, which only softens the spin. Repro (Chrome 152, built runtime, 1 s composition): a box slides from 0 to 0.2 s, then spins 360° from 0.4 to 0.6 s. The report is tier: filter, reasonCode: translation-only. Move the spin to 0.3–0.45 s so step 10 lands inside it and the report is tier: capture. Fix: compare at least the element's own transform's linear part on every step it is fast. Those steps already read its rect, and a spin in place changes it. The full ancestor chain can stay on the 10-step cadence.

heygengenesis · 2026-10-23T04:38:44Z

+  /** Who should draw instead of capture: it turned itself off, or the filter tier does as well for less. */
+  function handOff(): CaptureDefect | null {
+    if (disabled) return { code: disabledCode ?? "capture-failures", reason: disabled };
+    if (config.tier === "auto" && reader && translationOnly && filterTierAllowed()) {


nit: a translation-only hand-off scans the whole timeline twice. Capture's discover() decides translationOnly, then createFilterTier rescans from scratch, and with -events="on" every callback fires twice. Now that this is the default path for most compositions, passing the Discovery (movers, pads, dynamic) into createFilterTier would make it one scan.

heygengenesis · 2026-10-23T04:38:44Z

+| `data-motion-blur-shutter` | `240` | Shutter angle in degrees, 0–720. Larger is a longer smear. |
+| `data-motion-blur-samples` | at least 32 | A fixed number of samples per frame, 2–64. Left unset, each mover gets at least 32, and about one per 2 px of smear when that is more, so a fast move never shows steps. |
+| `data-motion-blur-events` | off | `on` runs timeline callbacks (`onUpdate` and similar) while blur looks for movers and at each sample. Needed when the motion is set from a callback, such as a `render(t)` function, which blur cannot otherwise see. |
+| `data-motion-blur-tier` | `auto` | `capture` or `filter` forces one of the two ways blur is drawn (below). |


nit: capture does not strictly force. chooseTier returns filter for tier: "capture" when capture cannot run, and a later scan defect hands off too. So the advice under Cost and limits ("Force one tier … for an exact match") does not hold in a preview browser without HTML in Canvas. Suggest wording such as "capture prefers capture and falls back to filter where it cannot run".

heygengenesis

Verdict: APPROVE — 1 resolved, 3 open, 0 new · 393e017e → b40dc365

Resolved: the Comments ratchet (2 comment lines removed from tier.ts; the check is green). Still open: the earlier inline important and 2 nits. The pixel suite passes locally on this head (16/16). Once #6107 takes the test fix, this needs a rebase and a fresh review.

— Genesis (pr-review)

heygengenesis

Verdict: APPROVE — 0 resolved, 3 open, 0 new · b40dc365 → 4ce56cea

Rebase only: the patch matches the approved head, minus the 2 test assertions that moved down into #6107.

— Genesis (pr-review)

The base branch was changed.

mintlify · 2026-10-23T20:46:27Z

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated
hyperframes	🟢 Ready	View Preview	Oct 23, 2026, 8:52 PM

💡 Tip: Enable Automations to automatically generate PRs for you.

… Studio stops opting out hyperframes check reports motion_blur_filter_fallback (info) with the reason capture could not run and the way back to it. The engine's per-session log line gains tier=, and its pixel test covers the fallback over a WebGL page. Studio no longer sends previewMotionBlur: every Chrome can preview blur now, through the fallback where it lacks html-in-canvas; the server still accepts false from an older Studio. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

…translates The scan now notes whether any mover turns or scales (the linear part of its transform chain changes) and whether it already forms a stacking context while it moves. When every mover only translates, the filter tier draws as well as capture and costs about what no blur costs: on an 8 s bench shot a render took 7.0 s against 9.9 s with capture and 7.3 s with no blur. Anything that turns or scales, or moves without a transform, stays on capture. data-motion-blur-tier="capture" or "filter" forces a tier. The report says which one drew and why (translation-only, tier-attribute); hyperframes check stays quiet for those two, which are choices, not fallbacks. The pixel tests run the shared assertions through both tiers. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

…d in limits The guide described blur turning off for WebGL, a root transform and similar. It now says the filter tier draws instead, how the tier is chosen and forced, what each does with a turn or a zoom, the measured cost of each, and which findings check emits. The skill note drops the off-for-WebGL advice and the 1.7x figure. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

…samples The shutter opens 240 degrees by default (was 180) and each mover gets at least 32 taps (was sized to the smear alone, down to 2), growing with the smear as before. These are the settings the shutter blur Jake built uses; on the card whip the default render is now 51 dB from his (34.6 dB at the old defaults). The guide's attribute table says so. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

github-actions · 2026-10-23T21:16:08Z

Edit accuracy: accurate 271 (base branch 271), smooth 131 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 (0)

Unstable (2)

seqplainkeys-none-center-r0-root-z100: tracking 0.06, pressJump 0, drop 0, reload 0, render 0.03, renderKey -, undo false, 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
seqplainkeys-none-pct-r0-root-z100: tracking 0.06, pressJump 0, drop 0, reload 0, render 0.03, renderKey -, undo false, 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 resolved, 3 open, 0 new · 4ce56cea → 819dff89

Rebase onto main with the same patch, plus 819dff89, which adds the reviewed deletion of useRenderQueueMotionBlur.test.tsx to ALLOWED_DELETIONS. CI is green.

— Genesis (pr-review)

vanceingalls mentioned this pull request Oct 23, 2026

feat(studio): motion blur panel with Off / Subtle / Strong #6112

Merged

This was referenced Oct 23, 2026

fix(core): capture blur strobing, backdrop box, glow drop-out and callback motion #6106

Merged

feat(core): capture blur falls back to per-element filters where capture cannot run #6107

Merged

heygengenesis Bot approved these changes Oct 23, 2026

View reviewed changes

vanceingalls force-pushed the feat/capture-blur-filter-tier branch from 09d41cc to 8e15e47 Compare October 23, 2026 05:41

vanceingalls force-pushed the feat/capture-blur-auto-tier branch from 393e017 to 00442fe Compare October 23, 2026 05:41

heygengenesis Bot approved these changes Oct 23, 2026

View reviewed changes

vanceingalls force-pushed the feat/capture-blur-auto-tier branch from b40dc36 to 4ce56ce Compare October 23, 2026 19:34

heygengenesis Bot previously approved these changes Oct 23, 2026

View reviewed changes

Base automatically changed from feat/capture-blur-filter-tier to main October 23, 2026 20:46

vanceingalls and others added 5 commits October 23, 2026 13:46

chore(core): trim tier.ts comments to the ratchet

0b4c3b8

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

vanceingalls force-pushed the feat/capture-blur-auto-tier branch from 4ce56ce to 0b4c3b8 Compare October 23, 2026 20:46

chore: allow deleting the Studio previewMotionBlur render test

819dff8

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

heygengenesis Bot mentioned this pull request Oct 23, 2026

fix(core): a hidden child of a mover stays out of the blur trail #6155

Merged

heygengenesis Bot approved these changes Oct 23, 2026

View reviewed changes

vanceingalls merged commit 79b57ab into main Oct 23, 2026
126 of 127 checks passed

vanceingalls deleted the feat/capture-blur-auto-tier branch October 23, 2026 21:20

heygengenesis Bot mentioned this pull request Oct 23, 2026

chore: release v0.8.190 #6158

Merged

Navigation Menu

Repository navigation

feat: automatic tier, 240°/32 defaults, check finding, and Studio stops opting out - #6108

feat: automatic tier, 240°/32 defaults, check finding, and Studio stops opting out#6108
vanceingalls merged 6 commits into
mainfrom
feat/capture-blur-auto-tier

vanceingalls commented Oct 23, 2026 •

edited

Loading

Uh oh!

heygengenesis Bot left a comment

Uh oh!

heygengenesis Bot Oct 23, 2026

Uh oh!

heygengenesis Bot Oct 23, 2026

Uh oh!

heygengenesis Bot Oct 23, 2026

Uh oh!

heygengenesis Bot left a comment

Uh oh!

heygengenesis Bot left a comment

Uh oh!

mintlify Bot commented Oct 23, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 23, 2026

Uh oh!

heygengenesis Bot left a comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

vanceingalls commented Oct 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

Evidence

Behaviour changes to be aware of

No visible change

Uh oh!

heygengenesis Bot left a comment

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot Oct 23, 2026

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot Oct 23, 2026

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot Oct 23, 2026

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot left a comment

Choose a reason for hiding this comment

Uh oh!

heygengenesis Bot left a comment

Choose a reason for hiding this comment

Uh oh!

mintlify Bot commented Oct 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Oct 23, 2026

Edit accuracy: accurate 271 (base branch 271), smooth 131 of those

Uh oh!

heygengenesis Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

vanceingalls commented Oct 23, 2026 •

edited

Loading

mintlify Bot commented Oct 23, 2026 •

edited

Loading