A fixed-control scroll mask and an ordinary tutorial cut. Each gets its own builder and a separate critic. Failed revisions remain evidence, not finished work.
Two code/SVG variants rendered. Original master absent; official AV1 media download returned HTTP 403. The retained lower-resolution excerpt misses the first four selected source frames and is not substituted.
Two code/SVG variants rendered. YouTube media returned HTTP 401/403. A separate official-webinar recovery found no exact accessible replay and retrieved no source bytes. Other events are not substitutes.
The fresh seventh-version critique ties both pairs. No defensible concrete defect survives the inspected native checks, but there is no candidate win. Moving-footage and native whole-frame review remain outstanding; another unchanged-output vote would not resolve those gaps.
Watch the comparisons
1 existing treatment refinement is shown separately. It is not a new group and does not increase the 36-group baseline.
1 of 1 playable groups
Existing treatment · Refinement in progress · Not completed
Atlas title: word-separation repair
A seventh local variant of an existing treatment, not a new draft group. It restores wider word breaks with a uniform 18px space advance while retaining the prior timing, glyph widths and original copy. These values are adaptation choices, not recovered source typography.
Official source · brief silent excerptOur code/SVG draft · original content
Paired playback is approximate. Pause and step for exact native-frame correspondence.
The fresh seventh-version critique ties both original-content comparisons. No defensible concrete defect remains in the inspected states, but this is not a candidate win. All 360 paired indexed frames were individually viewed fitted; 86 native-scale regions across 68 frame entries leave native spatial gaps. No native whole-frame or moving-video review is credited. This existing treatment remains under review, not a completed recipe.
Small pieces, choreography and limitations
Source cut pair 506→507, 16.867000→16.900000 s: the browser is replaced by a white title screen already showing five glyphs. The unseen first five glyph additions must not be invented. The preceding frame is context, outside this excerpt.
Source [507, 536), 16.900000–17.867000 s: first-line glyph additions accompany leftward displacement of existing glyphs. The complete first line is visible by frame 532; repeated wording does not imply identical pixels.
Source [536, 564), 17.867000–18.800000 s: the second line begins with three visible glyphs while the first line rises. Further glyph additions and leftward travel overlap; exact settling endpoints are not recovered authoring keyframes.
Source [564, 572), 18.800000–19.067000 s: the full phrase is present, but residual second-line travel and edge changes continue after the last glyph.
Source [572, 591), 19.067000–19.700000 s: a near-stationary reading interval contains three separate exact-RGB runs, not one identical-pixel hold. Small changes between the runs are not assigned a rendering cause.
Source [591, 597), 19.700000–19.900000 s: both lines enlarge slightly before the outgoing cut. The exact subpixel onset and easing remain unknown; the cut at frame 597 is outside the selected excerpt.
The fitted title surround and inspected native patches appear white. Uninspected native regions and faint measured residuals do not establish a separate blur trail, mask or wipe. Geist outlines and replacement copy remain editorial choices.
The fresh seventh-version critique ties both original-content comparisons. No defensible concrete defect remains in the inspected states, but this is not a candidate win. All 360 paired indexed frames were individually viewed fitted; 86 native-scale regions across 68 frame entries leave native spatial gaps. No native whole-frame or moving-video review is credited. This existing treatment remains under review, not a completed recipe.
The source is a retained, hash-verified native excerpt. Its full master is currently unavailable; this round does not refresh full-master pixel equivalence or whole-video coverage.
All 90 positions are retained. VP9 packet-duration fields differ from H.264 by up to 11 ticks at 16000 Hz; equal frame positions are not equal duration fields. The coded tail extends 5 ticks beyond the declared next-source-frame boundary.
The width multiplier 202/214 and 18px pre-scale space advance are adaptation choices, not recovered source font metrics. Pixel-bound measurements do not establish official CSS spacing or a quality improvement.
Publication is October 21, 2025. The metadata was retrieved after the recorded September 5, 2026 collection cutoff; contemporaneous possession is not established by this file.
Native indexed-image inspection is not moving-video review. No original audio was listened to; no sound-related claims are made. The outgoing cut is outside this excerpt.
Original text with code/SVG Geist outlines. No source footage or source-derived raster is an input to either candidate render. Editable downloads remain withheld pending package verification.
No playable group matches. Source inspection and rendering do not automatically publish an example.
Recorded updates
The two new candidate groups have four silent code/SVG renders but lack exact source comparisons. Neither is counted as a published addition; the accepted production baseline remains 36 draft groups.
Review 28 prefers the source over both sixth-version Atlas variants because words are too crowded. The builder returns with one fixed spacing repair, rendered once per original-content variant. Old movies and findings remain unchanged.
Fresh review 29 ties both seventh-version comparisons. The reviewer individually opens all 360 paired indexed frames fitted and 86 unresampled native regions across 68 frame entries. Suspected crowding, extra edges and clipped terminal letters do not survive the native checks. No native whole-frame or moving-video review is credited.
The local history now has 29 static review records: 15 candidate-preferred pairs, 25 source-preferred pairs and 18 ties. These are repeated comparisons, not distinct completed treatments. The requested win condition is still unmet; no tie is promoted to a win.
The Atlas reference is a retained native excerpt at frames 507 through 596, or 16.900 to 19.900 seconds end-exclusive. The full master is absent. Packet durations differ by up to 11 ticks at 16000 Hz, and the coded source tail extends 5 ticks past the next-frame boundary. These differences remain disclosed.
The seventh-version movies pass four local delivery configurations: paired playback, native-first Play, strict decoded-frame seeking, 0.5x speed, looping, fitted/native views, responsive layouts, reduced motion and no-JavaScript fallback structure. These are player checks, not motion-quality review.
The page separates the 11 checkpoint drafts from the 25 earlier drafts with direct navigation. This Atlas revision is an existing treatment refinement, not a 37th production group or two new groups for its content variants.
The September 13 separate progress preview preserves the 36-group production baseline. The user has now authorized one additional bounded staging copy for integrating this chapter into the original site; that allowance is not source-fidelity or completion approval.
This update adds navigation from the Atlas film and Opening studies as well as Motion, Editing, Techniques and the existing workbench. Earlier pages, comparison revisions, evidence and downloads remain intact. The Atlas revision is still one existing treatment, not a new draft group.
The recorded inventory remains 622 in-window videos at the September 5, 2026 17:13:40.602033 UTC cutoff. Source restoration and entire-timeline review remain outstanding. Full-video reviews, credited moving runtime, completed treatments and complete clean corpus passes remain zero.
A separate fresh inventory enumeration is in progress. Its results do not yet replace the recorded inventory or establish new source access, complete timeline inspection or clean review passes.
Snapshot 7 · 2026-09-13T01:44:16.007410+00:00
Checks for later published snapshots every 20 seconds.