Tanisea First-Act Semantic Sync v2.3

Tanisea first-act semantic sync v2.3

This release completes the precision pass requested for the first half of the film, with special attention to the 00:40–00:50 passage and the horizontal lyric progress rail. The approved second half and restrained two-tone visualizer design are preserved.

Exact semantic timing

• C1-06 now follows the performed source meaning: “Let” → “tremble” → “the earth.”
• C1-07 now follows “Through” → “walls” → “apartment,” matching the audible Russian concept order while keeping the English line stationary.
• C1-08 preserves its reviewed target order and releases “of fire” at sample 2,171,396 (00:49.238005) instead of holding it to the authored break.
• The reviewed alignment JSON remains unchanged. The correction is a presentation-layer mapping from existing source cues to their exact English semantic targets.
• Every corrected target is inactive before its nearest cue frame, fully emphasized on contact, and inactive at its exclusive cue end at both 60 and 120 fps.

Cleaner first-act handoffs and rail

• First-act lyric entrances use a one-proof-frame lead (368 samples), eliminating the visible 240 ms precontact stack.
• The incoming line is absent or zero-opacity immediately before contact and fully visible on contact; the outgoing line clears on that same boundary.
• The long horizontal lyric rail now advances by equal semantic cue stages, moves only while a reviewed cue is active, and pauses through measured inter-word gaps instead of drifting across the broad line envelope.
• V1 and C2 retain the approved cinematic attack, release, and handoff behavior.

Visualizer continuity

• The restrained v2.2 two-tone spectrum remains intact: ember bass, saturated teal upper bands, bounded transients, and no neutral bridge or pale high-frequency zone.
• The measured 64-band core, safe-area footprint, and minimum lyric clearance remain unchanged, keeping the visualizer calm and easy on the eyes.

Media and verification

• Production master: 1080×1080, 60 fps, 9,180 frames, 10-bit HEVC hvc1, BT.709, with the locked original AAC packet stream.
• Synchronization proof: 1080×1080, 120 fps, 18,360 frames, H.264 avc1, BT.709, with the same locked original AAC packet stream.
• Source verification passes strict typechecking, 1,205 tests across 20 files, browser-measured layout, and exact composition discovery.
• Canonical QA includes a soundtrack-bearing 39.5–50.5 s review range, 152 cue-contact frames, 12 contact sheets, 72 exclusive-end frames, six release sheets, selected encoded stills, and two independent full release-matrix executions.
• The supplemental workflow archive preserves the generated QA, visual review, alignment provenance, publication evidence, and repository records excluded from Git source control.

Use CHECKSUMS.sha256 for the core package and WORKFLOW-EVIDENCE.sha256 for the supplemental workflow archive. Earlier releases remain unchanged.

Workflow license, source credits and AI disclosure

Project-owned workflow contributions are offered under Creative Commons Attribution 4.0 International (CC BY 4.0), solely within the license scope and exclusions. This applies to licensable project contributions in existing source/workflow archives; it does not relicense their third-party contents.

We do not take credit for the original music, lyrics, recordings, remixes, animation, characters or source artwork. Those works, fonts and dependencies retain their original rights. Finished films, screenshots and AI-assisted covers are not offered as wholly CC BY 4.0 assets.

Original creators and source links · AI use disclosure · CC BY 4.0 legal text

Notice added 9 September 2026. Existing media and source archive bytes remain unchanged.