heygen-com--hyperframes/skills/embedded-captions/modes/cinematic

Cinematic mode (pure embed) — one engine, six DNAs

Cinematic mode compiles ../../dna/ through engine.html (make-composition.cjs). The old per-template HTML shells are retired — cinematic-cream maps to dna: "cream" automatically; the other archived templates (memory-wall / champion / portrait-header, in _archive/) remain as design references only.

Use this mode for pure-embed asks (no rail): brand film, hype, social reel, showcase. The DNA locks the visual language (type, palette scheme, blend, motion grammar, hero three-act); safe-zones v2 parameterizes it to the scene (sampled accent, light- direction contact shadow, depth-match blur); the agent decides layout only (planes, blocks, per-line typography within the DNA).

Workflow

1. bash scripts/prepare.sh <project> → matte ∥ transcript ∥ envelope → safe-zones v2
2. Pick a DNA (../../dna/README.md): bright hero band → ink, else by register (cream / editorial / keynote / documentary / loud). Recommend, let the user pick.
3. Author <project>/cinematic.json — "dna": "<name>" + thought-blocks (schema: scripts/make-cinematic.cjs header)
4. node scripts/make-cinematic.cjs <project> → plan.json → engine-compiled index.html
5. node scripts/preview-frames.cjs <project> → § Visual QA (failure checks + the 5 positive checks in ../../references/reference-bar.md)
6. bash scripts/render-and-composite.sh <project> → gates → final.mp4

What the engine generates (never author these)

• word timings from the transcript; accumulate-within-block / page-flip-between-blocks
• the hero hand-off + three-act orchestration (dim → RMS-coupled per-letter entrance → breathe + glow), per the DNA's hero block
• scene tokens: --accent (sampled), contact shadow, depth blur
• reading order, re-slot from measured heights, hero size/collision post-pass

What you DON'T do

• Override .cap color / blend / shadow / filter / motion curves — that's the DNA. Scene fights the look → pick a different DNA (bright → ink), never recolor.
• Hand-position the hero into a clean margin (it belongs ON the subject, ~30–55% occluded — safe-zones heroBands.best).
• Add full-frame grades/textures over the footage (hard rule: the video ships untouched).

Adding a DNA

dna/<name>.json — copy one, change the voice (see ../../dna/README.md § Adding). The engine consumes it with no code change. A DNA must be a distinct voice with a reason to exist, not a recolor.