mirror of
https://github.com/heygen-com/hyperframes.git
synced 2026-09-04 07:19:52 +00:00
WaterrrForeverandClaude Opus 4.8
05af482f22
* feat(skills): product-launch-video + consolidate motion knowledge into hyperframes-animation
- Add the product-launch-video skill: shot-sequence architecture where each
visual frame is a time-coded shot sequence picked from a blueprint menu and
paced to the voiceover (anti-PowerPoint). Includes the frame-worker sub-agent,
story/visual/motion-design references, and audio/captions/transitions/
stage-assets/assemble-index scripts.
- Consolidate motion knowledge in hyperframes-animation as the single source of
truth: promote the updated atomic rules (31 -> 36) and rename product-launch-
video's archetypes into hyperframes-animation blueprints (13 -> 15, replacing
the old set). product-launch-video, faceless-explainer, and pr-to-video now
reference them via ../hyperframes-animation/{rules-index,blueprints-index}.md
and the rules/blueprints dirs. Fixes the discrete-text-sequence broken links;
blueprints no longer ship per-id runnable examples, so example references in
the consumers were dropped.
- Default HeyGen TTS voice to Marcia (deterministic; was the API's first English
voice, which drifts on catalog re-sort). Override with --voice.
- assemble-index pre-assembly frame guards: auto-repair a sub-comp root missing
canvas dims; hard-fail on <video>/<audio> inside a sub-comp; hard-fail on a
timed non-root element missing class="clip" or overlapping same-track clips.
- Lint/CLI: lint media inside sub-compositions as an error; stop false-positive
caption layout/lint findings; contrast/layout-audit skip elements hidden by an
invisible ancestor.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(skills): clear CodeQL alerts in assemble-index.mjs
- script/style blanking regex now matches closing tags with trailing
whitespace (</script >, </style >) — js/bad-tag-filter (high).
- drop the existsSync precheck before reading/repairing a frame file; read
directly and handle ENOENT, removing the check->write TOCTOU window —
js/file-system-race (high).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
38 lines
4.8 KiB
Markdown
38 lines
4.8 KiB
Markdown
# spatial-pan-stations — Spatial Pan / Stations
|
||
|
||
**intent**: Pre-place a sequence of labeled stations on one oversized canvas, then traverse it with a single virtual camera — repeated lateral/diagonal pans that center each station in turn and reveal a callout at every stop, landing held on a final station.
|
||
|
||
**roles served**
|
||
|
||
- Hook (from hook-pan-timeline / #1 Hook_02): a horizontal timeline of evenly-spaced milestones, left-panned beat by beat, each marker getting a spring-popped callout, landing on the present moment ("evolution / milestone walk leading up to us").
|
||
- Problem (from problem-camera-pan-stations / #8 Problem_01): a connected web of pain "stations" linked by hand-drawn leading lines, diagonally panned station to station, ending on a tangled scribble knot ("too many disconnected steps — it's a mess").
|
||
- Product_Intro (from concept-demo-decode-pan): a two-shot strip bridged by ONE lateral pan — shot 1 holds a static phrase whose accent word 3D-flap-DECODES (the concept lands), then the camera pans across the strip (with background parallax) into shot 2, where a cursor drives a live typing demo. Pairs this pan with `cursor-ui-demo`'s focal-locked tracked typing.
|
||
|
||
**duration**: 7–10s (union of Hook 8–10s, Problem ~7s, concept-demo ~7s)
|
||
|
||
**shot structure**
|
||
One oversized flat canvas on a solid `[bg color]`; all stations/markers pre-placed in world space; `[accent color]` text + simple line-icons; one virtual `.world` camera pans ease-in-out between stops. Each station holds ~1.0s.
|
||
|
||
- Scene 1 (0.0–~1.0s): Camera opens on station 1 — `[label 1 / first step]` centered. A reveal lands on it (see variants). Camera then begins to PAN toward station 2, sliding station 1 out of frame.
|
||
- Scene 2 → Scene N-1 (~1.0s each): Camera PANS (ease-in-out) to center the next station; on arrival its `[label k]` (+ optional `[secondary label]`) is REVEALED with the role reveal. Repeat per station.
|
||
- Scene N (final, ~last beat): One last pan lands on the terminal station; the final `[callout / landing element]` reveals and HOLDS to the end. Camera goes static on the punchline.
|
||
|
||
- Variant — Hook: stations sit as evenly-spaced `[markers]` on a thin horizontal `[timeline]` (lower third); pans are LEFT-only along the single axis (timeline scrolls left). Each callout is a bordered `[callout box]` + downward triangle (offset drop-shadow) that SPRING-POPS up (scale 0→100%, bouncy overshoot, transform-origin at triangle tip) reading `[label k]`; a `[secondary label, e.g. year]` fades in and RISES above it. Some mid markers arrive as plain static text revealed by the pan alone (no box). Final scene lands on the `[present-day label]`, springs, holds.
|
||
- Variant — Problem: stations are scattered across a 2D web; pans are DIAGONAL, STEERED by `[accent color]` hand-drawn lines — each station has a rough write-on line/arrow that draws toward the next and the camera follows it (Scene 1 also draws a loop/circle around the headline's key word). Each station = a white `[line-icon]` above its `[label]`, revealed plainly by the pan (no spring box). Final scene: the accent line spirals into a dense chaotic SCRIBBLE KNOT centered on the field; camera holds static on the tangle (visual punchline).
|
||
|
||
**motion vocabulary**
|
||
repeated ease-in-out camera pans (horizontal-left for Hook, diagonal-steered for Problem) across one large static canvas; pre-placed stations sliding through frame via the pan; spring-overshoot callout pop with triangle-tip origin (Hook); rise-and-fade secondary label (Hook); plain labels/icons arriving via the pan alone; rough hand-drawn "write-on" leading lines/arrows + loop/circle key-word mark (Problem); terminal chaotic-scribble knot draw (Problem); static hold on the final station/punchline.
|
||
|
||
**rule mapping**
|
||
|
||
- camera pan / traverse across the canvas (primary) → `viewport-change` (single `.world` wrapper transform; PAN mode)
|
||
- sequencing the repeated pan beats into stops → `multi-phase-camera`
|
||
- centering each station as the pan target → `coordinate-target-zoom` (used as pan-to-target, no zoom)
|
||
- spring-overshoot callout pop, triangle-tip origin (Hook) → `spring-pop-entrance`
|
||
- rise-and-fade secondary label + plain per-station label/icon reveals via the pan → `discrete-text-sequence`
|
||
- hand-drawn leading lines / arrows / loop-circle key-word mark / terminal scribble knot (Problem) → `svg-path-draw`
|
||
- station line-icons (Problem) → `svg-icon-enrichment`
|
||
- static hold on the final station / punchline → (no motion; sustained held frame, no rule needed)
|
||
|
||
**camera modifier**: The pan IS the camera. One `.world` virtual-camera transform in PAN mode — `viewport-change` — sequenced across stops by `multi-phase-camera`, each stop targeted via `coordinate-target-zoom` (pan-to-target). No depth push-in (that distinguishes this from the cluster-push-in / dataviz-pushthrough blueprints).
|