render-hypermotion
Assembles a product-hypermotion + kinetic-typography ad: a vertical 9:16 sizzle (~20–30s) where ONE spectacular AI-gen hypermotion clip of the hero product carries the energy and punchy PIL-rendered spec cards carry the facts. Music-led, no VO — it reads as a high-production sizzle, not UGC.
The reusable IP is one-call-many-cuts + intercut: dice ONE 12–15s Seedance 2.0 hypermotion i2v into 5–6 segments (never a paid call per segment) and interleave PIL kinetic-typography spec cards between the cuts, capped by a real-logo end card, over a 124 BPM bass bed.
This capability ships the FREE, deterministic assembly — everything between the two paid
model calls. It is documentation-grade + config: the format has no runnable driver.
scripts/ carries the worked config + the step-by-step pipeline doc; the Soundboks
reference PIL impls (gen_kinetic_v6.py, gen_endcard_v10.py, assemble_v10.py) are
copied + adapted per run, not vendored.
- config.example.json — the Soundboks worked example: hero product, 5 spec callouts,
the 5-block Seedance prompt, per-card treatments, beat structure, end-card spec, music
brief, dims. Copy to
config.jsonand edit. - PIPELINE.md — the full config-field → step map: Phase 0 assets → Phase 1 PAID Seedance + music (parallel, gated) → Phase 2 FREE PIL cards → Phase 3 FREE dice/intercut/concat/mux → Phase 4 watch/QC. Names the atom/tool each step uses, plus 20s/25s/30s beat-structure variants.
- README.md — documents the FREE assembly (below).
The FREE assembly
- PIL kinetic-typography spec/CTA cards — full-frame 1080×1920 clips rendered frame-by-frame on a dark grain BG with the static Space Grotesk Bold TTF: italic skew (~6°), outline echo at 1.08× (bleed-safe), 3D extrusion (hero stat + CTA, side = brand accent), slam-with-shake, and color/inversion flash. No VO — the cards carry every fact.
- Real-logo PIL end card — the brand's actual logo PNG (base64-decoded out of the SVG, never typeset) with a slam-motion-blur entry, settle, continuous micro-motion (±1% scale + ±3px drift for the full hold — never freeze), an inversion flash at ~60%, and a cascade reveal of the spec-dot subtitle + CTA.
- FFmpeg dice + intercut — center-crop the ONE hypermotion clip 1:1 → 9:16, dice into 5–6 decreasing-length segments at Seedance's natural beats (crash-zoom → orbit → settle), then concat in the fixed intercut order: open on the intro card, alternate segment ↔ spec card, end on the CTA + brand end card.
- Explicit-map mux — mux the music bed as a SEPARATE pass with
-map 0:v -map 1:a(the default mapping silently ships ~1 kbps garbage audio; verifyffprobe≈ 192 kbps).
Run
# Phase 2 — FREE PIL cards (per config.text_cards + config.end_card) # render each spec/CTA card + the real-logo end card, 1080x1920, frame-by-frame → mov # Phase 3 — FREE dice + intercut + mux # center-crop 1:1→9:16, dice into 5-6 segments, concat in beat_structure order, # then explicit-map mux the music bed → master-final.mp4
Output: master-final.mp4, 1080×1920, ≈20–30s (25s default) h264 (+ 192 kbps aac music) —
14 segments: intro + 6 hypermotion cuts + 5 spec cards + CTA + real-logo end card. All
FREE, $0.
Contract
- Deterministic + FREE (PIL + FFmpeg); no paid calls, no AI-rendered logo or spec text.
- ONE hypermotion clip, diced into 5–6 segments — never a paid call per segment (cheaper AND identity-consistent: same camera/grade/subject in one call).
- The end card is the brand's real logo PNG, base64-decoded from the SVG — never typeset; it must micro-move for the full hold (never freeze).
- Spec cards carry every fact (no VO); outline echoes stay ≤1.08× so nothing bleeds off frame; use the static Space Grotesk Bold TTF (variable renders as Regular in PIL).
- The paid steps — the ONE Seedance 2.0 hypermotion i2v (5-block prompt with the mandatory ABSOLUTE CONSTRAINTS block, or the geometry drifts) + the ElevenLabs 124 BPM bass bed — are separate capabilities (create-video-fal, create-music-elevenlabs); the recipe orchestrates them and gates the spend.
