Assemble a stop-motion hand-swatch-cycle product-demo ad from a config — a sequence of still PLATES (one hand swiping a single-barrel product across a test surface, the barrel + swatch changing per plate while the hand, background, crop, and lighting stay locked) is PNG→mp4 loop-encoded at each plate's own stop-motion hold (fast motion frames 150–250ms, per-shade ~380ms, hero beats 1100–1800ms), concat-demuxed with HARD cuts into a silent master, closed on a Playwright HTML-rendered branded end card (serif tagline + sans subtitle + real logo SVG over a hero BG, never AI-rendered text), and muxed with a pre-sourced music track playing under the end card with a fade tail (no VO). This is the FREE deterministic assembly stage (loop-encode + concat-demux + end-card render + music mux); the master-anchor plate, shade plates, and end-card BG come from create-image-gpt-image-fal and the track from create-music-elevenlabs. Use for the stopmotion-hand-swatch-cycle format.
Install gooseworks: https://gooseworks.ai/creative-agent/install.md # then, in Claude Code, Cursor, or Codex: /gooseworks use the render-stopmotion-hand-swatch-cycle skill
Assemble a stop-motion hand-swatch-cycle ad from a config: a fast, tactile product demo where a single hand swipes ONE product barrel across a test surface, and the shade of the barrel AND the painted swatch stripe changes on every frame while the hand, background, crop, and lighting hold still — cycling the variant family so the viewer self-identifies their match, then a hero-pick payoff and a branded end card. The hand, background color, test surface, hero variant and music are the recipe's user choices; this capability only assembles the plates it is given. This capability is the FREE, deterministic assembly — the per-plate PNG→mp4 loop encode, the concat-demux, the end-card render, and the music mux.
scripts/config.example.json is the worked example — reference only, never the default (Kestrel
Art Supply "Swipe Your Color": an oil-pastel stick, a medium-tan hand, mustard-yellow background, round paper card; ~16.6s 1080×1920
9:16, ~24 cycle plates + finale beats + a 3s end card); scripts/PIPELINE.md maps every config
block to its source step and scripts/README.md documents the free assembly.
This is the FREE, deterministic assembly stage — it spends nothing. The paid inputs are separate
capabilities — the master-anchor plate + the per-shade / motion / finale plates + the end-card hero
BG (create-image-gpt-image-fal, gpt-image-2 EDIT mode, each plate anchored to the SAME
master-anchor — never chained), and the music track (create-music-elevenlabs, or a brand-supplied
mp3). Given the ordered plate PNGs + per-plate holds + the end-card hero BG + the brand logo SVG +
the music track, render-stopmotion-hand-swatch-cycle loop-encodes each plate at its hold,
concat-demuxes with hard cuts into a silent master, appends the HTML-rendered end card, and muxes the
music under it → the master. Re-cuts (re-timed holds, a re-ordered cycle, a swapped end card) reuse
the existing plates / track and spend nothing.
ffmpeg -loop 1 -t <pose_hold_ms>) to 1080×1920 @ 30fps crf18. This is stop-motion — discrete
held frames, no camera moves, no character animation. Do not animate the plates.ffmpeg -f concat) — NO dissolves. The
frame-swap tactility is the whole point; a crossfade erases it.concat demuxer over
the ordered plate list is correct and cheapest. (filter_complex is only needed when clips carry
mismatched audio — these don't.)concat demuxer overResearch supported advertising promises from a specific buyer's context, current product facts, customer language and observed ads. Turn them into static, video or copy ideas; match video ideas to actual recipes and save a product-scoped angle bank for write-video-ad-script. Product-led discovery can use verified facts without reviews. Reference persistence and reach are observations, not proof of performance.
Analyze ad campaign performance data (Google, Meta, LinkedIn) to identify what's working, what's wasting budget, and specific cut/scale/test recommendations. Runs statistical analysis, funnel diagnostics, and multi-channel budget reallocation with specific dollar-amount shift recommendations and scenario modeling.
For paid lead-gen and participant-recruitment ads, replaces vanity CPA with true CAC per qualified lead by joining ad-platform data with downstream funnel events, surfaces tracking gaps, and classifies every creative into Scale / Keep / Investigate / Cut.