capabilities

Render CGI Sizzle

Name: Render Cgi Sizzle
Author: Athina AI

Assemble a 3D-CGI app sizzle — nano-banana CGI plates (blank-glow floating phone + a CGI studio with rim-light + placeholder burst shapes; studio look and phone finish come from the recipe choices) plus PIL compositing of the REAL App Store screenshots onto the bezel + burst-out overlays, driven by Kling 3.0 i2v steady-float per beat with a per-beat Ken-Burns FFmpeg push-in fallback when Kling garbles the UI, then VO/music mix (sidechain duck, loudnorm) + 1.15x speed + anti-AI grain finalize. The on-screen UI, any faces in it, and the wordmark are ALWAYS real assets composited via PIL — never AI-rendered. The paid steps (plates, i2v clips, VO, music) are separate capabilities; this ships the config + PIPELINE + FREE assembly and the recipe orchestrates the spend. Use for the cgi-app-sizzle video format.

by Athina AI

Run in Gooseworks View skill in Github

Install

Terminal

npx gooseworks install --all

# then, in Claude Code, Cursor, or Codex:
/gooseworks use the render-cgi-sizzle skill

About This Skill

render-cgi-sizzle

Assembles the 3D-CGI app sizzle video format: a phone floats in a CGI studio while five or six app features demo one per beat — each beat bursts REAL App Store UI elements out of the phone in 3D with rim-light + bokeh, then everything collapses back into the screen at a value-prop climax + a brand end card over a music bed. It reads as an Apple-keynote product film, not UGC and not a physical-product shoot.

The worked example (scripts/config.example.json) is MasterClass: a gold phone in a smoky-black studio with an amaranth rim-light, an "Eryn" keynote-cadence VO, a premium-tech piano/synth bed, and a "200+ classes. One app." climax. Those are one example, not defaults.

Choices

Creative calls the user makes (the recipe's choices, asked in one round before any paid step). The demo's value is only an example.

studio_look — background, rim-light colour, bokeh; also the end-card canvas + bar. Asked of the user; the demo used smoky-black #0A0A0A + amaranth #E32652 rim-light.
phone — the phone model/finish in every plate. Asked; the demo used a gold/champagne titanium iPhone 15 Pro.
narrator_voice — the VO voice. Asked; the demo used ElevenLabs "Eryn".
vo_delivery — how the VO is written and read. Asked; the demo used a restrained Apple-keynote spec-sheet cadence.
music — the bed's genre/mood. Asked; the demo used a premium-tech underscore.

This capability ships the recipe (scripts/config.example.json) + a config→step map (scripts/PIPELINE.md) + the FREE-assembly how-to (scripts/README.md). It documents the FREE, deterministic assembly between the paid model calls:

nano-banana CGI plates (paid, separate cap) render only the phone shell + the chosen studio + rim-light + bokeh + placeholder burst shapes; the screen is left a blank warm glow on purpose, and every plate is --anchored on beat 1 so the phone/studio stay identical across beats.
PIL real-UI compositing (FREE) — auto-detect the bright phone-screen bbox in each plate and feather the REAL App Store screenshot into the bezel → scene-NN-composite.png; bake real burst-out overlays (e.g. climax portrait tiles, rim-light baked before rotation) around the plate. The on-screen UI + faces + wordmark are ALWAYS real assets — never AI-rendered, so no claim is invented and nothing reads fake.
Kling 3.0 i2v steady-float (paid, separate cap) drives each composite; the burst-out pops/settles while the phone + screen stay locked. Any beat Kling garbles drops to a FREE Ken-Burns FFmpeg push-in (zoompan, heavier on the climax) — the shipped demo used this path for the feature beats.
Assembly + finalize (FREE) — dice + intercut concat (timeline locked from the measured VO durations), audio mix (sidechain-duck the music under VO, loudnorm -14 LUFS master), PIL brand end card (real wordmark, never AI), then 1.15x speed + anti-AI grain master.

See scripts/README.md for the full FREE-assembly detail and scripts/PIPELINE.md for the config-field → source-step map.

Run

Config-and-PIPELINE capability (no re-built runnable pipeline here). Copy the structure of scripts/config.example.json → config.json, fill the creative fields from the recipe choices (never the demo's values), rebuild the brand/beats/screens, and follow scripts/PIPELINE.md:

VO first (locks the timeline) → nano-banana CGI plates → PIL screen composites → burst-climax overlays → Kling i2v clips (Ken-Burns fallback per garbled beat) → PIL end card → captions → sidechain-ducked mix → 1.15x speed + grain.

Output: 1080x1920, ~22s H.264 (+ AAC music). 5-6 feature beats + PIL end card.

Contract

REAL UI, always PIL — never AI. Every app screen, any face in it, and the wordmark is the real asset composited via PIL. AI plates only ever render the phone shell, studio, bokeh, and placeholder burst shapes. This is the format's whole credibility and the guard against invented claims.
Kling for the float, Ken-Burns fallback per beat. If a Kling beat garbles the burst-out UI, distorts the phone, or animates the screen, fall that beat to a FREE Ken-Burns push-in — never ship a garbled beat.
Anchor every plate on beat 1 so the phone/studio read identical across beats (one shoot).
Timeline locked from the measured VO durations, never planned word counts.
The paid steps — nano-banana plates, Kling 3.0 i2v beats, ElevenLabs VO + music — are separate capabilities (create-image-fal, create-video-fal, create-vo-elevenlabs, create-music-elevenlabs); the recipe orchestrates them and gates the spend.

What's included

studio_look* — background, rim-light colour, bokeh; also the end-card canvas + bar.

phone* — the phone model/finish in every plate. Asked; the demo used a gold/champagne

narrator_voice* — the VO voice. Asked; the demo used ElevenLabs "Eryn".

vo_delivery* — how the VO is written and read. Asked; the demo used a restrained

music* — the bed's genre/mood. Asked; the demo used a premium-tech underscore.

Video Production Harness

Full video production sequence with script review, actual ingredient choices, controlled generation, editing, evidence-based quality review, polish, captions and delivery. A host binding supplies project storage, authentic approvals, provider access and billing.

Render Street Interview

Build a vox-pop street interview video ad. An interviewer with a handheld mic asks passers-by one question about the brand's product, they give blunt wrong guesses, one gives the real answer, and the cut lands on a branded end card. Generates the takes through the GooseWorks fal proxy (Seedance 2.0 with native voice), then grades, re-cuts, captions and gates them locally. Use for the street-interview format.

Write Video Ad Script

Write the words of a short-form video ad (voiceover, dialogue, chat bubbles, on-screen lines) the way performance creative teams do instead of from a blank page. Builds the script from the buyers' own words, the beat sheet of an ad that already works and three deliberately different angles, filters them with a rule check and a second non-Claude model, and takes the strongest into the review with the other two as one-line swaps. Use it in every video ad run before any paid step, and whenever the user asks to write, rewrite or improve a video ad script or says a script sounds generic or AI-written.