vox-director

Files (52)

• .gitignore
• AGENTS.md
• LICENSE
• README.md
• README.zh.md
• SKILL.md
• SKILL.zh.md
• showcase-football.mp4 · assets/
• showcase-money.mp4 · assets/
• showcase-silicon-valley.mp4 · assets/
• showcase-tang.mp4 · assets/
• football.jpg · assets/thumbs/
• mexican.jpg · assets/thumbs/
• money.jpg · assets/thumbs/
• silicon-valley.jpg · assets/thumbs/
• tang.jpg · assets/thumbs/
• cr7-act.elements_spec.json · examples/
• money-15s.beats.json · examples/
• money-60s-9x16-english.beats.json · examples/
• ronaldo-9x16-kling.beats.json · examples/
• tang-30s.beats.json · examples/
• llms.txt
• package.json
• beat-layer.md · references/
• host-mode.md · references/
• local-engine.md · references/
• models-and-gotchas.md · references/
• prompt-guide.md · references/
• voices.md · references/
• aroll_assemble.py · scripts/
• aroll_clips.py · scripts/
• asr_beats.py · scripts/
• assemble.py · scripts/
• atlas_cloud.py · scripts/
• audio.py · scripts/
• clips.py · scripts/
• confetti.py · scripts/
• croll_keyframes.py · scripts/
• extract_elements.py · scripts/
• host.py · scripts/
• host_clips.py · scripts/
• host_plates.py · scripts/
• kenburns.py · scripts/
• keyframes.py · scripts/
• lipsync_score.py · scripts/
• mg_scrapbook.py · scripts/
• motion.py · scripts/
• provider.py · scripts/
• style_bakeoff.py · scripts/
• styles.py · scripts/
• text_overlay.py · scripts/
• vox-director.skill

SKILL.md (first lines)

# Vox Director

Turn a one-line topic into a finished **Vox-style paper-collage video**: a bold, punchy,
narrated explainer/ad where each beat is a torn-paper collage poster that comes alive, with
voice-over, music and captions. Runs on **one Atlas Cloud API key** + local **ffmpeg**.

The look is the modern editorial paper-collage popularized by Vox explainers and creators
like Stav Zilber / rom1trs: hand-cut paper cut-outs, torn edges, tape, halftone dots,
newspaper clippings, bold flat color per beat, big cut-out headlines.

## The core idea (read this first)

The Vox collage look and the collage motion are **two different steps**:

1. **The look is born in the IMAGE step.** Each beat is a finished collage *poster* made by a
   text-to-image model. All the collage DNA (torn paper, cut-outs, halftone, bold color,
   headline text) lives in that image. If the image isn't a rich collage, nothing downstream
   will save it.
2. **The motion is added after.** By default an AI video model animates the whole poster (the
   "living poster" path — simple, automated). For dramatic *piece-by-piece* assembly you cut
   the poster into parts and drive them with the local keyframe engine (advanced path).

Everything hinges on the prompts. **Before writing any image or video prompt, read
`references/prompt-guide.md`** — it has the exact prompt structures that make the difference
between "a real Vox collage" and "a moving PowerPoint".

## Prerequisites (check, don't skip)

- `echo "${ATLASCLOUD_API_KEY:+set}"` — if empty, tell the user to set it (get one at
  https://www.atlascloud.ai/console/api-keys) and stop.
- `command -v ffmpeg ffprobe` — required for assembly (`brew install ffmpeg` on macOS).
- `python3 -c "import PIL"` — Pillow, for captions/watermark overlays.

## Standard workflow (topic → film)

This is the default, most-automated path. Every stage is one script, all driven by a single
`beats.json` per project under `out/<project>/`.

1. **Topic → beat map.** First **read `references/beat-layer.md`** (the story layer) and pick a
   narrative `arc` that fits the topic (`timeline` for history, `pas`/`bab` for ads,
   `how_it_works` for explainers, `man_in_hole` for transformations, …). Then write
   `out/<project>/beats.json` following that arc: **beat-1 headline must be a ≤3s hook**; beat
   count per duration (30s→6–8, 60s→10–12); split each beat into **2 shots** (wide+detail) with
   **per-shot `camera_move` VARIED across adjacent beats** (never repeat; `static` on the payoff)
   and **rich `element_motion`** (see step 4). Each beat: `narration`, `title_cn`/`title_en`,
   `scene`, `bg`, `feel`, `hook`. This draft is the **first mandatory approval gate** — show the
   user the beat map before generating (the aspect-routing approximation in step 4 is the other
   one). Examples in `examples/`.

2. **Pick the visual style (hybrid — do this BEFORE keyframes).** Do not reuse one house style
   for every topic. Read `references/prompt-guide.md` (§5 theme presets); pick 3–4 **theme presets**
   (`styles.THEME_PRESETS`: `american-retro`, `swiss-modern`, `punk-zine`,
   `soviet-constructivist`, `wpa-propaganda`, `70s-groovy`, `chinese-ink`, `atomic-age`,
   `newsprint-editorial`) that fit
   the topic's era/culture/tone — **or compose a custom theme** by mixing the prompt-guide dimensions
   (medium/era/palette/type/finish) when none fit. Match the topic, **not** the language (an
   English film on Chinese history should look Chinese). A theme bundles the whole LOOK layer
   (idiom+palette+type+finish+mood+motion). Run a bake-off and let the user pick by eye — AI
   proposes, the library is the quality floor, the human decides. Set the pick as `"theme"`:
   `python3 scripts/style_bakeoff.py out/<project> american-retro,swiss-modern,punk-zine,atomic-age`
   Set the chosen name as `"collage_style"` in beats.json (keyframes.py reads it).

3. **Keyframes (the collage look).** `python3 scripts/keyframes.py out/<project>`
   Generates one collage poster per beat/shot with **google/nano-banana-2/text-to-image**,
   headline text baked in. Compose prompts with the 5-part structure in
   `references/prompt-guide.md`. Verify each poster looks like a *real layered collage*
   before animating — re-roll cheap ($0.08) here rather than paying to animate a weak image.

4. **Motion.** `python3 scripts/clips.py out/<project>`
   Animates each poster with **google/gemini-omni-flash/image-to-video**. Two independent axes
   (see `references/beat-layer.md` §3, tested on our stack):
   • **`camera_move`** — ONE move per shot. Safe/default: `{static, push_in, pull_out, pan, tilt,
     parallax}`. **Bold/experimental** `{orbit, dolly_zoom, roll, whip}` are **available, not
     banned** — they can warp the flat art, so pair with `constraints: loose` and **re-roll**.
     Any custom phrase also passes through.
   • **`element_motion`** — where the energy lives; **AI writes it per beat to fit that scene** (not a
     template). Make it RICH (several elements moving) — be bold. A **hero element flying across
     the frame** (paper bird/plane/coins) is a great **occasional** punch on a key beat, **not
     every shot** (a flyer in every frame reads as a formula).
   `motion_style` = amplitude `calm | punchy | max` (the theme sets a default). **`constraints`**
   = `strict` (default: defect guards on — flat-2D, one-way, no-morph; best for clean text-heavy
   explainers) or `loose` (let the model explore 3D/bold moves; re-roll the misses). **Headline
   text is hard-protected only on shots that have a title** (detail shots without a headline are
   free to go wild). For **real people / brand logos**, Omni & Seedance refuse — set
   `"video_model": "kwaivgi/kling-video-o3-pro/image-to-video"`.
   **Aspect routing** (`styles.resolve_video_aspect`, second approval gate): `clips.py` resolves
   `doc["aspect"]` against the chosen `video_model`'s own supported ratios — exact match wins;
   Omni is 16:9/9:16 only, Kling reference-to-video adds 1:1, Kling image-to-video/video-edit and
   Seedance just follow the input/ratio param. When there's no exact match it picks the nearest
   ratio but **stops and asks you to confirm** (set `"aspect_approx_confirmed": true` once you
   have) rather than silently reframing the film — every clip in one run shares the same resolved
   aspect so the finished film is never mixed.

5. **Voice + music.** `python3 scripts/audio.py out/<project>`
   One consistent narrator via **xai/tts-v1** + instrumental BGM via **minimax/music-2.6**.
   **Pick `voice_id` to fit the topic + language** (don't just keep the default) — see
   `references/voices.md` for the full roster (5 multilingual + ~66 native voices by language,
   with gender). Default `leo` (male, documentary). To narrate in a REAL person's own voice
   (the presenter of a C-roll photo, a brand voice), set `voice.clone_ref` to a local audio
   sample — narration switches to seed-audio voice cloning with a pinned-speaker,
   studio-clean template that keeps timing beat-stable (see gotchas: never hand seed-audio
   bare narration without that pin).

6. **Assemble.** `python3 scripts/assemble.py out/<project>`
   ffmpeg: normalize + concat all shots, lay the single narration ducked under the music,
   burn captions timed per beat, add the watermark. Output `out/<project>/final.mp4`.

7. **Verify.** You can't read an mp4 directly — extract frames to jpg and look:
   `ffmpeg -ss <t> -i final.mp4 -vf "scale=640:-1,format=yuvj420p" -frames:v 1 f.jpg`

### Cadence — how long shots should be

A common mistake is one long shot per beat. On a 9:16 / social piece especially, a static
10s shot reads as dead air. Aim for a **cut every ~4–6 seconds**:

- **Shots run 3–6s; never let a single shot exceed ~7s** — beyond that the AI motion has
  nowhere to go and it feels static.
- **A beat's narration is ~8–10s, so give each beat 2 shots** (a *wide* establishing shot with
  the headline + a *detail* cut-in without it). The narration plays continuously across both;

From Alisa0808/vox-director, MIT.