Step 4 — Edit & Generate
You have a vidsheet from Step 3. Now you drive it:
- Fill ingredient cells with the idea’s content (script, topic, B-roll prompt)
- Configure creation cards on each cell — which model, what prompt, what aspect ratio
- Configure layers on each video cell — background, avatar, captions, CTA, z-index
- Trigger generations on cells and layers
- Poll generation status until
completed
Vidsheet anatomy recap
Section titled “Vidsheet anatomy recap”Auto Content Engine├── Sheet (left) ← data grid│ ├── Rows (one per video)│ ├── Columns (content type, e.g. SCRIPT / VIDEO / CAPTIONS)│ └── Cells (intersection of row + column)│ └── Creation card (generation_type + data)└── Video composition (right) ← timeline └── Layers (per cell: bg, avatar, text, captions, music)Every cell has exactly one creation card — a declaration of how that cell generates content. Every video cell has a stack of layers that composite into the final video.
The 3-call pattern for every cell
Section titled “The 3-call pattern for every cell”1. PATCH cell value or card config — say what you want2. POST cell generate — trigger the AI job3. GET generation status — poll until completedRepeat for each cell you want to fill. For a typical 3-column video engine (script / voice / video), that’s 9 calls.
Trigger a generation
Section titled “Trigger a generation”curl -X POST "https://api.gen.pro/v1/autocontentengine/$ENGINE_ID/cells/$CELL_ID/generate?agent_id=$AGENT_ID" \ -H "X-API-Key: $GEN_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "generation_type": "video_from_text", "data": { "prompt": "San Antonio taco truck at golden hour, steam rising, handheld camera", "model": "veo_3", "aspect_ratio": "9:16", "duration": 10 } }'Response:
{ "generation_id": 789, "status": "pending" }Poll until complete
Section titled “Poll until complete”curl "https://api.gen.pro/v1/generations/789" \ -H "X-API-Key: $GEN_API_KEY"Status flow:
pending → processing → completed | failed | stoppedOn completed: result (text) or output_resources (media URLs) are populated.
The 9 creation card types
Section titled “The 9 creation card types”Every cell picks one:
| Type | Generates | Typical column |
|---|---|---|
| Text | Text (scripts, prompts, notes) | SCRIPT, CAPTION COPY |
| Image from Text | Image | THUMBNAIL, HERO IMAGE |
| Video from Text | Video (no source image) | B-ROLL |
| Video from Image | Video (animating an image) | AVATAR VIDEO |
| Video from Ingredients | Video (compositing multiple assets) | MONTAGE CLIP |
| Speech from Text | Audio (TTS) | VOICE |
| Lipsync | Video (sync mouth to audio) | TALKING AVATAR |
| Captions | Captions over audio or video | CAPTIONS |
| Media (Upload) | Pass-through of an uploaded asset | BACKGROUND MUSIC |
Regenerate workflow
Section titled “Regenerate workflow”Not happy with a generation? Update the card’s data.prompt or data.model and re-trigger. Generations are idempotent per generation_id but creating a new generation for the same cell is always allowed — the latest one becomes the cell’s displayed output.
See Regenerate Workflow for the stop/continue/re-run patterns and credit behavior.