Creation Cards

Each column in an Auto Content Engine sheet has a creation card type that determines what kind of content its cells generate. When you call the generate endpoint, the generation_type and data parameters you pass depend on the card type assigned to that column.

Card types

Card Type	What it does	Models	Detail page
Text	AI-generated text content (scripts, prompts, descriptions)	Gemini, OpenAI	Text
Media	Upload an image or video file directly	—	Media
Image from Text	Generate images from text prompts	Gemini, Gemini Pro, Midjourney, Grok	Image from Text
Video from Text	Generate video from a text prompt	Veo 3/3.1, Sora 2, Kling 1.6, Seedance, Grok	Video from Text
Video from Image	Generate video using a reference image	Kling 2.1/2.6, Veo 3/3.1, Sora 2, Seedance, Grok	Video from Image
Video from Ingredients	Generate video from a prompt combined with assets	Pika, Kling 1.6, Seedance, Veo 3.1, Grok	Video from Ingredients
Speech from Text	Convert text to speech with voice selection	ElevenLabs	Speech from Text
Lipsync	Sync an audio track to a video	sync.so, gen	Lipsync
Captions	Generate captions from audio or video	Gemini	Captions

Shared concepts

These concepts apply across multiple card types.

AI Text

Any text field — prompt, script, negative prompt — can itself be AI-generated. Pass a model and prompt within the text field’s configuration:

{
  "model": "gemini",
  "prompt": "Write a 30-second script about sustainable fashion"
}

Supported models: "gemini" and "openai".

Variables

Text fields support {{variable_name}} template substitution. Variables are defined at the sheet level (see Variables) and replaced at generation time.

{
  "prompt": "Create a {{product_type}} ad for {{brand_name}}"
}

Media fields can also reference variables via prompt_var, image_var, and similar fields — pulling values from other cells or global variables.

Aspect ratio

Available ratios vary by model. Common options:

1:1 — square
9:16 — vertical / portrait (TikTok, Reels, Shorts)
16:9 — horizontal / landscape (YouTube)

Check the individual card type page for model-specific ratio support.

Duration

For video card types, duration is specified in seconds. Available values depend on the model. Video cards can also accept optional resolution ("1080p" or "720p" where supported) and optional native_audio for models that support native audio. If omitted, the API uses the cell defaults or the closest supported model defaults. See each card type page for details.

How generation works

1. Trigger

Send a POST request with the generation_type and data for the card type:

curl -X POST "https://api.gen.pro/v1/vidsheet/101/cells/3000/generate?agent_id=42" \
  -H "X-API-Key: your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "generation_type": "text",
    "data": {
      "model": "gemini",
      "prompt": "Write a hook for a cooking video"
    }
  }'

The response returns a generation_id with status: "pending".

2. Poll for completion

Poll the generation status endpoint until the job finishes:

GET /v1/generations/{generation_id}

3. Status flow

pending → processing → completed
                     → failed
                     → stopped

Status	Meaning
`pending`	Queued, waiting to be picked up.
`processing`	Model is actively generating.
`completed`	Done. Check `result` and `output_resources`.
`failed`	Something went wrong. Check `failed_reason`.
`stopped`	Manually cancelled via the stop endpoint.

4. Retrieve output

On completion, the generation object contains the output:

Text cards: the generated text is in the result field.
Media cards (image, video, audio): URLs are in the output_resources array.

{
  "id": 9000,
  "status": "completed",
  "output_resources": [
    {
      "url": "https://cdn.gen.pro/outputs/abc123.mp4",
      "thumbnail_url": "https://cdn.gen.pro/thumbnails/abc123.jpg",
      "object_type": "video"
    }
  ]
}

Credits

Credits are pre-charged when a generation is created. If the generation fails or is stopped, credits are automatically refunded.