Skip to main content
estudjo can generate short cinematic product videos from a single product photo. Video generation uses the same POST /generate endpoint as images — the model you specify determines whether image or video logic runs.

Key Differences from Image Generation

Before sending a video request, note these differences from the image flow:
  • duration is required — supported values are "5", "8", or "15" (plain number strings, no unit suffix)
  • prompt is recommended — pass a direct prompt string instead of a scene for the most control over the output
  • prompt takes priority — when prompt is present, any scene or sceneId value is ignored entirely
  • category is optional — for video models it falls back to the model-wide default when omitted

Required Fields

If both prompt and scene/sceneId are absent from the request body, the API returns 400 invalid_input.

Example Request

The example below uses a direct prompt string — the recommended approach for video generation.

Response

Poll GET /status with the returned task_id to track the generation and retrieve the video URL once complete.

Error Reference

Next Steps

Poll for Status

Track your video generation and retrieve the finished file URL.

Image Generation

Generate static product images using the same endpoint.