Skip to main content
The Group Photo API inserts people from your input images into a reference scene. You provide 2–20 face images and a target background scene, and the API returns a composed group photo with those faces placed naturally into the scene. The call is synchronous — one request, one finished result.
The API holds the connection until the job completes. This can take up to four minutes. Most HTTP clients default to a 30-second timeout — raise yours to at least 300 seconds.

Before you start

  • A valid Bearer token. See Authentication.
  • Enough credits. Failed generations are refunded automatically.
  • Publicly reachable image URLs — all input and reference images are fetched server-side and cannot require authentication or have expired signed URLs.

Step 1 — Submit the request

Request fields

The scene prompt is generated automatically from the reference image. Use instructions to refine it, or prompt to replace it entirely.

Step 2 — Read the response

Credits are reserved when the job is accepted and refunded automatically if the generation fails.

Common issues

See Errors for the full code reference.