Shopify AI image editor scenes: one at a time vs n

Shopify's AI image editor creates one generated scene at a time; you can generate more sets, but unused scenes are deleted when you close the editor unless you save the images to your device. With the Sume image API you can request several scenes in a single call with n and read them back from the job result later, so closing a tab does not discard them.

Shopify's behavior is from its Help Center page; Sume's from the image generation docs, both read 2026-10-01.

How many scenes can one Sume request return?

The docs say to request up to 10 images per call with n, and that per-model ceilings are lower. The catalog example for Seedream 4.5 lists n as a range from 1 to 4, so read the n descriptor for your model rather than assuming 10.

{
  "model": "bytedance-seed/seedream-4.5",
  "prompt": "the same mug in four settings: kitchen, studio, desk, outdoor table",
  "n": 4
}

Where do the results live?

Sume jobs have standard status and result endpoints: GET /v1/jobs/{id}/status and GET /v1/jobs/{id}/result. A 200 is the image response and a 202 is a job envelope, which slow configurations (4K, high quality, large n) are most likely to return. Store the job id and fetch the images when you are ready.

Which approach fits which workflow?

The editor suits a one-off edit inside the Shopify admin. A request with n suits comparing scenes for many products; see batch image generation limits.