Google Merchant apparel images worn by people: reference-guided images

Google Merchant Center's image page says to provide images of products worn by people for clothing (apparel) products. Sume's openai/gpt-image-2.5 accepts up to 16 image references, so you can pass a garment photo and ask for a worn-by-model result.

What does Google ask for apparel?

Read on 2026-09-30: provide images of products worn by people for clothing (apparel) products. The page also says not to use an image with promotional elements or content that covers the product.

How do I pass the garment to the image API?

Use input_references, each entry an image_url object with a public HTTPS URL. The Image API docs state ChatGPT Image 2.5 supports up to 16 references and text-to-image. Ask for the framing you want in the prompt and choose a portrait aspect_ratio from the catalog, for example 3:4, so a standing figure fits the frame.

curl -X POST https://api.sume.com/v1/images \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-image-2.5",
    "prompt": "Full-length photo of a model wearing this jacket, plain backdrop",
    "aspect_ratio": "3:4",
    "input_references": [
      { "type": "image_url",
        "image_url": { "url": "https://example.com/jacket.jpg" } }
    ]
  }'

What does each requirement map to?

A quick mapping of Google's lines to the request.

What are the risks?

Generated wear can change a cut, a print or a colour. Compare the result with the physical item, and do not publish an image that misrepresents it. Check your Merchant Center diagnostics after upload, since this post only covers what Google's page lists.