AI clothing model generator: put your garment on a model

An AI clothing model generator puts your garment on a person who was never photographed in it: you give an image model a photo of the garment (a flat lay works) plus a photo or a description of the model, and ask for the model wearing it. Before the picture goes on a product page, check the print, the seams, and the color against the real garment.

The steps below use Sume's Image API and Format catalog docs, read on 2026-09-28.

What is flat lay photography?

A flat lay is a photo of a garment laid flat on a surface and shot from directly above. It needs no model, mannequin, or studio. It shows the whole garment, its print and color, but not how it hangs on a body. An AI model generator adds that part. A photo of the garment on a hanger or a mannequin works as input the same way.

How do I turn a flat lay into an on-model photo?

• Shoot the flat lay in even light, garment smoothed, whole piece in frame.
• Choose the person. Use a photo of someone who agreed to appear, or describe an invented model in the prompt.
• Host both photos at public HTTPS URLs and send them in input_references to POST /v1/images. ChatGPT Image 2.5 takes up to 16 references; models whose descriptor is {"min": 0, "max": 0} take none.
• Say which photo is which, what the model does, and what must stay the same on the garment.
• Pick a portrait ratio the model lists. In current code, ChatGPT Image 2.5 lists 4:5 and 3:4.

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 the man in the second photo wearing the shirt from the first photo, walking down a city street in daylight. Keep the shirt exactly as shown: its print, colors, collar, buttons, and pocket. Natural fit, tucked out.",
    "input_references": [
      { "type": "image_url", "image_url": { "url": "https://example.com/shirt-flat-lay.jpg" } },
      { "type": "image_url", "image_url": { "url": "https://example.com/model.jpg" } }
    ],
    "aspect_ratio": "4:5",
    "n": 3
  }'

Will the garment look exactly like mine?

Not guaranteed. A flat lay shows no drape, so the model guesses how the fabric falls, and it redraws the whole garment. Compare each version with the real piece:

• The print: scale, position, and any lettering.
• Seams, pockets, buttons, and the collar or neckline.
• Color, which shifts under the new light.
• Length and fit. The image shows a guess, not a size.

Can I make an on-model video instead?

Yes, with two catalog Formats that take a person photo and the garment. Virtual try-on video API shows the request.

What does it cost?

Each completed image is billed at its model's pricing line, so cost_usd × n is what you pay, and a failed or cancelled generation is not billed. Each model's price is on its GET /v1/images/models/{model_id}/endpoints record; the rest of the rate card is on API pricing. Result URLs are signed, so download the images you keep. For a garment with no person at all, see AI ghost mannequin.