Skip to content

How to make product photos with AI

Upload two to four photos of the product as references in the Photo section, describe the scene and choose the frame format. GPT Image 2.5 takes up to 16 references, Seedream 5.0 Pro up to 10. One run gives one to four variants, and the token price shows on the button. This keeps the packaging and shape closer to the real item.

Updated:

Step by step

  1. Shoot the product

    Even light, no glare on the packaging, the whole item in frame. Two to four angles give the model more data about the shape.

  2. Describe the scene

    Where the product stands, the background and the light. Put brand names and packaging text in quotes.

  3. Run and pick a result

    One run gives one to four variants. Save the best one and rerun the others with a fix.

Which model to use

GPT Image 2.5 takes up to 16 references, Seedream 5.0 Pro up to 10. If the item is small and the packaging text matters, start with GPT Image.

For a quick edit of a finished picture, use the editing tool in the Photo section.

Frequently asked questions

Can the product turn into a different item?

Sometimes small details change. Upload more angles and say what must stay the same, for example logo unchanged.

Can I put the product into my own background?

Yes, upload the background as a reference or describe it in words. A photo of the background helps when a specific interior matters.

How many variants can I get at once?

From one to four per run. You choose the number before you start, and the token price shows on the button.

Can I make a product photo without a shoot?

You can from a description, but then the product is invented. For a listing, upload at least one real shot.

Why did the packaging text get distorted?

Models often mix up small text. Make the product larger in the frame, put the text in quotes and check every variant before you publish it.