Pinterest Restyle beta: restyle a photo with a prompt on Sume
Pinterest Restyle is a beta in the US and Canada. To restyle your own photo, send it as a reference with an edit prompt and optional mask to Sume Images.

Pinterest Restyle is a beta feature, launching in the U.S. and Canada, that lets people use a photo to reimagine a space in the Pinterest app. If you want to restyle your own photo outside the app, Sume Images takes a photo as a reference, an edit prompt, and an optional mask, and returns new images.
Pinterest facts are from its newsroom post, read 2026-10-01; Sume facts are from the Image API docs. The two are different products, and this post does not compare their output.
What does Pinterest say Restyle does?
The post says Restyle lets people use a photo to reimagine a space, test ideas and compare options in their own home, and that it launches in beta in the U.S. and Canada. It is described as part of Pinterest Intelligence in the consumer experience.
How do I restyle a photo with Sume Images?
Send the photo in input_references with a prompt that names the change. The docs' own example is "make this scene look like a watercolor painting". Reference URLs must be public HTTPS.
const res = await fetch("https://api.sume.com/v1/images", {
method: "POST",
headers: {
Authorization: "Bearer " + process.env.SUME_API_KEY,
"Content-Type": "application/json",
},
body: JSON.stringify({
model: "openai/gpt-image-2",
prompt: "restyle this living room in a japandi style, keep the layout",
input_references: [
{ type: "image_url", image_url: { url: "https://example.com/room.jpg" } },
],
}),
});
console.log(res.status, await res.json());How do I keep part of the photo unchanged?
Use a mask. The docs list openai/gpt-image-2.5 (Flare) and openai/gpt-image-2.5-sunburst as supporting an optional mask_url along with up to 16 image references. Image 1.0 uses mask_image_url with image_urls instead; see which edit fields to use.
| Route or model | Reference field | Mask field |
|---|---|---|
POST /v1/images (GPT Image 2.5 models) | input_references, up to 16 | mask_url (optional) |
| Image 1.0 | image_urls, 1 to 10 | mask_image_url |
What should I check before relying on it?
Models whose input_references descriptor is min 0, max 0 reject references, so read the catalog entry first with GET /v1/images/models. Review each result yourself; a prompt-based edit can change details you wanted kept.
Sources
Related posts
More in Use cases
- Pinterest Showcase ads specs: title Pin plus four cards
A Pinterest Showcase ad is one 2:3 title Pin plus up to four static cards, with up to three features per card. Plan it as one generated image set.
- Pinterest video pin minimum length: 4 seconds, and which models fit
Pinterest's product specs set a 4-second minimum for video pins. seedance-2.5 starts at 4 seconds in Sume; wan-3.0 and Gemini Omni Flash can go shorter.
- Pinterest Visual Search Ads: product photo variants from one image
Pinterest Visual Search Ads put ads in visual search results. Make angle and context variants of one product photo with Sume's image-to-image reference input.
- Podcast video clip in vertical format: 1080x1920 by default
Timeline 1.0 renders 1080x1920 MP4 by default, so a podcast clip is vertical unless you set output width and height. Cut, segment and render steps.
Written by Sume