Blur a photo’s background

Drop a portrait or any photo. The subject stays sharp and the background is blurred. A slider sets how much. Your photo never leaves your device.

or paste a photo with Ctrl V

How do I blur the background of a photo?

Drop the photo above and the AI finds the subject. The tool blurs the rest of the picture, and the Blur slider sets how much. The blur is made from your own photo, and all of it happens in your browser, so nothing is uploaded.

  1. Add the photo. Drop it, pick it, or paste it with Ctrl V (⌘ V).
  2. Set the blur. Drag How blurry. Press V to slide between before and after.
  3. Fix the edge if needed. If the blur creeps onto the subject, paint with Restore (R). If sharp background shows, paint with Erase (E). A little Feather makes the join less harsh.

How is this different from portrait mode?

Portrait mode measures depth with the camera. This tool works from the picture alone, so it works on photos from any camera, old pictures and screenshots. The blur is even, like a flat backdrop. It does not grow with distance. So it looks best when the subject stands clear of what is behind it.

Can I use a different background instead?

Yes. Choose Color for a plain backdrop, Picture to put the subject on a photo of your own, or Transparent for a PNG or WebP with no background at all.

Questions

How do I blur the background of a photo?

Drop the photo, choose Blur under Background and drag the slider. The blur is made from your own photo, so the light and colors match, and the subject stays sharp.

How good is it on hair?

Good on most hair, because the edge is snapped to the photo and the old background’s color is taken out of it. Wispy hair against a similar color can lose strands. Use Restore on those, and a little Feather if it looks harsh.

Is my photo uploaded anywhere?

No. The AI model runs in your browser, on your graphics card when the browser allows it, and the photo never leaves your device. The tool downloads only the model file (about 100 MB, from Hugging Face, once) and the small program that runs it (from jsDelivr). Your browser keeps both for next time.

Why does the first photo take longer?

The first time, the browser downloads the model (about 100 MB) and starts it. After that it stays on your device, so the next visit starts in seconds. On a computer with WebGPU, a photo then takes a second or two. Without it, the model runs on your processor and takes much longer, so Automatic picks the smaller, faster model for you.

Which AI model does it use, and can I use the results commercially?

BiRefNet-lite, an open model by Peng Zheng and colleagues, released under the MIT license, which allows commercial use. The page names the exact files it downloads. The cut-outs are your pictures. We claim nothing on them.

Save what you find in one click.

Kortex is a Chrome extension for Gemini Notebook (formerly NotebookLM). It sends AI chats, web pages, posts and YouTube videos into the notebook you pick. It keeps your notebooks organized. It exports them back out.

Add to Chrome

No card needed.

Talk to the Kortex team

This opens a pre-filled email in Gmail. We usually reply within a day.