Skip to content
OneTooling

Blur Background

Blur the background of a photo like portrait mode, or swap it for a color. A small AI model finds the subject on your device. Nothing is uploaded.

Runs in your browser · nothing uploaded

Drop a file here, or

JPG, PNG, WebP, AVIF. Works best on photos with one clear subject, like a person or a product.

The first time, your browser downloads a small AI model (about 5 MB) and its engine (about 14 MB) from this site. They are cached, so later visits start quickly. The photo itself never leaves your device.

Why is there an ad? This tool is free and runs on your device. One small ad pays for hosting and new tools. No pop-ups, no autoplay, never inside the tool. How we make money

How to blur the background of a photo

  1. Choose a photo with a clear subject. The first time, the model downloads.
  2. Wait a moment while the subject is found.
  3. Set the blur strength, or switch to a solid color, and refine the edge if needed.
  4. Choose a format and press Download full size.

Blur strength is a share of the picture size, so a photo and its thumbnail look alike. JPG is a good choice for photos; use PNG for exact quality.

Questions

Is my photo uploaded?

No. A small AI model runs on your device through WebAssembly. Only the model files are downloaded from this site, once, and your browser keeps them for next time.

How big is the first download?

About 5 MB for the model and 14 MB for the engine that runs it. After that they load from your browser cache.

Part of the subject got blurred, or some background stayed sharp. What can I do?

Use the cut-off slider: raise it to blur more of the background, lower it to protect more of the subject. Edge softness and feather control how smoothly sharp meets blurred.

Does it work on any photo?

It works best on photos with one clear subject, such as people, pets, and products. Crowded scenes or subjects that match the background may need edge adjustments.

Is the saved photo full resolution?

Yes. The subject is found on a smaller version, then the blur and mask are applied to the original at full size.

Credits

Built on U2-Net (u2netp model) (Apache-2.0) and ONNX Runtime Web (MIT). OneTooling is not affiliated with these projects.