Background Blur

Keep your subject sharp and blur everything behind it for a natural portrait-mode depth effect.

Processed locally — your images never leave this device

Background Blur keeps your subject perfectly sharp while blurring everything behind it, reproducing the portrait-mode depth effect of a fast lens. It is the quickest way to clean up a busy or distracting background in a headshot, team photo or product shot without having to reshoot.

The model identifies the subject automatically, so there is no manual masking, no magic wand and no lasso work. Load the image, wait a few seconds for the first run to fetch and warm up the model, and the cutout appears. Because processing is local, the first image is the slowest — subsequent images reuse the already-loaded model and finish much faster. Fine hair, fur and semi-transparent edges are handled with soft alpha rather than a hard jagged cut.

Nothing you open here is uploaded. The entire operation runs inside your browser tab using the Canvas API and WebAssembly, which means there is no upload queue, no server-side file-size cap, no watermark and no account. You can even go offline after the page has loaded and the tool will keep working. When you close the tab, every trace of your file disappears from memory — which is exactly what you want when you are handling ID documents, contracts, private photos or unreleased product shots.

Because the work happens on your own hardware, processing is effectively instant for typical photos and still comfortable for 50-megapixel files on a modern laptop or phone. There are no daily limits, no paywalls after five files, and no queue behind other users. Everything is free, unlimited and ad-supported, and the layout reserves fixed space for ads so the page never jumps while you work.

How to use this tool

  1. Upload a photo with a clearly identifiable subject — a person, product or animal.
  2. Wait a few seconds while the segmentation model runs locally on your device.
  3. Set the blur strength applied behind the subject.
  4. Fine-tune the result and switch between transparent and filled output.
  5. Download the finished image as PNG (transparent) or JPG (flattened).

Technical specifications

Input formatsJPG, PNG, WebP
Output formatsTransparent PNG or flattened JPG
EngineIn-browser WebAssembly segmentation model
Manual workNone — the subject is detected automatically
ResolutionCutout returned at the source resolution
Privacy100% client-side — files never leave your device
Browser compatibilityChrome, Edge, Firefox, Safari 16+ (desktop & mobile)

Frequently asked questions

Yes. There is no credit system, no watermark and no account. The model runs on your device, so we have no per-image cost to pass on.