Blur background photo

Give any photo a soft, out-of-focus background. A depth model runs on your device, so you click your subject to set the focus and dial in how much blur you want — like a portrait lens, with no upload.

  1. Your photo

    Drop a photo here — or click to choose

    JPGPNGWEBP

    Your files are never uploaded — everything runs on your device.

  2. Depth model

  3. Adjust & export

How to blur the background of a photo

  1. Add your photo

    Drop a JPG, PNG or WebP. It loads straight into your browser — nothing is uploaded.

  2. Click what should stay sharp

    A depth model runs on your device and maps how near or far everything is. The first run downloads that model and the runtime it needs — on the order of 40 to 60 MB, depending on whether your browser uses its CPU or its GPU — and then caches it. Click your subject to set the focus, then dial the blur.

  3. Download

    Adjust the look live — the blur re-renders instantly — then download the result as a PNG.

What you get

Real depth, not a cut-out

A monocular depth model estimates the distance of every pixel, so the blur falls off smoothly with distance — like a real lens, not a hard mask.

Click-to-focus, live

Pick any focal point and watch it re-render instantly. The depth is computed once; changing focus or blur strength is just the shader.

Your photos are never uploaded

Depth estimation and rendering both run in your browser on your own device — no upload, no account, free and unlimited.

The model behind it

Powered by Depth-Anything-V2 (small) for the depth map and OpenCV YuNet for face detection, both permissively licensed. The blur follows estimated depth, so it falls off with distance instead of cutting a flat hole around the subject.

Frequently asked questions

How do I blur the background of a photo I have already taken?
Drop it in above. The tool estimates depth from the picture itself, so it does not need portrait-mode data from the camera — an ordinary photo taken months ago works the same as one taken today. Click your subject to set what stays sharp, then dial the blur and download.
How big is the first-run download?
On the order of 40 to 60 MB: the depth model plus the runtime that executes it. Which end you land on depends on whether your browser uses its CPU or its GPU — the GPU build is the larger, higher-precision one. It downloads once and is then cached, so every run after the first starts straight away, and the progress readout shows the real byte count while it happens.
Are my photos uploaded anywhere?
No. Both the depth estimation and the blur run entirely in your browser on your own device. Your photo is never uploaded, and nothing is kept once you close the tab.
How does it know what to keep sharp?
A depth model estimates how far away every pixel is, then you click your subject to set the focal plane. Everything nearer or farther than that gets progressively more blur, the way a real lens behaves.
What files can I use?
Any JPG, PNG or WebP photo. The result downloads as a PNG. Photos with a clear subject and some separation from the background look best.
Is it really free? Are there limits?
Yes — completely free, no account, no watermark and no per-photo limit. Because everything runs on your device, there is nothing for us to meter.
Which models does it use?
Depth-Anything-V2 small (Apache-2.0) estimates how far away each pixel is, and YuNet finds faces so the focus lands on the subject rather than the middle of the frame. Both run in your browser.