How to blur faces in a photo

The quick way

1. Open Blur Faces and drop a photo. Faces are detected automatically on your device.
2. Review the boxes. Add boxes for any faces that were missed, and move, resize or delete the rest.
3. Choose Blur, Pixelate or Solid, an ellipse or rectangle shape, and the strength.
4. Download the result. The covered pixels are replaced in the saved file and all metadata is removed.

How the detection works

Face detection uses UltraFace, a small neural network that runs in your browser with ONNX Runtime and WebAssembly. Thorough mode, about 1.6 MB downloaded once, checks the whole photo and overlapping tiles to catch small faces. Quick mode, about 1.3 MB, is faster and suits close-up portraits. The model file comes from this site and is cached by your browser; the photo is never sent anywhere.

It will miss some faces

No detector finds every face. Small faces in crowds, faces in profile or turned away, partly covered faces and faces in poor light are the usual misses, and posters or statues can be picked up as faces. Look over the whole photo before you download, and add boxes by hand where needed.

Blur, pixelate or solid?

• Blur looks natural and suits group photos where you want to avoid identification at a glance.
• Pixelate looks more obviously deliberate. Use a strong setting, because light pixelation can leave a face recognisable to people who know the person.
• Solid masks remove the face completely. Use them when identification could put someone at risk.

Faces are not the only giveaway

Name badges, school uniforms, tattoos, number plates and distinctive clothing can identify someone as well. Redact Image can cover any area with solid boxes, pixelation or blur, and Remove Metadata strips the time and place a photo was taken.