Blur Faces in Photo Online

How the Face Blur works

How to blur faces in photo online without uploading

Blurring faces by hand means drawing a box over each one and hoping you did not miss anyone in the background. This tool does the finding for you. A face detection model examines the image and returns the position of every face it is confident about, up to twenty in one picture, and each of those regions is then obscured with the style you choose. The detector runs in your browser through WebAssembly, so the image stays local, which matters when the whole point is protecting someone's identity.

Hide faces with picks the treatment. Blur applies a Gaussian blur scaled to the size of the face, so a large face in the foreground gets a proportionally heavier blur than a small one at the back. Pixelate replaces the region with large square blocks, the familiar look from television, which is harder to reverse than a light blur. Solid box paints the region black, which is the only option that removes the information entirely rather than degrading it; use it when anonymity actually matters rather than when you just want faces softened. Strength sets how heavy the blur or how large the pixels are, again relative to face size.

The detector returns the face itself, from brow to chin, and a blur that stops exactly there leaves hair, ears and forehead visible, which can still make someone recognisable. Margin expands the region outward by a percentage of the face size, with extra room above the eyes so the forehead and hairline are covered. Twenty percent suits most photos; raise it for a full head, lower it if the blur is spilling onto a neighbour. The stats say how many faces were found, and if the count looks low, faces that are very small, turned well away from the camera or partly hidden may not be detected, so check the result before sharing. Very large images are scaled to 2400 pixels on the long side before processing, and the output is a PNG.

Related tools