CalcTimer

Blur Faces in Photos

Every face in the photo is found and covered automatically. Tap your own face to keep it visible, work through a batch of photos at once, and nothing is uploaded to a server.

How to use Blur Faces in Photos

  1. Choose photos to work on. You can pick several at once or drop them onto the box.
  2. Every face found is covered straight away. The first run downloads the detection model, which takes a few seconds.
  3. Tap the face you want to keep, such as your own. Its outline turns yellow and the cover comes off. Tap again to put it back.
  4. For a missed face, or for a name badge or licence plate, press Add area and drag over it.
  5. Press Save this photo, or Save all (ZIP) when you have a batch.

Keeping one face visible

Photos start with every detected face already covered. From there you tap only the faces you want to show. A tapped face gets a yellow outline and a keep label; tap it once more and the cover returns.

There is a reason the tool does not try to recognise your face by itself. Recognition means storing facial features and matching them photo by photo, and a wrong match uncovers a stranger. Covering everything first and uncovering what you choose costs one tap and cannot expose someone by mistake.

Faces the detector tends to miss

Detection runs over the whole photo and again across overlapping tiles, so small faces in a group shot are usually picked up. Even so, the cases below get missed or mis-detected. Leave Show outlines on and look over the photo before saving.

SituationWhat happensWhat to do
Full profile, head bowedEyes, nose and mouth are hidden, so it may be missedAdd the area by hand
Very small, far-away facesA face only a few pixels wide can be missedAdd the area by hand
Masks, sunglasses, a hand over the faceThe confidence drops and it may be skippedAdd the area by hand
Faces on posters, photos, a TV screenTreated as real faces and coveredTap to uncover if it does not matter
Patterns that look like a faceOccasionally covered by mistakeTap to uncover

Mosaic or blur, and how strong

Mosaic breaks the area into square blocks; blur softens it inside an oval. Strength sets how many blocks the longer side is divided into, so fewer blocks means a heavier cover.

StrengthBlocks across the long sideResult
114Outline still readable; someone who knows the person may recognise them
211Lightly covered
3 (default)8Good for posting; hard to identify
46Firmly covered
54Only blocks of colour remain

For children and for strangers, use strength 3 or higher. A light blur leaves the shape of the face and the hair, which is often enough to recognise someone.

Cover a little more than the face

The detected area runs from the eyebrows to the chin. The Coverage slider grows it outwards, with a little extra at the top for the forehead and hair. Hairstyles and earrings identify people too, so going below the default of 25% is rarely worth it. An area you drag in yourself is covered exactly as drawn.

Before you post the photo

Posting a recognisable face without consent can become a real problem, especially with other people's children, diners at the next table, or spectators at an event. Faces are not the only giveaway.

Worth covering too: name badges and school uniform tags, licence plates, apartment or door numbers, delivery labels. Drag over each of them with Add area. What the law requires varies by country and by situation, so check locally if a photo worries you.

Photos never leave your device

Detection, covering and saving all happen inside your browser, and the photo is never sent to any server. The first run downloads a face detection program (Google's open MediaPipe face detector), and that program also runs entirely on your device.

The saved file carries none of the original capture data (EXIF). The GPS location stored by phone cameras goes with it, so a photo cannot quietly give away where you live.

Format and resolution

A JPG is saved as JPG at 92% quality, a PNG as PNG, and -mosaic is added to the file name. Resolution matches the original; only pictures above 16 megapixels are scaled down, because of the canvas limit on phones. On a phone, Share / save to Photos puts the result straight into your gallery where the browser supports it. An iPhone HEIC file copied to a computer may not open in some browsers, so set the camera format to Most Compatible or export it as JPG first.

Frequently asked questions

Are my photos uploaded to a server?

No. Detection, covering and saving all happen inside your browser. Only the detection model is downloaded the first time; the photos are never sent anywhere.

How do I keep my own face visible?

Every face starts covered. Tap your own face once and the cover comes off, with a yellow outline marking it. Tap again to cover it back up.

A face was missed, or something that is not a face got covered.

For a missed face, press Add area and drag over it. For anything covered by mistake, tap it once to uncover it. Undo reverses your last action.

Can I do several photos at once?

Yes. Choose a batch and every photo is checked automatically. Use the thumbnails at the top to review each one, then Save all (ZIP) to download them together.

Can a covered photo be reversed?

No. In the saved file the pixels underneath are gone. Your original photo is still on your device, so check the result before deleting it.

Does it work on a phone?

Yes. Tap faces to cover or uncover them, and press Add area to drag over anything else. Large batches take longer than on a computer.

Can I do this for a video?

Photos only for now. Video face blurring is in the works.