Docuboxer
By ··5 min read

Blur faces in a photo before you post it

How to blur or pixelate faces in a photo before you post it: which method is safest, how to do it in your browser, and what data stays behind.

To blur faces in a photo, open Blur faces in photos, drop in the image, and review the boxes the browser draws. Detection and the effect both run in your browser, so the photo is never uploaded. One limit to know: the detector can miss faces in profile, covered, or very small, and light pixelation or blur can sometimes be reversed.

Pixelate, blur, or block: which is safest?

A solid block is the safest option, because it erases the pixels in the marked area. Pixelation with large blocks is usually enough to keep anyone from recognizing the face at a glance. Light blur protects the least, because it keeps shapes and colors that can be reconstructed.

This is more than a hunch. In 2016, Hill, Zhou, Saul, and Shacham concluded that mosaicing and blurring are not viable ways to redact text. For faces, McPherson, Shokri, and Shmatikov showed that neural networks still identify pixelated or blurred faces. If the photo will be public, use a solid block.

  • Solid block: the strongest effect. It covers only the boxes you mark, so check that every face has one.
  • Pixelate: use large blocks. Small blocks keep more detail.
  • Blur: the weakest protection. It takes focus off a background, but it won't hide a person.

How do you blur faces without uploading the photo?

Open the tool, drop in the photo, and wait for the browser to find the faces. Each face appears in a box: add the ones that are missing, adjust the ones that aren't needed, and pick the effect. When you download, the copy comes out with no EXIF data, no location, and no thumbnail of the original.

  1. Open the tool: go to Blur faces in photos.
  2. Drop in the photo: the file stays in your browser. Photos over 24 megapixels are not processed: shrink them first with the image resizer.
  3. Check the boxes: add a box by hand for each face the detector missed.
  4. Choose the effect: blur, pixelate, or solid block, and the strength.
  5. Download the copy: the original photo is unchanged; the copy is what you share.

What a covered face doesn't hide

Covering a face is not enough if the rest of the photo identifies someone. A uniform, a badge, or a school sign can say more than a face. The original file also records the date, the camera model, and sometimes the GPS location. Check what it carries with the EXIF metadata viewer, and read what your photos reveal besides faces and how to remove the location from a photo.

Frequently asked questions

Does pixelating a face make it unrecognizable?

Not always. Large blocks usually stop people from recognizing the face at a glance, but that is no guarantee: in 2016, McPherson, Shokri, and Shmatikov trained neural networks that identify pixelated and blurred faces. If the face must be unrecognizable, use a solid block.

What happens to the photo's metadata?

The copy you download is re-encoded in your browser and loses the original's EXIF data, location and thumbnail. The original file still has them, so share only the copy.

Does it work with several faces in one photo?

Yes. The detector puts each face in its own box, and you can add, remove, or move boxes by hand. Check the whole photo before you download it: the detector can miss faces in profile, covered, or very small.

Is my photo uploaded to a server?

No. Detection and the effect run in your browser. The detection model is served from the site itself, and the photo is never sent to a server.

Is there a size limit?

Yes: photos over 24 megapixels are not processed. Shrink them first with Docuboxer's image resizer; the tool tells you if a photo is over the limit.

What if the detector misses a face?

Draw a box over it by hand. The tool itself suggests this for profiles, covered faces, and very small ones. Then check the copy before you post it.

Blur the faces now

No upload. Free, and it runs in your browser.

Blur faces →

Related tools

Sources