Pixelate a face in a photo

Cover a number plate, an address or a screen with solid blocks, or pixelate the whole frame for a retro look, without uploading the photo anywhere.

Drop your files here

.jpg, .jpeg, .png, .webp · up to 50 MB

Choose files

How to Pixelate Image

1
Add your image
JPG, PNG or WebP up to 50 MB, one photo at a time. The format is read from the file itself, so a photo saved with the wrong extension still lands in the right decoder.
2
Set the area and the block size
Under “What to hide”, keep “A selected area” and drag the frame over what must go, or pick “The whole image”. “Area shape” switches the frame between a rectangle and an ellipse. Bigger blocks hide more; “Block size” runs from 4 to 80 pixels.
3
Download the result
The stage shows the blocks as you change the settings. The file keeps the size and the format of the photo, with the EXIF block, including any GPS coordinates, dropped.

You place the ellipse, nothing finds the face

The page opens with “A selected area” under “What to hide”, “Ellipse” under “Area shape” and “Block size” at 24px. The ellipse starts in the middle, half the photo's width and half its height. The page does not detect faces, so drag the ellipse onto the face yourself. Pull a corner to resize it, or nudge it with the arrow keys; the blocks appear right under it. “More settings” takes exact numbers in percent of the photo: “Area left edge”, “Area top edge”, “Area width” and “Area height”. One pass hides one area.

How many blocks across the face

“Block size” counts the photo's own pixels, so what matters is how many blocks fit across the face. On October 21, 2026, we tried two phone photos of a mannequin head. With about 20 blocks across, the eyes, nostrils and lip shapes were still easy to make out. With 10 to 12, the eyes were two dark blocks and the mouth a faint band. With 7 or fewer, only darker patches marked the eyes. So make a block at least a tenth of the face's width, larger if an eye still keeps its shape. The ellipse gives that width: “Area width” 45 on a photo 1280 pixels wide is about 576 pixels, so 58 or more.

Why 24 pixels hides little on a phone photo

In the closer photo, 3024 × 4032 pixels, the face was 1,350 pixels wide. The default 24-pixel blocks put 56 across it, and even the eyebrows stayed visible. The face detector built into macOS still found the face on both photos. At 80, the largest “Block size”, 17 blocks remained, and the eyes, nose and lips still read clearly. A tenth of 1,350 is 135, past the end of the slider. So shrink a close-up first in Resize Image: “Exact pixels”, “Width” 1280, “Download JPG”. Then open this page and press “Use it” on the line “from your last step is still here”. In a 1280-pixel copy the face was 577 pixels wide, and 64-pixel blocks left 9 across.

One photo per pass, and what the file keeps

The page takes one photo at a time: drop two, and only the first opens. For a second face, download the result, press “Replace file” and add that copy. The ellipse and block size then start over. The result keeps the photo's size and format, a JPEG at quality 92, with -pixelated added to the name. Our 1.9 MB photo came back upright as a 1.6 MB JPG, about two seconds after “Download JPG”. Camera model and capture date were gone, and so was a GPS position we had added to test copies. The page makes a mosaic, not a blur, and only inside the ellipse: hair, clothes and background stay as they were. A person near an edge can be cropped out with Crop Image. The photo stays in your browser.

The photo never leaves your device

Decoding, averaging and re-encoding all happen inside this browser tab. Nothing is uploaded and there is no server queue. To continue in another tool, PulsarKit may store a local copy of your input or result in this browser. It can remain after you close the tab or restart the browser. Clear PulsarKit’s site data to delete it. The EXIF block is deliberately not copied into the result: a photo you pixelated to hide an address would otherwise still carry the GPS coordinates of where it was taken.

Frequently asked questions

Does my photo get uploaded to a server?

No. The file is read, decoded, pixelated and encoded again entirely in your browser using WebAssembly codecs. There is no upload step. To continue in another tool, PulsarKit may store a local copy of your input or result in this browser. It can remain after you close the tab or restart the browser. Clear PulsarKit’s site data to delete it.

Can pixelation be undone?

Not by this tool. The averaged blocks genuinely replace the original pixels and the original values are gone. In theory, someone who knows what the hidden content looked like can narrow down the possibilities from the block averages, which is why block size matters. Choose blocks large enough that a single block covers more than one character.

What is the difference between pixelating and blurring?

Blurring spreads each pixel into its neighbours and leaves a smooth, recoverable-looking smear; pixelation replaces whole square regions with a single flat colour. Pixelation destroys more information for the same visual area and reads unambiguously as "this was hidden on purpose", which is usually what you want on a screenshot or a document photo.

What block size is safe for hiding text?

As a rule of thumb, make a block wide enough to cover at least two characters. On a full-resolution phone screenshot that is around 24 to 32 pixels; on a small crop it may be 12. If you can still make out the shape of individual letters in the result, the blocks are too small.

Can I pixelate only part of the picture?

Yes, that is how the page opens: “A selected area” under “What to hide”, with a frame over the middle of the photo. Drag it over what must go, and pick “Rectangle” or “Ellipse” under “Area shape”. Outside the frame the picture stays as it was. In our check on October 21, 2026, a PNG kept every pixel outside the frame exactly; a JPEG, saved once more at quality 92, shifted them by under two levels out of 255 on average.

Are GPS coordinates removed?

Yes. The result is written without the EXIF block, so camera make, model, timestamp and any embedded location are gone. If you also want to clean up other files, Remove Image Metadata does the same job without touching the pixels.

How many files can I do at once?

One at a time, up to 50 MB: drop several and only the first one opens. For the next photo, use “Replace file”; the settings go back to what the page opened with. To hide a second spot in the same photo, open the downloaded result with “Replace file” and frame the next one.

Does this work on a phone?

Yes. Everything runs in the browser on iOS and Android as well as on desktop. A 3000 by 2000 photo takes well under a second; very large images take proportionally longer.