Pixelate Image

Drag over just the part that matters

Almost nobody wants a whole pixelated photograph. What people want is one thing hidden: a face in a group shot, a number plate, a house number, a name or email in a screenshot, a signature on a document being shared.

Switch to Region mode and drag a rectangle over that part. Everything else stays exactly as it was, at full quality. The preview shows the result before you commit, so you can check the box actually covers what you thought it covered — which is worth doing, because a box that clips the edge of a face or leaves half a digit visible is the most common mistake here.

Full image mode is there for the other reason people pixelate: as an effect. A deliberately blocky image for a thumbnail, a retro look, an abstract background.

What pixelation actually removes — and what it does not

This matters if you are doing it for privacy, and it is worth being precise, because a lot of "censoring" tools are vague here.

The original pixels are genuinely gone from the file you download. Each block is replaced by the average colour of the pixels it contained, and the image is re-encoded from that. There is no hidden original layer underneath, nothing to peel off, nothing to extract. That is a real and important difference from tools that simply draw a shape over an image or a page — where the thing underneath is still in the file and can be recovered.

What remains is the block averages, and that is where the residual risk lives. It is not extraction, it is inference. With small blocks, enough structure survives that a determined person can narrow down what was there — and this is most true of text and numbers, because the set of possible characters is small and a rough shape can be enough to guess between them. A lightly pixelated number plate or postcode is not as private as it looks.

So the rule is block size, and it depends what you are hiding.

Choosing the block size

The default of 8 is an effect setting. It looks pixelated, and on a face in a large photo it hides the person's expression but not necessarily their identity.

For hiding a face, go big enough that the region becomes a handful of flat squares rather than a recognisable mosaic — often 20 to 40 on a normal photo, more if the face is large in frame. The test is simple: look at the preview and ask whether you could tell who it was if you did not already know.

For text, numbers, plates and addresses, go further. Pixelation that leaves any hint of character shapes is not enough. If the region ends up as two or three uniform blocks, the information is gone.

And for anything genuinely sensitive — a document you are publishing, a screenshot going to strangers, anything involving someone else's safety — do not pixelate at all. Cover it with a solid shape. A filled rectangle carries no information whatsoever, which is a guarantee that no block average can give you. Pixelation is a style; a solid block is removal.

A privacy bonus worth knowing

Because the image is re-encoded from a canvas, the output carries no EXIF metadata — the camera model, the timestamp and, most importantly, the GPS coordinates are all dropped.

That matters more than people realise. A photo taken on a phone often has the exact location embedded in it, so pixelating someone's face while sharing the original file can hide their appearance and publish their address at the same time. Here the coordinates go with the face.

Nothing is uploaded, and here it is the point

The images people pixelate are the ones that need care: photographs with other people's children in them, screenshots of private messages, documents with account numbers, pictures where the location is the thing being hidden. Uploading those to a server in order to make them more private is a strange bargain, and it is the one every upload-based censoring tool asks for.

One format note: the result is saved as a JPEG, which is right for photographs. If your source is a PNG with a transparent background, that transparency will be filled in — this is a tool for photos rather than graphics.

Related tools

Add Text to Image can place a solid shape or label over an area, which is the stronger option when information must be genuinely gone. Crop Image removes the edge of a photo entirely — often better than hiding something near the border — and Compress Image reduces the finished file.