image tool
Pixelate image online
Choose a block size and the photo turns into a mosaic of squares, each one the average colour of the pixels under it. Switch to boxes to pixelate only a face, a name tag or a number plate, and the rest of the photo stays sharp.
Loading the tool…
- Runs on your device, nothing is uploaded
- Free, no sign in
- No watermark, no file limit
How it works
- 1
Drop the photo
The whole image is pixelated at 16 pixel blocks to start, the smallest size the page recommends for hiding text.
- 2
Mark what to hide
Press Only boxes I draw, then drag over each face, plate or name. Drag a box to move it; Delete removes the selected one.
- 3
Set the block size
Bigger blocks hide more. Under 16 pixels a warning appears, because small blocks over text can be read back.
- 4
Download
Show original compares, then Download saves the full-size result as PNG or JPG. Nothing is uploaded and nothing is stored, ever.
How the mosaic is made
The photo is divided into squares of the size you choose, starting from the top left corner. Every pixel in a square is replaced by the average colour of that square, so a 16 pixel block turns 256 pixels of detail into one flat colour. Squares at the right and bottom edges are cut short if the photo is not an exact multiple of the block size, and they are averaged over the pixels they actually cover. When you pixelate only boxes, each box is widened to whole blocks of the same grid, so the mosaic lines up with itself wherever boxes meet.
Why small blocks are not safe on text
A pixelated word still carries the average brightness of every block, and that pattern is different for every word. If an attacker can guess the font and the likely text, such as a surname, an account number or the letters on a plate, they can pixelate each guess the same way and look for the one that matches. Security researchers have published tools that do exactly this for small blocks. Larger blocks average away so much that many guesses look the same, which is why the page asks for 16 pixel blocks or more, and why a block should be at least as tall as the text itself.
Pixelate or blur
Both hide detail, and both can be attacked when the hidden thing is one of a few known possibilities. Pixelation at a large block size throws more away and makes it obvious to the reader that something was hidden on purpose, which suits plates, names and screens in a photo. A blur looks softer and suits faces in the background of a picture that is about something else. When the content is truly sensitive, such as a document number, do not rely on either: cover it with a solid box before sharing.
Limits worth knowing
Pixelation with small blocks over text can be recovered, so use 16 px blocks or larger, and larger again when the text is big in the frame. The warning under the slider appears whenever the block size is below 16. The grid is measured in pixels of the full-size image, so the preview is a scaled copy and the download is exact. Images over 8,192 pixels on the long side are refused with a plain message rather than freezing the tab.
Photos on consent and sign up pages
A classic form can carry image blocks with one of 10 border styles, and a photo of a past event makes a sign up page feel real. If people in that photo did not agree to appear, pixelate their faces here, then upload the hidden copy as the form's image.
Limits
- Pixelation with small blocks over text can be recovered; use 16 px blocks or larger. The tool shows a warning whenever the slider is below 16.
- Block size runs from 2 to 96 pixels of the full-size image, on a grid that starts at the top left corner of the photo.
- Boxes snap outward to whole blocks, so a pixelated area can be a few pixels bigger than the box you drew.
- Images larger than 8,192 pixels on the long side are refused with a message.
What it does not do
- It does not find faces or text for you. Every box is drawn by hand.
- It does not black out an area with a solid colour. For the strongest redaction, cover it completely in an editor.
- It does not keep the photo's EXIF data.
Questions people ask
What block size hides a face?
Aim for at least six to eight blocks across the face. A face 120 pixels wide in the full-size photo needs blocks of about 16 to 20 pixels; smaller faces in a crowd can use the same size because they are already covered by two or three blocks. Check the preview at full screen, and go bigger if you can still tell who it is.
Is pixelation reversible?
Not in the sense of rebuilding the original pixels: the averaging throws them away. But it can be guessed. When the hidden thing is text in a known font, or one of a few faces, the attacker pixelates every candidate and compares. Small blocks leave enough pattern to win that game, which is why the tool warns below 16 pixels.
Can I pixelate several areas at once?
Yes. Switch to Only boxes I draw and drag as many boxes as you need: one per face, plate or name badge. Every box uses the same block size and the same grid, so overlapping boxes merge cleanly. The readout under the preview counts the boxes, and Remove all boxes clears them if you want to start again.
Why is the pixelated area bigger than my box?
Blocks sit on one grid that starts at the top left of the photo, and a box is widened to whole blocks so that no block is cut in half. With 16 pixel blocks the area can grow by up to 15 pixels on each side. That makes the mosaic line up with itself and slightly over covers the thing you are hiding.
Does pixelating shrink the file?
A little, as PNG. Large flat blocks compress well, so a fully pixelated PNG is often much smaller than the original. As JPG the gain is smaller, and very small blocks can even grow the file because the hard block edges are expensive for JPG to store. The result line shows the saved size after each download.
Use the hidden photo on a consent page
Take what you made here into forms.
Create a form with OneCraftMore tools like this
- Blur image onlineBlur an image online with a Gaussian blur, all over or only inside boxes you drag, such as a face in a crowd. Runs on your device; the photo is never uploaded.
- Image cropperCrop with a draggable box and 44 pixel handles, locked shapes for square, 4:5, 9:16 and 16:9, thirds guides, quarter turns and arrow key nudging.
- Image compressorCompress to WebP or JPEG and compare the result against the original under a divider you drag. Or ask for a size in KB. Nothing is uploaded.
- Convert an imageSee your own picture measured in PNG, JPEG and WebP before you choose, then convert. Transparency is handled deliberately, and HEIC is explained.
- Convert image to black and whiteConvert an image to black and white using Rec. 709 luminance, add contrast with a slider, and save PNG or JPG. Runs in your browser, nothing is uploaded.
Background reading: the question page on this.
Written and checked by the OneCraft team. Last checked .