Hide faces, text and numbers in a picture. Runs on your device.
Pixelating turns a patch of a picture into a grid of flat squares, each one the average colour of the pixels under it. People use it to hide a face in a street photo, to blur out a name in a screenshot, or for the chunky retro look of a mosaic. This page opens with the Pixelate effect already selected, so you draw a box over the area, set the block size and download. Blocks run from 4 to 64 pixels on a side. A bigger block throws away more detail: a 40 pixel block over a face leaves only a few coloured squares, while 6 pixels still shows the outline of eyes and letters. When you add a picture the tool suggests a block of about one hundredth of the longest edge, at least 8 pixels. You can pixelate several areas and give each its own block size. One limit matters if the aim is to hide something private. Small blocks can sometimes be partly reversed, so for passwords, ID numbers or addresses switch the effect to Solid block, which replaces the pixels with one colour.
Solid block paints the area in one flat colour, black to start with. Nothing of the original is left under it, so it is the choice for a password, an ID or account number, an address or a face you must not show. Pixelate cuts the area into square cells and gives each cell the average colour of the pixels in it. Blur smears the area softly. Both keep a rough impression of what was there, which is why they look more natural on a face in a holiday photo and why they are the wrong choice for anything sensitive. Each region has its own effect, so one picture can have a solid block over a card number and a blur over a bystander. Regions are applied in the order of the list, each on top of the one before.
A blur or a coarse pixelation does not delete information, it mixes it. Short text, a number plate or a small face can sometimes be partly recovered when the blur is light, because the mixing follows a known pattern and software can work backwards from it. A bigger block size or a larger blur radius makes this harder, but nobody can promise where the safe point is for a given picture. Solid block has no such problem: the pixels are replaced, not changed, so there is nothing to work back from. This page says so next to the effect choice for that reason. If a picture would cause harm in the wrong hands, use Solid block, and pick a block size or radius only for pictures where a soft look matters more than certainty.
The file you download is drawn again from pixels and saved as a new image. The original pixels under every region are replaced in that new image, and the old ones are not stored anywhere in it. The camera details, the GPS location, the capture time and the other metadata that phones and cameras write into a photo are not copied, so they are gone too. A JPEG is saved as a JPEG at quality 0.92; a PNG or WebP is saved as PNG. The picture keeps its pixel size unless it is larger than your device can handle, and the page tells you before you save. The rotation flag that phones store is applied first, so the regions land on the same content you saw. A transparent PNG under a solid block becomes opaque in that area.
Your picture is opened by your browser's own image code and drawn on a canvas inside this tab. There is no upload step and no server that receives pictures. The page counts visits with Google Analytics; those counts say which tool was opened, never a file name, an image, a position or a size. While you work, the page shows a smaller copy so dragging stays smooth; when you save, the full-size picture is opened again and the regions are applied to it. Memory sets the limits: single images up to 100 megapixels, or 50 on a phone. This tool does not find faces or plates by itself, and it does not work on PDFs; to hide something on a PDF page, turn the page into an image first and hide it there.
Start with the suggested size, about one hundredth of the picture's longest edge, which is 40 pixels on a 4000 pixel wide photo. If eyes or the shape of the mouth still show in the preview, raise it toward 64. A face that fills 300 pixels is unreadable at 40 pixels, but a small face in a group photo needs a box tightly around it.
Not exactly, but it can sometimes be partly reversed. Each block keeps the average of the pixels it replaced, and software can compare many such blocks, for example across a video or a font known to be used, to guess the hidden content. Large blocks are much harder to reverse than small ones. For text such as a card number, use Solid block instead.
No. Only the boxes you draw are changed. Everything outside them keeps its original pixels. The JPEG is saved again at quality 92, so it is redrawn as a whole, and location and camera details are not carried into the saved file.
Sometimes, in part. A light blur or fine pixelation over a short number, a licence plate or small text can be partly recovered by software. A solid block cannot, because the pixels are replaced. For a password, an ID number or an address, use Solid block.
No. The file is drawn again from pixels, so the GPS location, camera model and capture date from a phone photo are not in it. A JPEG comes out as a JPEG at quality 0.92, and a PNG or WebP comes out as a PNG.
No. You draw each rectangle yourself. There is no detection, so check every face or plate in the picture before you share it. A new image opens with one example area in the middle, which you can move, resize or delete.
Not here. This tool works on JPEG, PNG and WebP. To hide text on a PDF page, turn the page into an image with the PDF to JPG tool, hide the area on the image, then turn the image back into a PDF. The text is then a picture, so it cannot be selected.
Touch and drag on the picture to draw a rectangle. Drag a rectangle to move it, or one of its handles to resize it. Each handle has a 44 pixel touch area. Touch outside the picture to scroll the page. An image can hold up to 50 regions. The Undo and Redo buttons keep up to 100 steps.
Yes. Choose Add region, then press an arrow key to move the region by about 1 display pixel. Shift with an arrow resizes it, Alt makes each step ten times larger, and Delete removes it. Ctrl or Command with Z undoes the last step.