Photo Censor: Pixelate, Blur or Hide Part of an Image

About the Photo Censor

The Photo Censor hides the parts of a picture you choose. Drop an image on the page, drag a rectangle or an ellipse over a face, a number plate, a street sign or an address, and pick what to do with it: pixelate it into coarse blocks, blur it, or cover it with a solid colour. Each region has its own effect, so you can pixelate one detail and black out another. Everything happens in your browser: your photo never leaves your device, nothing is uploaded, and there is no account.

The tool does not find faces for you. There is no face detection, so only the regions you draw are hidden. Look over the whole picture, including reflections and the background, before you share it.

How to use it

The file is saved as <name>-censored.<ext>, so street.jpg saved as PNG becomes street-censored.png. The effect is written into the pixels of the saved picture. It is not a layer on top, so nobody who opens the file can lift it off.

The three effects

• Pixelate splits the region into square blocks and paints each block with the average colour of the pixels in it. A larger block size means a coarser mosaic.
• Blur smooths the region with a Gaussian blur. The radius is the standard deviation in pixels, measured on your full-size picture.
• Solid colour replaces every pixel in the region with one opaque colour. It is the safest choice for text and numbers.

The maths behind the regions

For pixelate, a block covering $n$ pixels with colour values $v_i$ and opacities $a_i$ gets the opacity-weighted average, rounded to the nearest whole number:

$\bar v = \frac{\sum_{i=1}^{n} a_i\, v_i}{\sum_{i=1}^{n} a_i}$

Blocks start at the top-left corner of the region. When the region is not a multiple of the block size, the last blocks use the pixels they have.

An ellipse region is the ellipse inscribed in its own rectangle. A pixel at $(x, y)$ is inside it when its centre satisfies

$\left(\frac{x + 0.5 - c_x}{r_x}\right)^2 + \left(\frac{y + 0.5 - c_y}{r_y}\right)^2 \le 1$

where $(c_x, c_y)$ is the centre of the rectangle and $r_x$, $r_y$ are half its width and height. The corners of the rectangle stay untouched.

Worked example

Take a region of 4 by 4 pixels pixelated with blocks of 2. The top-left block holds four fully opaque grey pixels with values 10, 20, 30 and 40. Their average is $(10 + 20 + 30 + 40) / 4 = 25$, so all four pixels become 25. The other three blocks are averaged the same way, and the result is a 2 by 2 grid of flat squares. With blocks of 3 on the same region the first block would cover 3 by 3 pixels, and the last row and column of blocks would be 1 pixel wide.

Choosing a setting you can trust

Pixelate and blur remove detail, but a light effect can leave enough for a person or a program to guess what was there, especially for short text or a few digits. Use a block size or radius large enough that nothing readable is left, or use Solid colour for anything sensitive. When two regions overlap, the later region is applied on top of the earlier one.

Preview and saved file

The preview is drawn from a copy of at most four million pixels so that dragging stays smooth. The saved file is made again from your original at full resolution, with the region positions and the block size and radius scaled to match. JPEG is lossy and has no transparency, so transparent areas are filled with the background colour you choose, white by default. PNG is lossless and WebP is usually the smallest. Animated GIF and WebP files are saved from their first frame.

Related tools

To change the look of the whole picture, use the Image Filter. To cut the picture down before you hide anything, use the Image Cropper. To switch file types, use the Image Converter.

Limits worth knowing

There is no automatic detection of faces, plates or text. EXIF data such as the camera model or GPS position is not kept in the saved file. Images above 32 megapixels are refused with a message rather than risking a crash.

Frequently asked questions