How Does Pixelated Censorship Work and Why is Uncensoring a Pixelated Image Hard/Impossible to Do?



I’m not entirely sure how this works.

In: Technology

Say you have yourself a freshly finished painting of something. To “pixelate” it, you first subdivide the painting into regions. Within each region, you take the brush and smear it completely into a uniform colour in that area, and repeat for every region. Once smeared, the previous state and information is lost, and you’ll require a lot of work as well as likely other information in order to restore the previous, higher quality state.

Many different methods are available, one of them is to reduce the number of pixels in the area.

say if there was 100 pixels in an area, you reduce that to 10 pixels. So that some information is still there but the rest no longer exists.

Because you have written over the data(image) with new data (pixelation). The original data(image) no longer exists.

Much like whisking an egg, you can’t ever get just the yoke seperate again.

Pixilation works by basically averaging all the pixels in an area to come up with a single color, then filling the area all those pixels covered with that single color.

For example, suppose I have the numbers: 19, 43, 46, 10, 47, 26, and 15. I average them together to get 29.

Now given the number 29 can you figure out what all the individual pixel values were? Of course not, information was lost with the averaging operation. In this same way you can’t reverse the process of pixilation because it removes information from the image.

Lets talk about numbers, because it’s easier and pixels are just numbers anyway.

When you pixilate something, you take a couple pixels and average them together. So if we have the numbers 2, 5, 10, and 15, you average them together and get 8. Easy, right? But now, If I tell you that four numbers average to 8, can you tell me what the original four numbers were?

Bob (secretly) takes 4 adds 6 and 3 and 8 to it and then tells you the average value (arithmetic mean) is nearest to 5.

Next Alice asks you what the four numbers were that Bob was adding together and you have to guess them in the right order only knowing that the mean is about 5.

This is (approximately) what happens with a single pixelated image block if you interpret numbers that Bob knows as the brightness of the real image and the average as the pixel brightness you see.

In a video consisting of multiple sequential pixelated images with movement, there maybe slightly more information that Bob tells you but it is still not enough to get back to 4, 6, 3, 8.

Pixelation is usually done by taking a region of the image, scaling it down and then scaling it back up. It’s like if you open an image in MS Paint and shrink it by 90%, save it, and then scale back to its original size.

The scaling down causes loss of information because there are fewer pixels to display the image. When you scale back up you do not get this information back. It’s also why images look worse the further you zoom in.