Hacker News Reader

I don't get it. How is his code treating pixels differently according to intensity?

Because the code iterates per character, not per pixel.

If a pixel's value is '7' and in the redacted zone, the program will iterate over the single character and output '0'. If the pixel's value is actually '174', the program will iterate over all three characters, outputting '000'. Any normal PPM image viewer will display these identically, but high contrast areas like text will be recoverable.

What image format stores pixel values in text?

The Netpbm family of image formats, including PPM, have both binary and ascii standards.

https://en.wikipedia.org/wiki/Netpbm#PPM_example

Notably these formats are designed to be lossless and very simple to work with programmatically so they were used as the intermediate format in a lot of image processing workflows, even where the inputs and outputs are a more common format like JPG (which is extremely hard to manipulate directly).

Reply on news.ycombinator.com