[Instantnoodl]

Sorry for the confusion. The use-case is a little difference because the goal is to display the image as close to the original as possible with the limitation of only being able to use a forground color, background color and character per cell. The character is selected based on it's shape just like in the article. So if you get rid of the colors in Chafa you end up with something similar to the article. That's what I wanted to say :D

[araes]

Cool, and thanks for the explanation. Gotten interested in retro software recently, so may actually be helpful for trying to set up pictures in some of the retro consoles. Most do tend to be limited to foreground / background. The stuff listed here [1] is pretty representative of what's being dealt with.

[1] https://en.wikipedia.org/wiki/List_of_8-bit_computer_hardwar...

Note: If you happen to know how to do multi-color dithering with some of these that would actually make significant improvements on some of these old picture hardware tests.

[Instantnoodl]

Isn't your problem more about Color quantization than about dithering? If you have big character cells like in a terminal dithering won't help you much. For each cell you want to find the best shaped cell and a foreground and background colour that are the closest to a colour from the supported palette.

But maybe I didn't understand your real problem yet