Hi HN!
This is a repo to convert images into ASCII and ANSI art.
The general technical approach is to use CLIP as a judge on the ASCII art (akin to CLIPasso, for those familiar), while trying to make the glyph space differentiable in a sensible way.
I've tested it on macOS and Linux/CUDA. A render takes about an hour and 8 Gb of memory, there's a fast preview mode if you want to watch the optimization process make progress.
Paper writeup is in progress -- any feedback would be welcome :)