Two post-processing nodes for ComfyUI: Dither 1-bit turns an image into two colors, ASCII Art turns it into characters in a monospace font. No model, no GPU, pure NumPy + Pillow. Both nodes keep the input resolution and output a mask.

Dither 1-bit

Patterns: diagonal stripes, reverse diagonal, horizontal and vertical lines, crosshatch, Bayer, halftone dots, Floyd-Steinberg, Atkinson. Adjustable pixel size, pattern size, smoothing, black point, white point, gamma, invert and both colors.

ASCII Art

Maps brightness to characters from a charset of your choice (dark to bright), optionally sorted by real glyph density. Adjustable font size, line spacing, black point, white point, gamma, invert and colors. Color modes: fade, fixed or taken from the source image. Uses Consolas by default, any .ttf/.otf can be set. Also outputs the plain ASCII text.

Install

Extract the zip into ComfyUI/custom_nodes/ and restart ComfyUI. You will find the nodes under X4Lch3mist/image.

All my node zips share the folder ComfyUI-X4Lch3mist, so you can combine them. The complete package is on GitHub: https://github.com/X4Lch3mist/ComfyUI-X4Lch3mist

If you switch to the GitHub version later, delete the ComfyUI-X4Lch3mist folder first.

Tips

Dark images: start with black_point 0.15–0.25 so the background stays clean.

Hard diagonal stripes: diagonal stripes, pixel_size 2, pattern_size 6–10, smoothing 0–0.2.

Newspaper look: halftone dots, pixel_size 1, pattern_size 6.

License: GPL-3.0