dither.py

dither.py is a simple library I wrote to implement various dithering algorithms as a learning exercise. The following algorithms are currently implemented:

• Threshold - basic per-pixel quantization
• Ordered Dithering - quantization based on matrix templates
  • Bayer 4x4
  • Bayer 8x8
  • Cluster 4x4
  • Cluster 8x8
• Error Diffusion - adaptive forward quantization
  • Floyd-Steinberg
  • JaJuNi
  • Fan
  • Stucki
  • Burkes
  • Sierra
  • Sierra-2
  • Sierra Lite
  • Atkinson
• Randomized - randomized quantization
  • Per-pixel random
  • Block random

Several palettes are also available:

• Grayscale
  • 1-bit (black and white) through 7-bit (128 level)
• Gamma-corrected Commodore 64 (without the color limitation of either HiRes or MultiColor modes)
• CGA (using the RGBI monitor brown color)
  • Mode 4, low intensity
    • Palette 1
    • Palette 2
  • Mode 4, high intensity
    • Palette 1
    • Palette 2
  • Mode 5, low intensity
  • Mode 5, high intensity
• EGA
• Websafe (most common 6-bit variant)

Use dither.py to dither an input image with a given palette:

python dither.py -m floyd_steinberg -p ega images/parrot.jpg

or use the -a flag to create a collage with all combinations of methods and palettes: