Kids, dither is ready! This release introduces a full dithering framework with a wide range of algorithms, a new preset palette, UI refinements, and a modernized build pipeline.

New Features

• Dithering Framework
  • Added a new dithering framework supporting multiple algorithms:
    • Random and Gaussian Random
    • Simple threshold
    • Error diffusion: Floyd-Steinberg, Atkinson, Jarvis–Judice–Ninke, Stucki, Burkes, Sierra, Sierra Lite, Two-Row Sierra
    • Ordered: Bayer 2x2 / 4x4 / 8x8, Cluster 4x4 / 8x8
  • Added a configurable error diffusion factor for fine-tuning diffusion-based ditherings.
  • Dithering is incompatible with the “keep original palette” option. This is due to the fact that all dithering algorithms require an indexed palette, and the only way to compute a full palette for a given image is to look up each pixel, which would be computationally infeasible.
• Added the Pico-8 preset palette.

Improvements

• Reorganized the input/output configuration into an accordion layout to save vertical space.
• More consistent use of localization throughout the UI (fixes #6).
• FH04 segmentation is now the default segmentation algorithm.

Build & Infrastructure

• Upgraded to JDK 26.
• Switched to Reckon for versioning.
• Updated project dependencies.
• Updated the Gradle wrapper.
• Added an automated release pipeline. (unfortunately, this isn’t functional in this release yet, but we hope to make it ready for the next one)
• Adjusted .gitignore.