FloydSteinbergvirheiden

Floyd–Steinberg dithering is a widely used error diffusion algorithm for converting images from a high-color or grayscale palette to a limited palette, commonly black and white or a small set of colors. The method preserves overall image luminance by distributing the quantization error of each processed pixel to neighboring pixels that have not yet been processed.

The algorithm processes the image in a top-left to bottom-right scan. For each pixel, it maps the original color to the nearest color in the target palette and computes the quantization error as the difference between the original and quantized colors. This error is then distributed to four neighboring pixels using the fixed weights: 7/16 to the pixel to the right, 3/16 to the bottom-left, 5/16 to the pixel below, and 1/16 to the bottom-right. The diffusion continues as the scan proceeds, ensuring that the total error is preserved across the image.

History and usage: Floyd–Steinberg dithering was introduced in 1976 by Robert Floyd and Louis Steinberg and has since become a standard technique in digital image processing for producing visually pleasing dithering patterns. It is commonly applied to grayscale conversion and to color images by processing each color channel separately or by mapping to a chosen palette prior to diffusion.

Variants and alternatives: The basic approach has inspired numerous diffusion kernels (such as Jarvis–Judice–Ninke, Stucki, Sierra, Atkinson, and Burkes) that differ in their diffusion patterns and efficiency. Floyd–Steinberg remains favored for its balance of simplicity, quality, and computational efficiency.

Limitations: The method can introduce visible dithering artifacts, especially in large uniform areas or near palette limits, and may require buffering or streaming approaches for real-time implementations.