Median Cut vs K-Means vs NeuQuant vs Octree

Pixel conversion samples a photo onto a grid of cells, then paints each cell from a smaller palette. The grid decides how coarse the picture is. The color budget decides how many distinct colors remain.

Each cell takes an average of the photo underneath it. Palette reduction groups those averages into graphic blocks, so the result reads as pixel art instead of a tiny photograph. Dithering is optional. It mixes neighboring colors so gradients stay readable after the palette shrinks.

Set the cell width and height, pick a styled look, and set max colors. Leave dithering off for flat regions, or turn it on when soft shading matters. You still recolor the palette and paint on the grid. For the full convert-then-edit loop, see photo to pixel art.