algorithmic modeling for Rhino
Testing four different dithering algorithms written for Grasshopper 2.0 In each case the palette is created by putting all the colours into a kd-tree, then picking the most densely populated leaf-nodes and using the mean colour in those nodes. This allows for the creation of palettes with any amount of unique colours. It is also considerably faster than a k-means approach, although it is probably a lot slower than a median cut approach. Median cuts though seem to only be able to generate palettes with sizes that are powers of two.