This is created by an algorithm inspired by fejesjoco (in the same challenge). Start with a black pixel in the center. Iterate every pixel in the image, from closest to the center to furthest from the enter. For each pixel, calculate the average color of all placed pixels out of the 8 neighboring pixels. Find colors in the color cube that are not already taken and as close as possible to the calculated color. The closeness measure is the distance between gray-scale values. If multiple equally close colors are found, take the ones closest to black, using the negative-infinity norm (min(r, g, b)). If multiple of these are found, choose a random one. Place the found color at the found point.