In 1996 Steven Worley introduced a noise function built not from smoothed
random gradients (as in Perlin noise) but from a scattered set of
feature points. For every point on a surface, the algorithm
finds the distance to the nearest feature point (F1) — and often
the second-nearest too (F2) — and uses those distances to shade
the surface. The result looks like cracked mud, cells under a microscope,
reptile scale, or stone mosaic, depending on how you combine the distances.
Worley noise is also called "cellular noise" or "Voronoi noise" because the F1 boundaries are exactly a Voronoi diagram. It is widely used in film and game production for stone, water caustics, animal skin, and alien organic surfaces — often layered with Perlin/simplex noise for extra detail.
A tiled 3D surface shaded and displaced by Worley (cellular) noise: the value at every point is derived from its distance to the nearest scattered feature points, producing cell-like, cracked-earth or mosaic patterns.
Each surface point is shaded by its distance to the nearest (F1) and second-nearest (F2) feature points. Different combinations of these distances — and different distance metrics — produce very different cell shapes and edge textures.
Adjust the number of feature points, pick a distance metric and noise mode, and drag the relief slider to push the pattern into 3D terrain. Toggle the point markers or animate them to see the cells shift live.
Worley noise's F1 boundaries are mathematically identical to a Voronoi diagram — the same construction used to model soap-bubble foam, giraffe spots, and cracked mudflats in nature.