Worley noise

Example picture generated with Worley noise's basic algorithm.
When computing Worley noise, each seed is located within its own grid square. (Grid is not a part of the noise)
Worley noise can be differentiated once to generate a normal map. (Grid is not a part of noise)

Worley noise, also called Voronoi noise and cellular noise, is a noise function introduced by Steven Worley in 1996. Worley noise is an extension of the Voronoi diagram that outputs a real value at a given coordinate that corresponds to the distance of the nth nearest seed (usually n=1) and the seeds are distributed evenly through the region. Worley noise is used to create procedural textures.

Worley noise of Euclidean distance is differentiable and continuous everywhere except on the edges of the Voronoi diagram of the set of seeds and on the location of the seeds.