🔲 Worley Noise — Cellular Texture

Worley noise, or cellular noise, is a procedural texture technique that scatters a set of feature points through space and assigns each pixel a value based on its distance to the closest one or two points. It produces natural-looking cellular patterns such as cracked stone, water caustics, and organic cell structures rather than the smooth blobs typical of Perlin noise.

F1 is the distance from a pixel to its single nearest feature point, and F2 is the distance to the second-nearest one. Shading by F1 alone produces smooth, rounded cells; F2−F1 highlights the thin borders between neighbouring cells as crack-like ridges; and F1·F2 blends the two into a softer, more organic look.

Each feature point owns exactly the region of pixels that are closer to it than to any other point — which is the textbook definition of a Voronoi cell. Shading by F1 draws the interior of each Voronoi cell, while the F2−F1 mode draws precisely the edges of that same Voronoi diagram, so Worley noise is essentially a continuously shaded Voronoi tessellation.

Density sets the average number of feature points seeded into each grid cell before the nearest-distance search runs. A higher density packs in more points, giving smaller and more numerous cells, while a lower density spreads points further apart, producing large open cells separated by long cracks.

Euclidean distance is straight-line distance and produces the classic round, organic cells. Manhattan distance sums the horizontal and vertical offsets, which turns the cell boundaries into diamond shapes. Chebyshev distance takes the larger of the horizontal or vertical offset, which squares off the cells into blocky, grid-aligned shapes.