Poisson disk sampling
Generates points that are randomly spread yet never closer together than a given radius — an even "blue noise".
Purely random points clump; a grid comes out unnaturally regular. Poisson disk sampling lands in between: each new point is rejected if it falls within radius r of an existing one, so the distribution is dense but free of clusters. Robert Bridson's 2007 algorithm does it in linear time using a background grid and a list of active points to seed new ones around.
It is used to scatter trees, tufts of grass and rocks across procedural landscapes, for stippling, and to place sample points for soft shadows, ambient occlusion and depth of field — anywhere you want randomness without clumpiness.