Jump flooding algorithm (JFA)
A GPU algorithm that computes approximate Voronoi diagrams and distance fields in a logarithmic number of passes.
Described by Guodong Rong and Tiow-Seng Tan in 2006. On each pass every pixel samples its neighbours at a step size that halves each time — big jumps first, then finer — propagating information about the nearest seed across the surface. After log₂(n) passes every pixel knows its closest seed.
The result is approximate — small errors can appear at pass boundaries — but very fast, which makes JFA popular for distance fields, outlines, soft shadows and GPU-driven text rendering.