An animation of Mitchell’s best-candidate algorithm, which produces samples with blue-noise spectral characteristics that are useful for minimizing aliasing. Unlike uniform random sampling, best-candidate samples are more evenly distributed, with fewer samples close together. (A similar, but more efficient, algorithm is poisson-disc sampling.)
For each new sample, the best-candidate algorithm generates a fixed number of candidate samples, shown in gray. Here, 10 candidates are generated. The best candidate, shown in red, is the one that is farthest away from all previous (non-candidate) samples.