Multi-Class Poisson Disk Sampling

Sampling has been a core process for a variety of graphics applications including rendering, imaging, and geometry processing. Among the plethora of sampling patterns, Poisson disk distribution remains one of the most popular thanks to its spatial uniformity and blue noise spectrum. However, research so far has been mainly focused on Poisson disk distribution with one class of samples. This could be insufficient for common natural as well as man-made phenomenon requiring multiple classes of samples, such as object placement, imaging sensors, and stippling patterns. We extend Poisson disk sampling to multiple classes of samples where each individual class as well as their union exhibit Poisson disk properties. We propose algorithms to generate such multi-class Poisson disk samples, study their statistical characteristics, and demonstrate applications in object placement, sensor layout, and color stippling.