ABSTRACT
This chapter aims to understand why sampling is important in ray tracing; and shows how several common sampling techniques are implemented. It discusses the characteristics of good sampling techniques; and presents a have at readers' disposal a flexible sampling architecture. Compared with the best sampling techniques, the worst techniques can require three times as many samples per pixel to reduce noise to an acceptable level. Multi-jittered sampling was developed by K. Chiu et al. to improve the 2D distribution of the samples in n-rooks sampling, while preserving the even 1D projection. J. M. Hammersley sampling was developed in the 1960s, and as such, it’s a lot older than jittered, n-rooks, and multi-jittered sampling. Each object will have a pointer to its own sampler object, and let’s suppose that the sampling is multi-jittered. Each pixel whose rays hit an object that is shaded with the light will use samples from both sampling objects.
