This paper provides a very general summary of how stochastic sampling of an image reduces aliasing but adds noise to the resulting image. The reader is forced to accept the author’s viewpoint because he provides no reasoning or rationale, with hard information, on why this form of sampling was developed or how. This makes it difficult for the reader to verify any of the author’s statements and conclusions, follow up on the topic, or apply the sampling technique to actual images. A large bibliography is provided, which leaves it up to the reader to do the same research in order to arrive at the same conclusions.
A quick review of discrete sampling of an image in computer graphics is presented, along with how uniform point sampling results in aliasing. Then, using the human eye as a basis, the author briefly discusses a Poisson disk distribution as a good random, nonuniform sampling distribution. The technique of jittering (adding noise to a sample location) is discussed in general terms, including how it can approximate a Poisson disk distribution. Jittering that is applied to distributed ray tracing is presented in general terms. The paper concludes with “examples” that are simply the results of stochastic sampling. The author provides little information about how this discrete sampling method is actually used to generate the pictures.