Monte carlo and quasi-monte carlo methods for computer graphics

Abstract

Some computer graphics applications, such as architectural design, generate visually realistic images of computer models. This is accomplished by either explicitly or implicitly solving the light transport equations. Accurate solutions involve high-dimensional equations, and Monte Carlo (MC) techniques are used with an emphasis on importance sampling rather than stratification. For many applications, approximate solutions are adequate, and the dimensionality of the problem can be reduced. In these cases, the distribution of samples is important, and quasi-Monte Carlo (QMC) methods are often used. It is still unknown what sampling schemes are best for these lower dimensional graphics problems, or what "best" even means in this case. This paper reviews the work in MC and QMC computer graphics, and poses some open problems in the field. © 2008 Springer-Verlag Berlin Heidelberg.