- Published 2006

In the particle model light is assumed to be composed of photons originating at the light sources and moving within the environment possibly getting scattered by the medium and refracted or re ected by the objects in the environment until they are eventually absorbed Using Monte Carlo methods for simulating the behaviour and keep ing track of a su ciently large sample of photons in the environment illumination e ects can be estimated In this paper we rst describe the algorithm used for Monte Carlo simulation and show how the results are used for computing global illumination e ects and for synthesising realistic images A distinctive feature of the algorithm is that as the simulation progresses with in creasing number of samples the computed illumination is progressively re ned A detailed comparative performance analysis with a straight forward implementation of the radiosity method shows that the results are equally good and the computation times comparable The major advantages over the other well established methods of ray tracing and radiosity stem from the fact that complex environmental behaviour can be more easily accommodated In particular we discuss techniques for dealing with complex extended light sources and large unevenly lighted surfaces without any need for apriori discretisation and the issues related to the rendering of non di use environments Finally we describe a variation in the simulation algorithm for variance reduction and compare the results with the earlier algorithm

@inproceedings{Pattanaik2006ComputationOG,
title={Computation of Global Illumination by Monte Carlo Simulation of the Particle Model of Light},
author={Sumanta N. Pattanaik},
year={2006}
}