A new State of the Art report co-authored (among others) by past VCL member Gurprit Singh and VCL Director Wojciech Jarosz reviews a long thread of research on point sampling distributions for computer graphics.
The lighting of a virtual scene is often approximated via a long, stochastic process called Monte Carlo rendering, which takes many samples of the lighting throughout the scene and averages them. How the sample points are chosen can have a palpable effect on how long it takes for the image to converge, or have no “graininess” left due to sampling noise. This STAR report investigates the various methods that have been devised for reducing noise faster.
The report drew from experts all over the world, including collaborators Cengiz Öztireli (Disney Research, Zurich), Abdalla G.M. Ahmed (KAUST, Saudi Arabia), David Coeurjolly (Université de Lyon, CNRS, France), Kartic Subr (University of Edinburgh, UK), Oliver Deussen (University of Konstanz, Germany), Victor Ostromoukhov (Université de Lyon, CNRS, France), and Ravi Ramamoorthi (University of California, San Diego).
For more details, head over to the project page.