A real time high performance approach to volumetric light transport

Our research consists of developing novel techniques to further improve the rendering time and quality of volumetric light transport, in special the effect of multiple scattering on path tracing. Our proposal is to explore the human perceptual system to speed up the computation using a visual perception metric. To determine which parameters are feasible of change in a way not perceptible by the human eye, the CIE metric states that two images will look similar if the Euclidian distance in the CIELAB color space is smaller than 2.3. With this threshold, we will be able to determine which parameters of our technique play a heavier role and thus balance them out to get better performance while keeping quality. These new rendering techniques may be incorporated into motion pictures, animations and games giving better visuals to participating media such as clouds, smoke and dust.

Faculty Supervisor:

Pierre Poulin

Student:

Partner:

Universidade Federal do Rio Grande do Sul

Discipline:

Computer science

Sector:

Technology; New and Digital Media; Other

University:

Université de Montréal

Program: