Generating photorealistic images requires an accurate and unbiased simulation of light transfer in a scene that is typically modeled using geometric primitives (e.g. a huge collection of triangles), and using a collection of reflectance properties that represent the way surfaces and volumes interact with light.
A classical approach to estimating the color of a given pixel consists in accumulating contributions from randomly generated light paths connecting the light sources to the camera sensor. Once a few valid paths have been found, random perturbations (or mutations) may be used, so as to generate new samples at fairly low cost.
Studying the structure of light transport operators would bring a global understanding of the structure of light paths space in a scene. A consequence is to make efficient sampling of this space more easy to achieve, and to find representations of light transport that are both economic and consistent with the current geometry and meterial.
The main objective of the CALiTrOp project is to go over a general functional and geometrical analysis in order to identify and model the properties of the light transport operators useful for efficient rendering.