Multiple importance sampling for path guiding

• Applicant: Dr.-Ing. Pascal Grittmann
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 563229620

Numerous applications benefit from light transport simulation in virtual scenes. Those include product and architecture visualization, video games, or VFX in movies. The demand for faster and more robust methods to render images of such virtual scenes is far from satisfied. A promising research direction to that end is the use of adaptive methods, so called "path guiding" solutions. These automatically adapt the rendering algorithm to the scene that is currently being rendered. A vital ingredient to make path guiding - and other rendering algorithms - sufficiently robust and generic is multiple importance sampling (MIS). MIS allows us to combine multiple techniques into one joint solution. While path guiding is reasonably well explored in isolation, the optimal use of MIS in the context of path guiding is still underexplored. The goal of this project hence is to explore this key aspect further and answer the question: How can path guiding methods reap maximum benefits from MIS? This poses a number of research challenges, including the optimal allocation of the available sampling budget across techniques, and the adaptation of techniques to avoid harmful redundancies.

• DFG Programme: Research Grants