Efficient sampling for computing complex illumination in real-time

This thesis delves into advanced techniques for real-time computation of complex direct and indirect illumination, focusing on Reservoir-based SpatioTemporal Importance Resampling (ReSTIR). It offers a practical implementation of the ReSTIR algorithm for direct illumination, demonstrating a robust approach for biased and unbiased many-light sampling that leverages hardware-accelerated ray tracing. The study includes qualitative and performance tests validating the algorithm's effectiveness in various settings. Furthermore, we propose several optimizations to enhance computational efficiency and rendering quality. The thesis concludes with a discussion of potential future work, outlining promising directions for further improvements and expansions of the ReSTIR methodology.