The hosts discuss NVIDIA and University of Toronto's research on novel rendering techniques utilizing 3D Gaussian splatting. They introduce 3DGRT, a ray-tracing method for particle-based scenes, enabling advanced effects and handling distorted cameras more effectively than rasterization. To address 3DGRT's speed limitations, the creators of 3DGUT, which uses an unscented transform within a rasterization framework to also support distorted cameras and secondary rays. The 3dgrut GitHub repository offers the official implementations of both 3DGRT and 3DGUT, along with a hybrid approach called 3DGRUT, combining their strengths. The NVIDIA Toronto AI Lab page lists researchers and projects, highlighting their work in machine learning for computer graphics and mentioning the publications introducing 3DGRT and 3DGUT.