Peter Kocsis successfully defends PhD thesis on material estimation and lighting control

Matt Niessner congratulated Peter Kocsis on his successful PhD defense, describing a thesis on generative priors for material estimation and lighting control. Niessner says Kocsis’s Intrinsic Image Diffusion samples plausible ways to separate an image into its underlying components rather than predicting a single answer. He also highlights Kocsis’s work on lighting control and 3D material reconstruction, and says Kocsis is continuing this research at NVIDIA.