Augmented reality: Full 3D scene relighting, object insertion and material editing with a single image or video as input.
Inverse rendering: Estimation of shape, material and lighting in indoor scenes with complex light transport.
3D reconstruction: Implicit-explicit representations with complex topology (neural implicit evolution, neural mesh flow, topological derivatives).
Photorealistic digital twins: Automated 3D content creation from videos with high-quality appearance, shape, material and lighting.
Complex light transport: Physically-motivated estimation of shape, material and lighting with complex global interactions.
Applications: Inverse rendering and 3D content generation applied to use-cases across indoor scenes, robotics, faces and view synthesis.