SwitchLight: Co-design of Physics-driven Architecture and Pre-training Framework for Human Portrait Relighting
This work addresses the challenge of realistic human portrait relighting for computer vision applications, though it appears incremental as it builds on existing physics models and data limitations.
The paper tackled the problem of human portrait relighting by co-designing a physics-guided architecture and a self-supervised pre-training framework, achieving a new benchmark in relighting realism.
We introduce a co-designed approach for human portrait relighting that combines a physics-guided architecture with a pre-training framework. Drawing on the Cook-Torrance reflectance model, we have meticulously configured the architecture design to precisely simulate light-surface interactions. Furthermore, to overcome the limitation of scarce high-quality lightstage data, we have developed a self-supervised pre-training strategy. This novel combination of accurate physical modeling and expanded training dataset establishes a new benchmark in relighting realism.