Adaptive and lightweight bidirectional rendering methods

• Applicant: Dr.-Ing. Pascal Grittmann
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 540768663

Monte Carlo rendering is used in a wide range of applications. From entertainment uses to product design and advertisement to engineering visualizations. The demand for ever faster rendering of ever more complex scenes is continually growing. To satiate that need, this project aims to explore and enhance the practical utility of bidirectional rendering methods. These can efficiently render some very difficult scenes with complex indirect illumination such as caustics. However, in practice, bidirectional methods are rarely used. Instead, the go-to default method is unidirectional path tracing. That is so because bidirectional methods can be inefficient and wasteful on many practical scenes, as they do not naturally focus computation on visible effects. We aim to combat that problem by designing methods to automatically adapt bidirectional rendering algorithms to the scene. To that end, we plan to develop learning-based methods that focus computation on those regions of the scene that need it the most.

• DFG Programme: Research Grants