Abstract: In recent years, significant progress has been made in using neural networks to represent and learn complex objects or functions, which has strongly promoted the leapfrog development of graphics computing and spatial computing. This talk will introduce our latest research progress in neural graphics rendering technology and architecture. The architecture decouples the global light energy radiation transmission process of the scene, using a fully connected hypernetwork to learn the environmental light energy transmission variation model with objects as insertion units, and establishes a light energy radiation transmission incremental model with light sources as insertion units using Transformer architecture. It creates a real-time neural rendering pipeline for global full-spectrum illumination and develops a universal lighting large model, effectively breaking through the GPU architecture limitations represented by NVIDIA's RTcore, enabling real-time high-quality rendering of global full-spectrum lighting effects on chips with ordinary rasterization and AI inference capabilities. The talk will conclude with a vision for the future development of this new technology framework.

Speaker Bio: Hujun Bao, Distinguished Professor of Changjiang Scholars and Special Expert of Zhejiang Province, mainly engages in research in computer graphics and mixed reality. He has received funding from the National Outstanding Youth Fund and the National Innovation Research Group Science Fund, and as the chief scientist, he has undertaken research on virtual reality and mixed reality in two 973 Program projects. Significant innovative breakthroughs have been made in aspects such as three-dimensional modeling of mixed reality scenes, mobile three-dimensional registration and fusion, adaptive high-fidelity rendering, and natural digital avatar interaction. He has led the development of a cloud-edge collaborative mixed reality software platform, successfully achieving industrial application. His research achievements have won one National Natural Science Award and four provincial and ministerial level science and technology awards.