Agenda
Basics to Beyond: Sparse Volume GI in 'Arena Breakout: Infinite'
Ruochen Ying (Senior Graphics Programmer, Tencent Games)
Hantao Zheng (Graphics Programmer, Tencent Games)
Yuzhi Liang (Graphics Programmer, Tencent Games)
Pass Type:
Academic Pass, Festival Pass, Game Changer Pass, Indie Pass, Start-Up Pass -
Get your pass now!
Track:
Programming
Audience Level:
Intermediate
Format:
Lecture
Vault Recording: TBD
Audience Level: Intermediate
This session shares the design of the GI system shipping in Arena Breakout: Infinite, an online first-person extraction shooter. It addresses challenges including large maps (2 km × 2 km), dynamic time of day and weather, and scalability (from 60 to 120 fps) across a wide range of hardware(from GTX 960 to RTX 5090). The solution features a sparse, geometry-adaptive tree structure for organizing volumetric lighting samples, that scales to hundreds of millions of high-density lighting samples per map, together with an efficient runtime sampling strategy that reduces light leakage and provides seamless reconstruction over long viewing distances. It also implements custom compression and streaming strategies to meet VRAM constraints and support incremental relighting. Throughout, the session shows how project constraints can guide a practical, tailored solution. It also discusses experiments with learned lighting representations, including MLP-based far-field GI and neural compression.
Takeaway
Attendees learn how the developers of Arena Breakout: Infinite design the game's GI system to meet ambitious visual goals within the tight performance and memory budgets of a competitive online FPS. They gain practical techniques for sparse volume design, seamless irradiance reconstruction, streaming, compression, and learned lighting representations.
Intended Audience
This is intended for graphics programmer, technical artists, etc., who are required to have a certain foundation in rendering.