Matt Stanton, a graduate student from Carnegie Melon, will be giving a talk about his research.
Game graphics, physics, and AI are still largely limited by the processing power of individual PCs and consoles. In this talk, I will discuss how we can use massive cloud-based precomputation to create games that are not bound by these constraints. Using player models learned from recorded gameplay traces, we can efficiently sample game state spaces in an offline process, then use the sampled data to build an interactive game runtime. Our game dynamics can be arbitrarily complex, allowing us to create immersive, engaging games that could not be made any other way.