IAP SIE | AI with Social Influence for Enjoyment in Gaming

This lecture provides a high-level survey of how real-time rendering in video games is evolving as machine learning begins to challenge the traditional rasterization-based pipeline. It begins with a brief overview of classic graphics methods that have defined interactive rendering for decades, establishing the assumptions and constraints that shaped modern GPUs and game engines. The lecture then surveys emerging applications of machine learning in rendering, highlighting where learned methods already augment or replace parts of the pipeline. The second half of the lecture focuses on a deeper dive into neural rendering, examining how learned representations, inference, and temporal prediction enable new rendering paradigms that infer images rather than explicitly compute every pixel. The session concludes by discussing open challenges, production realities, and the implications of these techniques for the future of real-time interactive graphics. Connect with CSAIL Alliances On our site: https://cap.csail.mit.edu/ On LinkedIn:   / mit-csail