High-Fidelity Physics Research

Welcome to the homepage for NVIDIA’s High-Fidelity Physics Research group lead by Ken Museth. We conduct research in the intersection of physics simulations, computer graphics, and machine learning. As the name suggests, we tend to prioritize high accuracy and fidelity over real-time performance and bridge the gap with sparse and neural representations as well as other machine learning techniques. We are researching both traditional (forward) and inverse physics problems, including solvers for fluids (gas and water), hair, cloth, muscles, soft-bodies, oceans, and granular materials like sand. Broadly speaking we are keenly interested in exploring novel ways of introducing more accurate physics in neural understandings and encodings of the world surrounding us. Our group consists of both researchers and engineers located across Asia, Europe and North America. If you are interested in joining our team feel free to reach out to any of us.


April 2024 - We are excited to join forces with Sanja Fidler and her AI research team.

March 2024 - Technical Achievement Award, to Lucas Miller, Christopher Jon Horvath, Steve LaVietes and Joe Ardent for the creation of the Alembic Caching and Interchange system.

March 2024 - Scientific & Engineering Award to Ken Museth, Peter Cucka and Mihai Aldén for the creation of OpenVDB and its ongoing impact within the motion picture industry.

August 2023 - ACM SIGGRAPH Practitioner Award to Ken Museth for the creation and continued development of OpenVDB.


Quickly discover relevant content by filtering publications.