Video-Driven Graph Network-Based Simulators
This addresses the need for efficient physics simulations in design, cinematography, and gaming, but it is incremental as it builds on existing Graph Network-based Simulators.
The paper tackles the problem of physics simulation in visual applications by inferring physical properties from short videos, eliminating explicit parameter input, and demonstrates that video-derived encodings capture these properties with a linear dependence on motion.
Lifelike visualizations in design, cinematography, and gaming rely on precise physics simulations, typically requiring extensive computational resources and detailed physical input. This paper presents a method that can infer a system's physical properties from a short video, eliminating the need for explicit parameter input, provided it is close to the training condition. The learned representation is then used within a Graph Network-based Simulator to emulate the trajectories of physical systems. We demonstrate that the video-derived encodings effectively capture the physical properties of the system and showcase a linear dependence between some of the encodings and the system's motion.