MPMWorlds: Material-Point-Method Simulations for Inferring and Extrapolating Physical Dynamics
2026-06-01 • Graphics
GraphicsComputer Vision and Pattern RecognitionMachine Learning
AI summaryⓘ
The authors created a collection of 2D physics simulations showing different materials and movements to test how well computers can guess the physical behavior from videos and predict what happens next. They tried two methods: one that writes computer code to recreate the physics and another that generates videos based on visual info. They found the code-writing method is better at making realistic and stable future predictions but has trouble understanding physical details from images. The video-generating method picks up shapes better but makes less realistic predictions about what happens next.
Material Point Methodphysical simulationvideo diffusioncode generationdeformable objectsfluidskinetic objectsphysical parametersvideo extrapolationvisual input
Authors
Žiga Kovačič, Kevin Ellis
Abstract
To study the ability to infer physical dynamics from videos and extrapolate them forward in time, we assemble a dataset of 2D Material Point Method (MPM) physical simulations covering rich physical phenomena such as deformable objects, fluids, kinetic objects, and emitters. We study code generation and video diffusion approaches on this dataset, identifying their strengths and weaknesses by varying the amount of physically relevant side information. The code generation model, beyond giving a working demonstration of automatic synthesis of MPM simulations, reveals that such an approach struggles with inferring physical parameters from visual input, but relative to video diffusion, produces physically and temporally stable extrapolations forward in time, while the video diffusion model more strongly identifies geometric properties from visual input but produces physically implausible extrapolations.