Reproduce “MatPhys: Learning Material-Aware Physics Parameters for Deformable Object Simulation from Videos” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yang Yang, Yiyan Wang, Zheming Liu, Naoya Iwamoto
MatPhys predicts physical parameters for deformable-object simulations from video using semantic parts and material information. It groups reconstructed 3D Gaussians into parts, combines part-level material priors with video motion and spring geometry, and predicts stiffness, damping, and contact parameters in one forward pass. On PhysTwin future-frame prediction, it reports CD 0.008 and silhouette IoU 79.8%, ahead of NeuSpring across 22 evaluated scenarios.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.