DeformMaster: An Interactive Physics-Neural World Model for Deformable Objects from Videos

TL;DR

DeformMaster is a novel interactive physics-neural world model that learns to simulate and render deformable objects from real videos, capturing physical dynamics, material behavior, and appearance for realistic future predictions.

Contribution

It introduces a unified framework combining physics-based modeling with neural residuals to accurately simulate deformable objects from real-world videos.

Findings

Outperforms state-of-the-art baselines in future dynamics prediction

Supports action rollout, material variation, and novel-view synthesis

Demonstrates high-fidelity rendering of deformable objects from real videos

Abstract

World models for deformable objects should recover not only geometry and appearance, but also underlying physical dynamics, interaction grounding, and material behavior. Learning such a model from real videos is challenging because deformable linear, planar, and volumetric objects evolve under high-dimensional deformation, noisy interactions, and complex material response. The model must therefore infer a physical state from visual observations, roll it forward under new interactions, and render the resulting dynamics with high visual fidelity. We present DeformMaster, a video-derived interactive physics-neural world model that turns real interaction videos into an online interactive model of deformable objects within a unified dynamics-and-appearance framework. DeformMaster preserves structured physical rollout while using a neural residual to compensate for unmodeled effects, grounds sparse hand motion as distributed compliant actuator for hand-continuum interaction, represents material response with spatially varying constitutive experts, and drives high-fidelity 4D appearance from the predicted physical evolution. Experiments on real-world deformable-object sequences demonstrate DeformMaster's ability to roll out future dynamics and render dynamic appearance, outperforming state-of-the-art baselines while supporting novel action rollout, material-parameter variation, and dynamic novel-view synthesis. Project page: https://can-lee.github.io/deformmaster-web/