FracGen: Learning How Objects Stretch and Tear with Physics-Informed Video Generation

2026-09-29

AI Summary

This study addresses the inherent trade-off between high fidelity and computational efficiency in generating physically plausible, controllable fracture videos from a single image. To this end, we propose FracSim, a framework that constructs a Material Point Method (MPM) simulation engine integrated with a continuous damage model to synthesize paired training data at low cost. Furthermore, we design a diffusion model that jointly predicts RGB frames and physical fields, incorporating a physics-aware loss to constrain the network toward learning intrinsic physical states rather than merely fitting surface appearances. Our approach significantly outperforms existing baselines in both physical realism and visual fidelity, enabling fine-grained control over tear locations, crack propagation velocities, and deformations. This work establishes an efficient new paradigm for physics-driven fracture video generation.

Abstract

We introduce FracGen, a fracture-aware video generation model that produces plausible, controllable fracture dynamics from a single image of an intact object, conditioned on physics signals. To train FracGen, we build FracSim, a fracture-aware simulation framework that augments material point method (MPM) simulation with a continuum damage model, producing paired fracture videos and dense, pixel-aligned physical fields at no additional cost beyond standard rendering. FracGen leverages these maps in two ways: it is trained to jointly predict them alongside RGB video, encouraging the model to capture physical state rather than surface appearance; and it is supervised with physics-informed losses that encourage consistency among the predicted maps. As a result, FracGen captures distinct material-specific fracture behavior without expensive test-time simulation or per-scene tuning, while offering fine-grained control over where an object tears, how fast the crack propagates, and how much deformation precedes failure. We further introduce a benchmark for evaluating the physical plausibility of generated fracture video, and show through extensive experiments that FracGen outperforms existing video generation baselines in both physical and visual fidelity. Results are best viewed in our project website: https://fracgen.github.io/.

Problem

• fracture dynamics
• video generation
• physics-informed
• material-specific fracture
• controllable simulation

Innovation

• Physics-informed video generation
• Material point method (MPM)
• Fracture dynamics
• Continuum damage model
• Physical consistency loss