PhysConvex: Physics-Informed 3D Dynamic Convex Radiance Fields for Reconstruction and Simulation

2026-02-21

AI Summary

Existing methods struggle to simultaneously achieve visual realism and physical consistency in dynamic 3D scenes, particularly when modeling complex material deformations and dynamics. This work proposes PhysConvex, a novel approach that introduces a boundary-driven dynamic convex primitive representation to construct a deformable radiance field, unifying visual rendering and physics-based simulation. By incorporating reduced-order dynamics via neural skinning eigenmodes, physics-informed constraints on convex primitives, and Newtonian mechanics–guided time-varying degrees of freedom control, PhysConvex enables shape- and material-aware non-uniform deformation modeling and accurately captures evolving boundaries. Experiments demonstrate that the method achieves high-fidelity reconstruction of geometry, appearance, and physical properties from monocular video alone, significantly outperforming state-of-the-art approaches in both reconstruction quality and simulation accuracy.

Technology Category

• Computer Vision: Low Level & Physics-based Vision
• Search and Optimization: Non-convex Optimization
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

Abstract

Reconstructing and simulating dynamic 3D scenes with both visual realism and physical consistency remains a fundamental challenge. Existing neural representations, such as NeRFs and 3DGS, excel in appearance reconstruction but struggle to capture complex material deformation and dynamics. We propose PhysConvex, a Physics-informed 3D Dynamic Convex Radiance Field that unifies visual rendering and physical simulation. PhysConvex represents deformable radiance fields using physically grounded convex primitives governed by continuum mechanics. We introduce a boundary-driven dynamic convex representation that models deformation through vertex and surface dynamics, capturing spatially adaptive, non-uniform deformation, and evolving boundaries. To efficiently simulate complex geometries and heterogeneous materials, we further develop a reduced-order convex simulation that advects dynamic convex fields using neural skinning eigenmodes as shape- and material-aware deformation bases with time-varying reduced DOFs under Newtonian dynamics. Convex dynamics also offers compact, gap-free volumetric coverage, enhancing both geometric efficiency and simulation fidelity. Experiments demonstrate that PhysConvex achieves high-fidelity reconstruction of geometry, appearance, and physical properties from videos, outperforming existing methods.

Problem

• dynamic 3D reconstruction
• physical consistency
• material deformation
• visual realism
• physics-informed simulation

Innovation

• Physics-informed
• Dynamic Convex Radiance Fields
• Continuum Mechanics
• Reduced-order Simulation
• Neural Skinning Eigenmodes