A longstanding goal in computer vision is to model motions from videos, while the representations behind motions, i.e. the invisible physical interactions that cause objects to deform and move, remain largely unexplored. In this paper, we study how to recover the invisible forces from visual observations, e.g., estimating the wind field by observing a leaf falling to the ground. Our key innovation is an end-to-end differentiable inverse graphics framework, which jointly models object geometry, physical properties, and interactions directly from videos. Through backpropagation, our approach enables the recovery of force representations from object motions. We validate our method on both synthetic and real-world scenarios, and the results demonstrate its ability to infer plausible force fields from videos. Furthermore, we show the potential applications of our approach, including physics-based video generation and editing. We hope our approach sheds light on understanding and modeling the physical process behind pixels, bridging the gap between vision and physics.
Abstract
We propose a differentiable inverse graphics framework to recover invisible forces from videos by integrating object modeling, physics simulation, and optimization:
- Objects are represented with 3D Gaussians and assigned physical properties via Vision-Language Models.
- Forces are modeled as a causal tri-plane, capturing underlying dynamics.
- Object motions are simulated using a differentiable physics simulator.
- A sparse tracking objective and pulsed-force optimization enable differentiable force recovery from videos.