Latent Intuitive Physics: Learning to Transfer Hidden Physics from A 3D Video

TL;DR

This paper presents a transfer learning framework that infers hidden fluid properties from a single 3D video and enables novel scene simulations without explicit physical parameters.

Contribution

It introduces latent intuitive physics using a learnable prior conditioned on particle states, allowing for unseen scene simulation and future fluid dynamics prediction.

Findings

Successful simulation of new scenes with learned physics

Accurate future prediction of fluid behavior

Effective supervised particle simulation results

Abstract

We introduce latent intuitive physics, a transfer learning framework for physics simulation that can infer hidden properties of fluids from a single 3D video and simulate the observed fluid in novel scenes. Our key insight is to use latent features drawn from a learnable prior distribution conditioned on the underlying particle states to capture the invisible and complex physical properties. To achieve this, we train a parametrized prior learner given visual observations to approximate the visual posterior of inverse graphics, and both the particle states and the visual posterior are obtained from a learned neural renderer. The converged prior learner is embedded in our probabilistic physics engine, allowing us to perform novel simulations on unseen geometries, boundaries, and dynamics without knowledge of the true physical parameters. We validate our model in three ways: (i) novel scene simulation with the learned visual-world physics, (ii) future prediction of the observed fluid dynamics, and (iii) supervised particle simulation. Our model demonstrates strong performance in all three tasks.