SurfPhase: 3D Interfacial Dynamics in Two-Phase Flows from Sparse Videos

TL;DR

SurfPhase is a novel method that reconstructs 3D interfacial dynamics in two-phase flows from sparse videos, enabling accurate visualization and velocity estimation of sharp, deformable liquid-vapor interfaces.

Contribution

It introduces a combined approach using dynamic Gaussian surfels, signed distance functions, and a video diffusion model to improve 3D reconstruction from limited camera views.

Findings

High-quality view synthesis from two cameras

Accurate velocity estimation of interfaces

Effective reconstruction of sharp, deformable interfaces

Abstract

Interfacial dynamics in two-phase flows govern momentum, heat, and mass transfer, yet remain difficult to measure experimentally. Classical techniques face intrinsic limitations near moving interfaces, while existing neural rendering methods target single-phase flows with diffuse boundaries and cannot handle sharp, deformable liquid-vapor interfaces. We propose SurfPhase, a novel model for reconstructing 3D interfacial dynamics from sparse camera views. Our approach integrates dynamic Gaussian surfels with a signed distance function formulation for geometric consistency, and leverages a video diffusion model to synthesize novel-view videos to refine reconstruction from sparse observations. We evaluate on a new dataset of high-speed pool boiling videos, demonstrating high-quality view synthesis and velocity estimation from only two camera views. Project website: https://yuegao.me/SurfPhase.