GaussianFluent: Gaussian Simulation for Dynamic Scenes with Mixed Materials

Bei Huang, Yixin Chen, Ruijie Lu, Gang Zeng, Hongbin Zha, Yuru Pei, Siyuan Huang

Keywords: 3dgs, simulation, animation, reconstruction

TL;DR: GS diverse simulation with interior generation

Abstract: 3D Gaussian Splatting (3DGS) has emerged as a prominent 3D representation for high-fidelity and real-time rendering. Prior work has coupled physics simulation with Gaussians, but it predominantly targets soft, highly deformable materials such as rubber and snow, leaving brittle fracture in objects like watermelons unresolved. This stems from two obstacles: the lack of a volumetric interior with coherent textures with GS representation, and the absence of fracture-aware simulation methods for Gaussians. To overcome these, we introduce GaussianFluent, a framework for realistic simulation and rendering of dynamic object states. First, it synthesizes consistent, photorealistic interiors by densifying internal Gaussians guided by generative models. Second, it integrates an optimized Continuum Damage Material Point Method (CD-MPM) to enable correct brittle fracture at real-time speeds. Finally, a Blinn-Phong model is used to shade the dynamically evolving fracture surfaces. Experiments show GaussianFluent delivers photo-realistic, real-time renderings of these state changes with structurally consistent interiors, highlighting its potential for downstream applications.

Primary Area: applications to computer vision, audio, language, and other modalities

Submission Number: 5755