Abstract
Snow simulation is always a challenge in the computer graphics community due to its combined nature of solids and fluids. In the past, researchers usually applied different solvers to computationally simulate the behavior of snow at different phases, which made the simulation both slow and complicated. In 2013, the material point method, abbreviated as MPM, was first introduced for snow simulation, eliminating the need for multiple solvers. This paper investigates the history of the application of MPM to snow simulation in computer graphics specifically, and offers an overview of its evolution since the pioneering work by Stomakhin. It aims at showing the current state-of-art as well as any limitations. Nowadays, the development of MPM and snow simulation focuses on improvements of the stability and physical accuracy of the method itself, and the generalization of the application scope from snow to arbitrary granular materials. The trade-off between efficiency and accuracy remains a problem, thus it introduces more potential research directions, ranging from developing simpler mathematical models for better physical accuracy to incorporating machine learning techniques to accelerate the simulation process.
Mingrui Wang
CS PhD Student
My research interests include computer graphics, physics-based simulation and deep Learning.