Ice Melting Simulation using Metaballs and Marching Cubes on GPUs

碩士 === 國立交通大學 === 多媒體工程研究所 === 101 === In this thesis, we propose a novel ice melting simulation method based on a particle-based model. We have two types of particles: ice particles and water particles. The ice particles represent the ice model and the water particles represent water. In our metho...

Full description

Bibliographic Details
Main Authors: Lii, Shing-Yeu, 李幸宇
Other Authors: Wong, Sai-Keung
Format: Others
Language:en_US
Published: 2012
Online Access:http://ndltd.ncl.edu.tw/handle/52215001872957445194
Description
Summary:碩士 === 國立交通大學 === 多媒體工程研究所 === 101 === In this thesis, we propose a novel ice melting simulation method based on a particle-based model. We have two types of particles: ice particles and water particles. The ice particles represent the ice model and the water particles represent water. In our method, ice particles have an attribute named virtual water volume, which is used to represent the fluid volume around this ice particle. An ice particle may increase its amount of virtual water volume when the latent heat of the ice particle increases. We simulate fluid flowing on the surface of the ice model by transferring the virtual water volume between ice particles. A water particle is produced if the virtual water volume in an ice particle is larger than default volume of the water particle. For smoothly shrinking ice model, we propose a new method to calculate the potential field. The latent heat of ice particles and virtual water volume are taken into account for computing the potential field. We use marching cubes to construct the polygonal mesh for the ice model. To render the ice model and water particles, we propose a ray tracing method to render them. The ice model is represented by the polygonal mesh, and water particles are constructed into metaballs. We use the bounding volume hierarchy (BVH) to create hierarchy constructions to accelerate the process of ray tracing. Our method has been implemented on GPUs. Experiment results show that our method is efficient and compute realistic ice melting simulation.