Grid and particle based hybrid fluid simulation on the GPU

OData support
Supervisor:
Dr. Umenhoffer Tamás
Department of Control Engineering and Information Technology

The topic of my thesis is grid and particle based hybrid fluid simulation on the GPU. Fluid simulation is one of the most famous topics of computer graphics. Realistic simulation of fluid like materials can be achieved with fluid simulation. One of these materials is water, which is the most obvious one. Although it is not only used to simulate water, it is also used for smoke, cloud and fire simulation as well, because these materials behave like fluids. Realistic modelling of fluids and fluid like materials is very complex, with a fairly easy mathematical model, but computation of the model takes too much resource and time, this is why approximation is necessary. In order to achieve the best results the original momentum equations should be used, and solved as easily as possible.

During my work I tried to implement a method, which combines different fluid simulation methods and as a result creates the highest quality simulation with the most efficient way. In order to do that, I had to learn about the grid based and particle based approach of fluid simulation, about the positives and the negatives of each types. After that I could find the advantages of each method, thus I was able to create a more efficient hybrid solution.

In my thesis I will present the equations describing fluid movements, grid based and particle based simulation approaches and the theory of the hybrid simulation. All of the topics from before will have two and three dimensional solutions as well. For the implementation I have used OpenGL shaders, which gave me the opportunity to generally program the graphics card. This way I have shaders not only for displaying purposes; I use them for computations too, which gives nearly or real-time performance to this problem, which has extreme parallelization possibilities.

Downloads

Please sign in to download the files of this thesis.