Fluid Simulation in Computer Graphics

The target of this practical course is to develop a physical fluid simulator based on Smoothed Particle Hydrodynamics (SPH) including visualization techniques. The students will work in small teams and build upon a software framework written in C++ with OpenGL. Moreover, we will provide a server with version control system in form of a git repository. Within each team the students will be fully responsible for the distribution of the workload. This also requires the group to manage the software architecture of their project in order to be able to include the code contributions of each team member in a successful and efficient way. Besides the implementation all groups have to prepare a written report that elaborates the implemented methods, implementation details, results including comparisons and images representing the according simulation results.

Numerical Simulation

In order to simulate compressible as well as incompressible fluids, the Navier-Stokes equation has to be discretized in space and time and subsequently solved. The students will learn about the spatial discretization method SPH and explicit time integration schemes in order to numerically solve the underlying equations. Furthermore, advanced methods extending the basic SPH simulation will be implemented to enrich the model by complex physical phenomena such as incompressibility, viscosity, surface tension and solid-fluid interaction.

Graphics

All simulation results will be visualized with spherical particles using the OpenGL graphics pipeline. Moreover, the students are required to visualize physical quantities such as velocity, density or pressure using either particle coloring or general geometric primitives. Depending on the progress of the students we will provide the possibility to export the simulation results using an appropriate file format in order to visualize a reconstruction of the fluid's surface using a photorealistic offline renderer.

Modelling

When the implementation of the basic simulation methods are completed, the students will have to model complex scenarios to demonstrate the applicability of the implemented method in real-world examples. The groups can therefore either use geometric modelling tools such as Blender or Meshlab followed by particle sampling or generate the scenarios using their developed code.

Organizational Details

• Meetings: In every meeting each group has to present their implementation and results of the current topic and should discuss their problems and insights. This presentation should be prepared to take approx. 10-15 minutes. Please remember that attending the meetings is mandatory.
• Report: Length of approx. 15 pages excl. cover, table of contents, bibliography, etc.
• Implementation: Besides the code of the provided framework, the implementation should only consist of original code of the group members. However, we would like to encourage the groups to discuss occurring problems and to exchange information about their solutions.
• Additional meetings: If there are any group-individual meetings necessary, please contact the organizers via e-mail in order to get an appointment.

Requirements

• Good knowledge and practical experience in C/C++ and object-oriented programming
• Basic knowledge of numerics, algorithms and data-structures
• Good knowledge in modern OpenGL
• Knowledge of Computer Graphics on the level of the lecture "Basic Techniques of Computer Graphics"

Material

Will be announced in the introductory meeting.