Für eine korrekte Darstellung dieser Seite benötigen Sie einen XHTML-standardkonformen Browser, der die Darstellung von CSS-Dateien zulässt.

You are here

  • English
  • Deutsch

BA/MA Thesis

Below you will find a list of topics for a thesis, and feel free to contact us.

 

Titlesort descending Thesis Type Supervisor Requirements Short Description
Anisotropic Smoothed Particle Hydrodynamics MA M. Sc. Rene Winchenbach Computer Graphics IV The goal of this method is to evaluate possible benefits that anisotropy can bring to SPH. The student would have to implement such a method, e.g. arxiv.org/abs/astro-ph/9512078 , in our existing simulation framework using CUDA and C++. Due to the difficulty of the method itself, prior knowledge in SPH is definetly beneficial.
Blur kernel estimation by deep learning MA Tak Ming Wong Recently, computer vision community developed several motion deblurring and single image super-resolution methods, which model the deblurring process by deep neural network. However, estimating an explicit blur kernel still plays an important role in image processing perspective. The aim of this work is to study in blur kernel estimation on conventional RGB image using state-of-arts deep learning technique.
Influence of Body Sampling on the Interlinked SPH BA Rustam Akhunov The goal of this work will be the testing of the influence of different rigid body sampling on Interlinked SPH rigid body solver. The candidate will learn about the modern body sampling techniques and apply them to the provided rigid body solver in order to check the effect of different surface structuring.
Modern Material Models for Monte-Carlo Path Tracing MA Martin Lambers Computer Graphics III In this thesis, the candidate will review the relevant literature both on modern material models and on importance sampling of materials for Monte-Carlo path tracing, and construct a physically plausible material model that is able to reproduce the behavior of a wide variety of real-world materials while still supporting proper importance sampling as far as possible.