Master Thesis MSTR-2016-11

BibliographyDumitrescu, Otilia-Camelia: Visual Debugging for Particle-based Simulations of Fluids.
University of Stuttgart, Faculty of Computer Science, Electrical Engineering, and Information Technology, Master Thesis (2016).
57 pages, english.
CR-SchemaI.7.2 (Document Preparation)
Abstract

Visualizations represent an important tool that we have at our disposal when it comes to analyzing large data sets. A significant amount of data comes from simulations such as fluid, weather, biology and chemistry simulations. Due to increases in computation power the simulations have become more comprehensive, resulting in a larger amount of data. Increased volumes of the simulations require more specialized tools that can offer an insight so we can better understand the phenomena that is reproduced. The present thesis presents a visual debugging plug-in for Particle-based simulations of fluids that can help the researchers to better explain the simulation scenario and to identify possible errors. Moreover, the tool can be used to comprehend modeling and development of new techniques. The environment in which I have implemented the plug-in is MegaMol, a system software focus on visualizing particle-based simulations. There are four modules that I have implemented to enhance MegaMol functionality. In order to import a specific multidimensional data set I have created the BGEODataSource module which converts Houdini geometry formats into MegaMol Particle List Data (MMPLD). By doing this, the simulation data is available for other modules that are already implemented. To explore different particles that have certain properties I have created the ScatterPlot module that offers a way to select and visualize interesting regions of the attribute space. The user can select two attributes that will generate a scatter plot and interact with it by brushing. In order to get another perspective on the data I have implemented the ParallelCoordPlot module which allow the user to identify different patterns and trends between various attributes. By choosing distinct attributes we can see the correlation between different properties and clusters within a specific value range. The modules mentioned above work in the 2D space for observing the feature space. In SimpleSphereP ickingRenderer module we can select particles in the 3D space that will serve as input data for the ScatterPlot and ParallelCoordPlot. This is done by a simple selection of the region of interest.

Full text and
other links
PDF (14877614 Bytes)
Department(s)University of Stuttgart, Institute of Visualisation and Interactive Systems, Visualisation and Interactive Systems
Superviser(s)Weiskopf, Prof. Daniel; Huber, Markus; Krone, Dr. Michael; Reinhardt, Stefan
Entry dateAugust 1, 2018
   Publ. Computer Science