Refine
Document Type
- Article (4)
- In Proceedings (4)
- ZIB-Report (2)
- Book chapter (1)
- Doctoral Thesis (1)
- In Collection (1)
- Poster (1)
Language
- English (14)
Keywords
- Datenanalyse (1)
- Tensorfelder (1)
- Visualisierung (1)
- anisotropic sampling of tensor fields (1)
- exploratory visualization (1)
- tensors of second order (1)
- visualization (1)
This work is concerned with adaptive screen-space sampling for volume ray-casting. The goal is to reduce the
number of rays being cast into the scene and, thus, the overall number of sampling points. We guarantee reliable
images through explicit error control using an error estimator that is founded in the field of finite element methods
(FEM). FEM theory further provides a well-founded theory to prove the efficiency of the presented algorithm via
convergence analysis. We, therefore, compare the convergence behavior of our method against uniform subdivisions
and a refinement scheme that was presented in the context of CPU volume ray-casting. Minimizing
the number of sampling points is of interest for rendering large datasets where each evaluation might need an expensive
decompression. Furthermore, with increasing screen resolutions high-resolution images are created more
efficiently with our method.
Three-Dimensional Second-Order Tensor Fields: Exploratory Visualization and Anisotropic Sampling
(2013)
Tensors provide a powerful mathematical language to describe physical phenomena. Consequently, they have a long tradition in physics and appear in various application areas, either as intermediate product or as output of simulations or measurements. The potential of tensors to describe complex anisotropic behavior, however, concurrently complicates their interpretation. The central research question of this thesis is how three-dimensional tensor fields of second order are visualized effectively so that, as a long term goal, their interpretation becomes easier. The focus of this thesis lies on the class of indefinite tensors. The methods that are proposed in this thesis fall into two main categories: (1.) the interactive exploration of the three-dimensional tensor data, and (2.) the geometric reduction of the data to two-dimensional planes or triangulated surfaces. In both cases, possible visualization approaches are presented. For interactive exploration of the data, we propose to combine diagram views with three-dimensional hybrid visualizations. We show that this facilitates familiarizing with the data and leads to exciting analytic queries. If a geometric data reduction is possible, we focus on glyph- and texture-based methods. In this context, the thesis is concerned with methods to improve their quality. Therefore, we propose two algorithms for the efficient creation of anisotropic sample distributions. Moreover, we present a novel visualization method that works on planar slices as well as on triangulated surfaces. The basic idea of this method is to use anisotropic sample distributions for the efficient computation of anisotropic Voronoi cells, which then are used as base elements for texture mapping. Hence, the usage of textures to encode the tensor’s various degrees of freedom becomes possible. We evaluate our methods for the interactive exploration on stress tensor fields from structure simulations. To show the ffectiveness of novel visualization methods, various datasets are presented.