Tobias Germer

RenderBots—Multi-Agent Systems for Direct Image Generation

The term stroke‐based rendering collectively describes techniques where images are generated from elements that are usually larger than a pixel. These techniques lend themselves well for rendering artistic styles such as stippling and hatching. This paper presents a novel approach for stroke‐based rendering that exploits multi‐agent systems. RenderBots are individual agents each of which in general represents one stroke. They form a multi‐agent system and undergo a simulation to distribute themselves in the environment. The environment consists of a source image and possibly additional G‐buffers. The final image is created when the simulation is finished by having each RenderBot execute its painting function. RenderBot classes differ in their physical behavior as well as their way of painting so that different styles can be created in a very flexible way.

Uncertain 2D Vector Field Topology

We introduce an approach to visualize stationary 2D vector fields with global uncertainty obtained by considering the transport of local uncertainty in the flow. For this, we extend the concept of vector field topology to uncertain vector fields by considering the vector field as a density distribution function. By generalizing the concepts of stream lines and critical points we obtain a number of density fields representing an uncertain topological segmentation. Their visualization as height surfaces gives insight into both the flow behavior and its uncertainty. We present a Monte Carlo approach where we integrate probabilistic particle paths, which lead to the segmentation of topological features. Moreover, we extend our algorithms to detect saddle points and present efficient implementations. Finally, we apply our technique to a number of real and synthetic test data sets.

Procedural Arrangement of Furniture for Real-Time Walkthroughs

This paper presents a procedural approach to generate furniture arrangements for large virtual indoor scenes. The interiors of buildings in 3D city scenes are often omitted. Our solution creates rich furniture arrangements for all rooms of complex buildings and even for entire cities. The key idea is to only furnish the rooms in the vicinity of the viewer while the user explores a building in real time. In order to compute the object layout we introduce an agent‐based solution and demonstrate the flexibility and effectiveness of the agent approach. Furthermore, we describe advanced features of the system, like procedural furniture geometry, persistent room layouts, and styles for high‐level control.

Uncertain topology of 3D vector fields

We present a technique to visualize global uncertainty in stationary 3D vector fields by a topological approach. We start from an existing approach for 2D uncertain vector field topology and extend this into 3D space. For this a number of conceptional and technical challenges in performance and visual representation arise. In order to solve them, we develop an acceleration for finding sink and source distributions. Having these distributions we use overlaps of their corresponding volumes to find separating structures and saddles. As part of the approach, we introduce uncertain saddle and boundary switch connectors and provide algorithms to extract them. For the visual representation, we use multiple direct volume renderings. We test our method on a number of synthetic and real data sets.

Camera textures

In this paper we introduce a novel real-time rendering technique for camera deformations that can be applied to lens distortions and non-realistic projections. Our technique is based on vertex shader textures and presents a hybrid approach working in both, image and object space. As the primal deformation is achieved by a vertex deformation in object space, our technique does not exhibit any artefacts known from pixel-based scaling techniques.Additionally, we present a novel approach for an automatic lens-object tracking that assists the user in manipulating and interacting 3D objects within the scene. Here, we developed a GPU-based technique for fast bounding box determination that is defined by a set of arbitrary vertices. We also discuss different techniques allowing the user to accurately select these vertices in a nonlinearly distorted environment.

SpringLens Distributed Nonlinear Magnifications

We present a flexible, distributed and effective technique to model custom d istortions of images. The main idea is to use a mass-spring model to create a flexible surface and to create distor tions by changing the rest-lengths. A physical simulation works out the displacements of this particle grid. We pro vide intuitive tools to interactively design such nonlinear magnifications. In addition, our system enables da ta-driven distortions which allows us to use it for automatic nonlinear magnifications. We demonstrate this with an applic tion for labeling of 3D scenes.

Closed stream lines in uncertain vector fields

We present a method for the detection and visualization of closed stream lines topologically acting as sources or sinks in uncertain 2D and 3D vector fields. For their detection, we apply a Monte Carlo simulation which generates particle distribution functions representing sinks and sources. We show that in the uncertain case there is no structural difference between critical points and closed orbits. This allows the application of critical point extractors to closed stream lines as well. We show applications for some synthetic and real world examples.

Stream Surface Parametrization by Flow-Orthogonal Front Lines

The generation of discrete stream surfaces is an important and challenging task in scientific visualization, which can be considered a particular instance of geometric modeling. The quality of numerically integrated stream surfaces depends on a number of parameters that can be controlled locally, such as time step or distance of adjacent vertices on the front line. In addition there is a parameter that cannot be controlled locally: stream surface meshes tend to show high quality, well‐shaped elements only if the current front line is “globally” approximately perpendicular to the flow direction. We analyze the impact of this geometric property and present a novel solution – a stream surface integrator that forces the front line to be perpendicular to the flow and that generates quad‐dominant meshes with well‐shaped and well‐aligned elements. It is based on the integration of a scaled version of the flow field, and requires repeated minimization of an error functional along the current front line. We show that this leads to computing the 1‐dimensional kernel of a bidiagonal matrix: a linear problem that can be solved efficiently. We compare our method with existing stream surface integrators and apply it to a number of synthetic and real world data sets.

Mutual text-image queries

This paper presents a novel approach to support students to learn a comprehensive domain-specific terminology and to understand textual descriptions of complex-shaped objects. We implemented an experimental system where learners can interactively explore textual descriptions and 3D visualizations. We propose a method for hierarchical content representations of text documents and views on 3D models. Based on these data structures, user interactions on texts and interactive 3D visualizations are transformed into queries to an information retrieval system. This enables us to coordinate the content of both media, to focus the attention of the user on the most salient graphical objects, and to suggest potential relevant text segments in large text documents and appropriate views on 3D models to illustrate the spatial relations between the relevant domain objects of the query. Finally, we demonstrated this concept in an interactive tutoring environment based on standard textbooks on human anatomy.

As-Perpendicular-as-possible surfaces for flow visualization

We define APAP surfaces, surfaces that are as perpendicular as possible to steady 3D vector fields, and present a method to construct discrete representations of them. Since, in general, a perfectly perpendicular surface to a vector field does not exist, we propose and minimize an error metric to enforce perpendicularity as much as possible. Our algorithm constructs an APAP surface by deforming a seed surface anchored in a domain point. In the discrete setting this minimization results in iteratively solving linear least-squares problems and integrating a locally scaled version of the vector field. The definition of the error metric and its numerical minimization guarantee that the minimum zero is attained for the perfectly perpendicular surface if it exists. Otherwise, the minimization converges to the same local minimum independent of the seed configuration, and the resulting surface is - in a least-squares sense - as perpendicular as possible to the flow. We apply these APAP surfaces as an interactive flow visualization tool which we demonstrate on a number of synthetic and real flow data sets.