University of Bergen | Faculty of Mathematics and Natural Sciences | Department of Informatics | Visualization Group
You are here: Department of Informatics > Visualization Group > Projects > PhysioIllustration > Comparative VIsualization of Spatial Ensemble Data
 > about
 > team & contact info
 > research
 > publications
 >  projects
 --- >  PhysioIllustration
 > teaching
 > seminars
 > resources
 > network
 > events
 > links
Comparative VIsualization of Spatial Ensemble Data

Nowadays, generation and study of spatial ensemble data is becoming more and more important for the research in the central scientific fields, such as medicine, biology, physics. A comparative visualization is highly suitable for the exploration of the ensemble members, especially if the spatial aspect is of upmost importance. The usefullness of the comparative visualization highly depends on the number of compared members, and on the visibility of the important characteristics of the members. We have created a model, which creates a comparative visualziation of ensemble data, showing as much of ensemble members as possible, and preserving important characteristics of the individual members. Here you can find a prototype implementation of our model available as Unity project.

A Fractional Cartesian Composition Model for Semi-spatial Comparative Visualization Design


Feel free to download the unity project, which contains a prototype implementation of our model. The project can be opened by the Unity Editor. In the prototype project hierarchy there are ensmeble input data (stored in xml format), 3 scenes which covers the demonstration of project and the scripts.

Three scenes (ivy, parp, city) covers the demonstration of the model described in the paper (Wall-covering Ivy, PARP polymerization, Parametrized Cities). Every scene have predefined GameObject (*manager) for accessing user interface of the model (for ivy example see Fig. 1).

The first step of the interaction with the model is to load an ensemble (Fig. 1a)---then the characteristics for the example are shown (Fig. 1b). Here, the user adjusts the importance of the characteristics for the optimization. After pressing "Build Samples" (Fig. 1c), our prototype picks ensemble members (randomly), applies the available abstractions and computes the selected characteristics in order to approximate the cost function for the optimization. For all the examples in this paper, we used an empirically-determined sample size of 10 ensemble members. Clearly, this is only a prototype implementation of the otherwise more general framework and ample opportunities for optimization are given here.

After the initial setup, the user interacts with the model parameters, characteristics, and axes (see Fig. 1d, b, e, respectively). After each parameter change, a preview of how the ensemble members will be positioned and how much visual space is available for each member is shown. If the user is satisfied, he/she can then create the final visualization (Fig. 1g).

 Last change: Ivan Kolesar, 2014-08-11