Visualization techniques for vector field data

In this work a technique for visualizing steady flow is proposed. Using this technique, we first convert the vector field data into a scalar level-set representation. We then analyze the dynamic behavior and subsequent distortion of level-sets and interactively monitor the evolving structures by means of texture-based surface rendering. Next, we combine geometrical and topological considerations to derive a multiscale representation and to implement a method for the automatic placement of a sparse set of graphical primitives depicting homogeneous streams in the fields. Using the resulting algorithms, we have built a visualization system that enables us to effectively display the flow direction and its dynamics even for dense 3D fields. The ideas developed and presented in this work will be influential for future developments:

• We have derived a multiscale representation for flow fields. This can lead to a multiresolution framework for flow, where only principal streams on a lower resolution level and the difference information needed for the next finer levels is included.
• We have demonstrated that flow field direction and speed can be visualized very effectively via texture-based rendering and by automatic extraction and placement of the principal streams. This enables us to appropriately visualize dense 3D flow fields.
• The multiscale representation might result in even more efficient integration schemes for particle tracing. By employing the multiscale nature of the analyzed flows, as well as their geometric and topologic structure, traditional integration schemes can be extended in order to control the step size adaptively and thus to improve their efficiency.

Kontakt: Ruediger Westermann