Interactivity is an important aspect of a visualization system since it permits rapid exploration and visual analysis of the relevant data
sets and as such avoiding costly pre-computations is of key importance. The volume shading methods were designed to account for this and to enable interactive visualizations of volumetric and geometric data sets by exploiting the parallel computing powers available on modern graphics processing units.
There are four main themes for better illumination of volumetric data: translucent shading without using surface normals, occlusion shading, combined geometric and volumetric occlusion, and depth of field for volume rendering. Volumetric directional occlusion shading method allows computation of a subset of the global ambient occlusion solution by restricting the occlusion computation to a viewer-oriented cone. This enables an efficient implementation in a slicebased direct volume rendering system, while at the same time providing plausible
depth cues similar to those of a full ambient occlusion computation.
Rendering of combined occlusion effects from both volumetric and geometric structures extends the benefits of enhanced depth perception to data sets combining geometry and volumes. Those originate in many varied fields of scientific
visualization, where tube shaped structures are situated within an associated scalar volume. Common examples are DTI fiber tractography or streamline tracing. The method for computing depth of field effects supports an alternative way of providing supplementary depth cues within the context of direct volume
rendering. They are especially useful for data sets, such as those arising from combustion simulation, where the presented occlusion shading method is less effective due to lack of sufficiently opaque, surface-like structures.
Research Seminar Announcement
In the course of the Visual Computing research seminar Prof. Dr. Nils Thuerey from the Technical University of Munich will give a talk with subsequent discussion about the topic Fluid Simulations and Neural Networks.
We would like to invite everyone interested to join us. The research seminar will take place in on Friday 30.06.2017, 13:15 in G29-R335.