Michaela Schmitt(TU Kaiserslautern)
hosted by Seminar Series on Scientific Computing
"Recent approaches in Opacity Optimization"
In the field of scientific visualization flow visualization is a long existing technique. Scientists and engineers aim to derive information from visualizations of flow data, but, less important structures frequently obscure features of interest.Thus, visual clutter and occlusion are common problems when it comes to inspecting flow data and different approaches have been developed to remedy for this problem.Two major approaches exist to solve this problem, one of them being seeding algorithms. By smartly placed seeding points the resulting visualization can be optimized. However, this approach is not performant for real time interaction since it is costly to recompute seed lines.The second class of approaches are selection algorithms. Instead of optimizing the choice of seed points, the set of lines is precomputed and clutter then reduced by smart selection of the lines computed earlier. Therefore, the generation of lines (e.g. numerical integration) is only performed during the preprocessing and will not be recomputed which makes it easier to interactively apply changes. Belonging to the selection-based algorithms is opacity optimization. It aims to maximize the opacity of important structures while removing clutter by minimizing the opacity of less important structures.
In this talk we will take a look on recent advances in the field of opacity optimization, focusing on the work from Zeidan et al., which proposed a novel technique utilizing moment-based techniques for signal reconstruction. Optimized opacities are calculated per fragment on an underlying, possibly arbitrary geometry.We will compare it to other recent work and conclude with a discussion of the advantages and disadvantages found in the different approaches.
|Time:||Thursday, 11.02.2021, 11:30|