EGPGV: Eurographics Workshop on Parallel Graphics and Visualization
Permanent URI for this community
Browse
Browsing EGPGV: Eurographics Workshop on Parallel Graphics and Visualization by Author "Dachsbacher, Carsten"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Hybrid Online Autotuning for Parallel Ray Tracing(The Eurographics Association, 2019) Herveau, Killian; Pfaffe, Philip; Tillmann, Martin Peter; Tichy, Walter F.; Dachsbacher, Carsten; Childs, Hank and Frey, SteffenAcceleration structures are key to high performance parallel ray tracing. Maximizing performance requires configuring the degrees of freedom (e.g., construction parameters) these data structures expose. Whether a parameter setting is optimal depends on the input (e.g., the scene and view parameters) and hardware. Manual selection is tedious, error prone, and is not portable. To automate the parameter selection task we use a hybrid of model-based prediction and online autotuning. The combination benefits from the best of both worlds: one-shot configuration selection when inputs are known or similar, effective exploration of the configuration space otherwise. Online tuning additionally serves to train the model on real inputs without requiring a-priori training samples. Online autotuning outperforms best-practice configurations recommended by the literature, by up to 11% median. The model predictions achieve 95% of the online autotuning performance while reducing 90% of the autotuner overhead. Hybrid online autotuning thus enables always-on tuning of parallel ray tracing.Item Moment-Based Opacity Optimization(The Eurographics Association, 2020) Zeidan, Mahmoud; Rapp, Tobias; Peters, Christoph; Dachsbacher, Carsten; Frey, Steffen and Huang, Jian and Sadlo, FilipGeometric structures such as points, lines, and surfaces play a vital role in scientific visualization. However, these visualizations frequently suffer from visual clutter that hinders the inspection of important features behind dense but less important features. In the past few years, geometric cluttering and occlusion avoidance has been addressed in scientific visualization with various approaches such as opacity optimization techniques. In this paper, we present a novel approach for opacity optimization based on recent state-of-the-art moment-based techniques for signal reconstruction. In contrast to truncated Fourier series, momentbased reconstructions of feature importance and optical depth along view rays are highly accurate for sparse regions but also plausible for densely covered regions. At the same time, moment-based methods do not suffer from ringing artifacts. Moreover, this representation enables fast evaluation and compact storage, which is crucial for per-pixel optimization especially for large geometric structures. We also present a fast screen space filtering approach for optimized opacities that works directly on moment buffers. This filtering approach is suitable for real-time visualization applications, while providing comparable quality to object space smoothing. Its implementation is independent of the type of geometry such that it is general and easy to integrate. We compare our technique to recent state of the art techniques for opacity optimization and apply it to real and synthetic data sets in various applications.