Interactive Streak Surface Visualization on the GPU

Abstract

In this paper we present techniques for the visualization of unsteady flows using streak surfaces, which allow for the first time an adaptive integration and rendering of such surfaces in real-time. The techniques consist of two main components, which are both realized on the GPU to exploit computational and bandwidth capacities for numerical particle integration and to minimize bandwidth requirements in the rendering of the surface. In the construction stage, an adaptive surface representation is generated. Surface refinement and coarsening strategies are based on local surface properties like distortion and curvature. We compare two different methods to generate a streak surface: a) by computing a patch-based surface representation that avoids any interdependence between patches, and b) by computing a particle-based surface representation including particle connectivity, and by updating this connectivity during particle refinement and coarsening. In the rendering stage, the surface is either rendered as a set of quadrilateral surface patches using high-quality point-based approaches, or a surface triangulation is built in turn from the given particle connectivity and the resulting triangle mesh is rendered. We perform a comparative study of the proposed techniques with respect to surface quality, visual quality and performance by visualizing streak surfaces in real flows using different rendering options.

IRIS: Illustrative Rendering for Integral Surfaces

“…Using a GPU-based approach, Schafhitzel et al [29] presented a point-based algorithm that does not compute an explicit mesh representation but rather uses a very dense set of particles, advected at interactive speeds, in combination with point-based rendering. Recently, Krishnan et al [21], Bürger et al [5] and von Funck et al [34] presented approaches for time and streak surface computation. While the former authors focused on the CPU treatment of large CFD datasets, the latter designed their approach for GPUs with the aim of real-time visualization for smaller datasets.…”

“…[10,21]), and interaction with the surface in a near real-time setting -possibly even during the surface computation -is highly desirable. For less complex data, the recent work of Bürger et al [5] describes a real-time computation approach that leverages the computing power of GPUs, and we aim at retaining the applicability of the methods described in this paper in such a scenario. Similarly, the dynamic and evolving nature of time and streak surfaces attractively captures the temporal characteristics of flows; as such, the ability to animate integral surfaces is pertinent to our considerations.…”

“…triangles or quadrilaterals), with a set of two texture coordinates associated with each vertex. This representation need not be connected and we explicitly accommodate some computation approaches (such as the method by Bürger et al [5]) that generate surfaces as (partial) primitive soup. The normal-variation transparency approach described in Section 3.2 requires continuously varying normals over the mesh; in this case, the normal must be specified per vertex.…”

Particle advection performance over varied architectures and workloads

“…Multiple studies ( [19], [20], [21], [22]) have focused on GPU implementations of particle advection problems for desktop machines with a single GPU. In all cases, the particle advection workloads considered required significant computational resources, and the GPU was found to be superior when compared to a CPU.…”

Interactive Virtual Probing of 4D MRI Blood-Flow

“…Lastly, we incorporate integral surfaces. Garth et al [7] have described a generic hardware-accelerated approach for generating pathsurfaces, while Bürger et al [3] presented a real-time technique for the generation of streak surfaces. Recently, Born et al [2] and Hummel et al [10] introduced distinct illustrative visualization styles for integral surfaces.…”