Efficiently using graphics hardware in volume rendering applications

Abstract

OpenGL and its extensions provide access to advanced per-pixel operations available in the rasterization stage and in the frame buffer hardware of modern graphics workstations. With these mechanisms, completely new rendering algorithms can be designed and implemented in a very particular way. In this paper we extend the idea of extensively using graphics hardware for the rendering of volumetric data sets in various ways. First, we introduce the concept of clipping geometries by means of stencil buffer operations, and we exploit pixel textures for the mapping of volume data to spherical domains. We show ways to use 3D textures for the rendering of lighted and shaded iso-surfaces in real-time without extracting any polygonal representation. Second, we demonstrate that even for volume data on unstructured grids, where only software solutions exist up to now, both methods, iso-surface extraction and direct volume rendering, can be accelerated to new rates of interactivity by simple polygon drawing and frame buffer operations.

Volumetric video compression for interactive playback

“…Current state-of-the-art graphics hardware enables real-time volume and isosurface rendering [28] and decompression [14].…”

Fast re-rendering of volume and surface graphics by depth, color, and opacity buffering

“…In our implementation, the compositing is being done outside the geometry engine. However, we invision that it will be possible to use the primitive operations available in the OpenGL pipeline (Neider et al, 1992) to perform the compositing wholly in the frame graphics (Westermann and Ertl, 1998). Amongst a number of issues being investigated is the performance of the operations involved in the merging process, such as the depth culling of the geometry every time its position and orientation is changed.…”

A refined data addressing and processing scheme to accelerate volume raycasting

“…Two of them are hardware based; the first one utilizes commodity graphics-cards; the second one utilizes special purpose hardware, e.g. VolumePro and Vizard; the third is CPU based [1][2][3][4][5][6]. Purely hardware based solutions provide real-time performance and high quality; however they are limited in their functionalities: Basic visualization systems are supported by hardware volume rendering solutions; consequently they are the mostly applied approach in practice.…”