Dynamic Isosurface Extraction and Level-of-Detail in Voxel Space

摘要

A new visualization representation is described, which dramatically improves211u001einteractivity for scientific visualizations of structured grid data sets by 211u001ecreating isosurfaces at interactive speeds and with dynamically changeable levels-211u001eof-detail (LOD). This representation enables greater interactivity by allowing an 211u001eanalyst to dynamically specify both the desired isosurface threshold and required 211u001elevel-of-detail to be used while rendering the image. A scientist can therefore 211u001eview very large isosurfaces at interactive speeds (with a low level-of-detail), 211u001ebut has the full data set always available for analysis. The key idea is that 211u001evarious levels-of-detail are represented as differently sized hexahedral virtual 211u001evoxels, which are stored in a three-dimensional binary tree, or kd-tree; thus the 211u001elevel-of-detail representation is done in voxel space instead of the traditional 211u001eapproach which relies on surface or geometry space decimations. Utilizing the 211u001evoxel space is an essential step to moving from a post-processing visualization 211u001eparadigm to a quantitative, real-time paradigm. This algorithm has been 211u001eimplemented as an integral component of the EIGEN/VR project at Sandia National 211u001eLaboratories, which provides a rich environment for scientists to interactively 211u001eexplore and visualize the results of very large-scale simulations performed on 211u001emassively parallel supercomputers.