Volume Rendering for Computational Fluid Dynamics: Initial Results

Abstract

This work describes a direct volume rendering technique adapted for visualization of results from Computational Fluid Dynamics (CFD) flow solvers. The problem is difficult for at least three reasons: (1) the shapes of the grids, (2) vector as well as scalar data fields, and (3) the types of features of interest. In addition, the data set sizes are often between ten and one hundred megabytes, which is large even by volume rendering standards. The method, a ray casting implementation, is chosen for its robustness and adaptability. The software described here differs from other volume renderers in the handling of vector field data and in the techniques for reducing the time required for intersecting rays with grid cell faces. Vector fields are rendered using mappings from the vector field space to three dimensional color spaces plus opacity. Several mappings are considered and the relative merits analyzed. Two techniques for improving the speed of the ordinarily slow ray casting are described. Future developments and improvements are also discussed.