Fast Volume Rendering of Curvilinear Data Sets Using a Splatting Approach
We develope a new algorithm, that visualizes curvilinear grids fast and accurately. Our algorithm is based on the concept of splatting. In order to reconstruct the underly- ing function we use a Gaussian reconstruction kernel. In computational domain the sampling of the data field is regular and therefore we have spherical reconstruction kernels. However, the mapping into physical space is non-linear and the spherical reconstruction kernels from computational space are mapped to what we call ?bean- shapes?, which are usually not ellipsoids. We approximate these bean-shapes with ellipsoids and project them to the screen. Our algorithm is similar to Mao?s splatting algorithm, but is much more efficient and less computationally involved. It also more faithfully represents the underlying data.
- Torsten Möller
- Roger A. Crawfis
- Nelson Max