Multi-Dimensional Transfer Function Design for Scientific Visualization

Direct volume rendering of field data can be accomplished through a correct choice of transfer functions that map data values to visual properties such as transparency and color. Such one-dimensional transfer functions, especially when specified through trial and error selection, often prove inadequate for producing correct and informative visualizations. In this paper, we consider multi-dimensional transfer functions for scalar fields which utilize field gradient magnitude and second derivative information, in addition to data values, to discriminate features with overlapping ranges. Our automatic method is further distinguished by our setting of voxels opacities, based on their spatial distance to boundaries in the specified volumetric scalar field data.