Volume visualization: advances in transfer and opacity function generation for interactive direct volume rendering

Abstract

Direct volume rendering is an advanced technique for visualizing the structure of volume data sets. It maps directly from the data set to the rendered image without any intermediate geometric calculations. This paper presents a method to generate high-quality renderings by designing opacity functions that (1) are correctly correlated to particular data values of interest, (2) discriminate regions of interest from remaining data values, and (3) dynamically control the transfer function by relating values located in the raw volume data set to color and specified opacity values. Gradient constructs are formulated to capture more effectively subtle features within the volume. Gradient calculations are performed on directional derivatives along gradient directions. The method described herein is applied to data generated at the GE Aircraft Engines in Evansdale, OH which are available through the Stanford's Volpack library.