Data Centric Transfer Functions for High Dynamic Range Volume Data

Abstract

Creating effective transfer functions for high dynamic range scalar volume data is a challenging task. For data sets with limited information about their content, deriving transfer functions using mathematical properties (gradient, curvature, etc.) is a difficult trial and error process. Traditional methods use linear binning to map data to integer space for creating the transfer functions. In current methods the transfer functions are typically stored in integer look-up tables, which do not work well when the data range is large. We show how a process of opacity guidance with simple user interface can be used as the basis for transfer function design. Our technique which uses opacity weighted histogram equalization lets users derive transfer functions for HDR floating point easily and quickly. We also present how to adopt these techniques for real-time interactive visualization with minimal pre-processing. We compare our techniques with traditional methods and show examples.