OpenGLSL-based Raycasting - Comparison of Execution Durations of Multi-pass vs. Single-pass Technique

Abstract

Real time volume rendering of medical datasets using raycasting on graphics processing units (GPUs) is a common technique. Since more than 10 years there are two established approaches for realizing GPU ray casting: multi-pass (Kruger and Westermann, 2003) and single-pass (Röttger, et al., 2003). But the required parameters to choose the optimal raycasting technique for a given application are still unknown. To solve this issue both raycasting techniques were implemented for different raycasting types using OpenGLSL vertex and fragment shaders. The different techniques and types were compared regarding execution times. The results of this comparison show that there is no technique faster in general. The higher the computational load the more indicates the use of the multi-pass technique.