An efficient method for volume rendering using perspective projection

Abstract

Use of the perspective projection adds important perceptual cues for image comprehension. However, it has not been widely used in volume rendering because of the lack of efficient computational algorithms and concern over the nonuniform sampling rate imposed by perspective ray divergence. This paper introduces two new techniques to help make perspective projection more feasible in rendering volume data. First, a method is presented for efficient slice-by-slice processing of volume data, allowing high resolution data sets by eliminating typical memory constraints. Second, an adaptive "ray splitting" approach is described which ensures that the entire volume is sampled within user-specified limits. Additionally, we present results using distributed ray tracing to achieve depth of field effects.