Shear-Warp Deluxe: The Shear-Warp Algorithm Revisited

Abstract

Despite continued advances in volume rendering technology, the Shear-Warp algorithm, although conceived as early as 1994, still remains the world s fastest purely software-based volume rendering algorithm. The impressive speed of near double-digit framerates for moderately sized datasets, however, does come at the price of reduced image quality and memory consumption. In this paper, we present the implementation and impact of certain measures that seek to address these shortcomings. Specifically, we investigate the effects of: (i) post-interpolated classification and shading, (ii) matched volume sampling on zoom, (iii) the interpolation of intermediate slices to reduce inter-slice aliasing, and (iv) the re-use of encoded RLE runs for more than one major viewing direction to preserve memory. We also study a new variation of the shear-warp algorithm that operates on body-centered cubic grids. We find that the reduction of the number of voxels that this grid affords translates into direct savings in rendering times, with minimal degradation in image quality.