A New Framework for Visualizing a Time-Varying Unstructured Grid Dataset with PBVR

Abstract

In this paper, we propose a new framework for a large-scale time-varying unstructured grid dataset visualization based on the Particle-Based Volume Rendering (PBVR). A previous rendering system can handle a time-varying grid dataset which is composed of only hundreds of thousands of unstructured grid cells. To overcome the limitation, we employed the PBVR in which the computational complexity of rendering is dominated not by the number of cells but by that of particles. Regardless of the number of cells, the PBVR can efficiently render a large-scale grid dataset by reducing particles in the location or the time frame which is not so important. In this system, we develop a hierarchical framework for managing the particles for efficient rendering. Moreover, by compressing the particle data, we also succeed in reducing the loading time of the whole time step particle data from the hard disk. We applied the system to rendering of 465 time steps of 3,676,828 hexahedral cell grid dataset to confirm the effectiveness.