Visapult: A Prototype Remote and Distributed Visualization Application and Framework

Remote visualization is the process of creating pictures from data, where the data and viewer are on different platforms, often separated by a wide area network. Traditional approaches to remote visualization have tended to fall into two broad categories. Render-remote approaches are implementations where data storage and access, visualization, rendering and viewer involve a single machine. Render-local approaches require transmission of data to the viewer’s machine, where it is then visualized and rendered. The former approach has the potential to address large-data problems, but sacrifices interactivity. The latter approach has the potential to preserve interactivity, but assumes that the data will fit entirely on the local workstation, and that it is practical to transmit the data across the network. Our Approach We present Visapult, a prototype visualization application and framework that strikes a balance between render-local and render-remote approaches. Visapult was designed to address the needs of scientific researchers who produce and need to visualize hundreds of gigabytes, if not terabytes, of time-varying data. With data of this scale, neither transmission across a network nor local storage is practical, yet interaction with the visualization is crucial. We did not want to sacrifice the gains in understanding that result from motion parallax and stereoscopic presentation [1]. To achieve these goals, we have implemented an