Widening the remote visualization bottleneck

With the advent of generic remote visualization solutions, accessing distant graphics resources from the desktop is no longer restricted to specially tailored applications. However, current solutions still suffer from poor interactivity due to limited network bandwidth and high latency. In this paper we present several improvements to our remote visualization system in order to overcome these drawbacks. Different image compression schemes are evaluated with regard to applicability for scientific visualization, showing that image compression alone cannot guarantee interactive frame rates. Therefore, we furthermore present different techniques for alleviating the bandwidth limit, including quality reduction during user interaction or animations. Latency aspects are addressed by employing multi-threading for asynchronous compression and utilizing features of modern programmable graphics adapters for performing the image compression entirely or the graphics card. The presented ideas are integrated in our remote visualization system without violating universality.