A Breakthrough in Remote Visualisation - November 2006
J. T. O'Brien; R. S. Kalawsky, Research School of Systems Engineering, Department of Electronic & Electrical Engineering, Loughborough University.
A breakthrough has been achieved at Loughborough University with adaptive visualisation rendering techniques for application scientists. High fidelity visualisations continue to demand large, expensive, resources to achieve the best insight for application scientists. Experiments in fluid dynamics, molecular chemistry and medical imaging are generating extremely large multi gigabyte data sets. Interactive 3D visualisation of this data is only possible by exploiting high-end rendering hardware. Access to this hardware presents a significant financial barrier for scientists. However without this access, achieving insight into their research is severely limited. Currently scientists must have physical access to these resources to interactively explore the visualisations. Research carried out by the author at Loughborough University has developed new tools to provide remote interactive access to high-end rendering hardware, without demanding that the scientists learn a new visualisation system or adapt their own applications. Remote visualisation is provided by a transparent process that scientists can harness from their workstations. The system reduces financial and technical barriers for scientists wishing to perform large scale visualisation whilst allowing research centres to develop new revenue streams from their visualisation hardware.
The remote render system uses an innovative adaptive rendering technique to provide guaranteed quality-of-service visualisation for the user. User specified parameters, including the relative importance of frame rate and visualisation quality, are used to inform the remote rendering system about how the visualisation should best be adapted to meet quality-of-service requirements and match available rendering resources. Rendering facilities at both the remote visualisation centre and local workstation are utilised to achieve faster and more responsive interaction for the user. Rather than simply act as a video stream for the high-end visualisation system, local rendering hardware on the scientists' workstation is exploited to introduce a local interaction loop that can be explored without further updates from the remote visualisation. This process increases the responsiveness of the visualisation and allows novel display modalities to be explored.
If stereo display (True 3D) is required to increase spatial awareness for viewers by exploiting human binocular vision, each rendered scene in a stereo display requires twice as much computation as a normal monoscopic display. This further exacerbates the visualisation performance requirements for the application. The remote render system allows distributed rendering resources to be combined, rendering the separate viewpoint of each eye in parallel. This process allows cheap stereographic hardware found in workstation graphics cards to display far larger visualisations than would normally be possible.
The author is currently in the process of implementing a version of the remote render system for a local NHS hospital. The system will provide clinicians in the hospital with collaborative access to 3D visualisations of large Computed Tomography (CT) and Magnetic Resonance Images (MRI) using both existing and new display devices around the hospital, without the need to invest heavily in visualisation hardware. The affordable access to stereoscopic displays delivered by the remote render system is being investigated as tool for improving patient diagnosis and for surgery planing, where spatial awareness of objects within a patient is particularly important. Initial exposure of the tool to clinicians in the DaVinci network (www.davinci-net.org) has received excellent reviews; surgeons in particular have remarked at how remote stereo rendering could offer large improvements for their work. An example of this is post operative surgery, where a surgeon must locate valves and other aids present in a patient.
The author is a PhD student at Loughborough University and the remote render system has been developed as part of the PhD research. The system developed is currently a prototype. It is envisaged that work with application scientists will help develop the system into a marketable product. The system offers clear economic benefit for researchers in both scientific and medical disciplines where the visualisation of very large data sets is a necessity. The adaptive rendering techniques used provide clear competitive advantage over other commercial systems.