The recent development of highly powerful programmable graphics cards has made it possible to use Ray Tracing graphics techniques for interactive simulation applications.

Previously, Ray Tracing (RT) techniques have been used for animated film production and non-real time applications where images may be generated in minutes or hours. The same techniques can now be used to generate images in 1/30^th or 1/60^th second for use where the image must change in response to operator input - for example in ground vehicle or flight simulators.

Use of Ray Tracing techniques potentially allows the same scene to be rendered across a variety of wavebands (optical, thermal, electromagnetic) using broadly similar algorithms, reducing the number of different databases needed and promoting correlation, interoperability and reuse as well as higher image quality. Another potential benefit is reducing the workload, time and most importantly the cost incurred for synthetic natural environment database production, especially in the advanced texturing techniques now used to produce high quality â€˜serious' games.

This paper describes work funded by the UK Ministry of Defence, and undertaken by XPI Simulation Ltd. and Lockheed Martin UK Insys Ltd., to evaluate the feasibility and implementation of ray tracing techniques for improving the quality and usability of simulation graphics. The paper describes the problems encountered in achieving real time rates for simulation and training applications - for example the performance hits incurred handling moving 3D models with articulated parts, along with some solutions to these issues that have been found.

The paper describes the future potential of real time ray tracing techniques given the steadily improving cost / performance ratio of PC graphics hardware, and the effect that implementation of RT techniques will have on the high performance graphics used in simulation and training visualization.

Menu

Craig Hutchinson