Advanced graphics processors, multi-core processors, and game physics engines have contributed to the rapid growth of games that inch ever closer to replicating real world physical interactions. Vehicles collide with objects and display realistic damage, structures hit by weapons crumble realistically, and trees bend and sway in the wind and snap off when fired at with small arms. The sources of these effects are sophisticated computational physics algorithms available to mainstream game developers through physics engines such as those developed by Havok and Ageia.

In order to simplify the game environments where these effects are used, many assumptions are made about the physical properties and the interactions of objects in these environments. If the game is being used purely for entertainment or if it is being used as a training tool where the realism of the interactions is not critical, these simplifying assumptions are acceptable. However, there are valid reasons to replace simplifying assumptions of the environments and physical responses of objects in the game with more realistic physical models. If the game is being used for experimentation where certain effects can alter the outcome of an experiment, better models may be warranted. Physics-based effects are also valuable for training where accurate weapons effects are important to training requirements.

In this paper, we discuss our research into the implementation of real-time physics models in a game environment. The objectives of this research are to demonstrate the feasibility of using physics-based weapons effects models in a game engine and to develop an approach for optimizing the models for real-time response. The US Army Research & Development Engineering Command Simulation and Training Technology Center (RDECOM -STTC) is currently researching and evaluating these real-time models to support experimentation and training.