Evolving behaviours for a real-time autonomous camera

The quality of autonomous camera systems is crucial for the elegance and success of 3rd-person perspective computer games. We must ensure that these systems are fully capable of correctly performing camera movements in real-time within the interactive computer game environment. Our previous work demonstrated a highly effective constraint-weighting based camera system for producing believable autonomous camera movements. However, it had a complex interface for artists and game designers. To alleviate this problem, we describe a novel use of genetic algorithms for the purpose of evolving camera profiles to match the desired properties, by searching the space of possible camera profiles. As part of the experimental study, we implemented a 3D game engine to demonstrate the effectiveness of the use of constraint-weighting techniques for camera control. We also demonstrate the effectiveness of genetic algorithms in evolving camera profiles.