Henshall, Gareth I.Teahan, William J.Cenydd, Llyr apTao Ruan Wan and Franck Vidal2017-09-212017-09-212017978-3-03868-050-5https://doi.org/10.2312/cgvc.20171284https://diglib.eg.org:443/handle/10.2312/cgvc20171284The advent of high-fidelity Virtual Reality (VR) allows us to experience virtual characters with an ever increasing sense of presence. VR is capable of enhancing the perception of many aspects of virtual characters, including their volume and texture, speed, weight and temperament. The intimate nature of such experiences also requires a greater attention to behavioural realism, player interaction and animation detail. In this paper, we describe our research in developing techniques for optimising complex procedural animation systems for VR. We start by describing a dolphin animation system and how we use crowd-sourcing and genetic algorithms to search the parameter space for realistic behaviour. We then explain some key differences we have found between dolphins optimised using a desktop monitor and ones optimised in VR. Finally, we describe our current work which aims to further enhance the speed and intuitiveness of animation optimisation in VR.10.2312/cgvc.2017128495-96