George Baron Undergraduate Dissertation 2016/17

High Fidelity Visual Rendering of GPU Animated Fish Schools

Supervised by P.Richmond

Abstract

This dissertation aims to examine whether incorporating large-scale, agent-based models with high fidelity visual effects can be sufficiently achieved on modern GPU hardware with the processing capabilities it lends, and whether doing so provides a more immersive experience than the pre-calculated animation approaches that are more commonly used. The aim of this report is to introduce agent-based modelling and the software available for it, examine various techniques that can be used to render large-scale simulations of fish schools and discuss various approaches to rendering high-fidelity underwater visual effects. Further to this, extending the visualisation of the simulation to virtual reality hardware will also be discussed.