This thesis is dedicated to the loving memory of my dad

The aim of this thesis is to investigate computational models of non-verbal social inter-action for the purpose of generating synthetic social behaviour in animations. To this end, several contributions are made: A dynamic model, providing multimodal control of animation is developed and demonstrated using various data formats including mo-tion capture and video; A social interaction model is developed, capable of predicting social context/intent such as level of interest in a conversation; and finally, the social model is used to drive the dynamic model, which animates appropriate social behaviour of a listener in a conversation in response to a speaker. A method of reusing motion captured data by learning a generative model of motion is presented. The model allows real-time synthesis and blending of motion, whilst providing it with the style and realism present in the original data set. This is achieved by projecting the data into a lower dimensional space and learning a multivariate probability distribution of the motion sequences. Functioning as a generative model, the probability density estimation is used to produce novel poses, and pre-compute