Exaggeration is often used in art and entertainment to capture the interest and attention of an audience. We present an approach to automatically exaggerate the distinctive features of 3D faces which contain skin detail down to the pores. Mesh adaptation and model simplification are used to produce two low resolution approximations of the face. The detail of the original high resolution face is captured by achieving two sets of model parameterizations: the high resolution face with respect to the face obtained using model simplification (simplified model) and the simplified model with respect to the model obtained using mesh adaptation (working model). The working model is exaggerated using a vector-based algorithm that automatically identifies the prominent features with respect to an average face. The resulting model and the parameterizations drive a two-stage model reconstruction process that generates the high resolution exaggerated model which preserves the original level of deta...