Generating accurate visualizations of human faces presents unique challenges, particularly in capturing both large and fine-scale details. Most existing methods are either heavily data-driven, necessitating vast amounts of often inaccessible data, or they fail to capture intricate details due to the limitations of geometric face models. However, a new approach bridges this divide by employing strategies from traditional computer graphics. This innovative method models new expressions by blending the appearance from a limited set of extreme facial poses. The blending process involves measuring local volumetric changes in these expressions and reproducing their appearance when a similar expression is encountered. This technique not only enhances the representation of new expressions by adding detailed effects to smooth volumetric face deformations but also demonstrates its potential application beyond faces.