We re-designed a library of software to build statistical models of the shape and grey-level appearance of objects in images. In addition we have developed a novel new algorithm for iteratively improving the fit of such a model to an image. It relies on learning the relationship between parameter displacements and residual image errors, allowing a prediction of the current offset of a model when matching an image. The algorithm allows us to fit full appearance models to images much more rapidly than before. This has been successfully applied to face images [10,11,12] and medical images .