
Principal Component Analysis (PCA) is an established statistical technique for extracting the major axes of variation from a high dimensional set of data. It can be applied to images, here of faces and fish.
There are three sets of PCA animations available: