This paper describes new methods that are used to adapt a 3D polygon model of a face, which is then used in a facial recognition system. In this paper, a new method is proposed, which can adapt a 3D model of a face, while using a minimum number of 2D images of a face (full face, profile, halfturn). Control points and distances between them are used to extract eyes, nose and other fragments from a face, and to find links between them in different projections. The results of the proposed algorithm are compared with a Faceworx program.