The goal of facial animation is to synthesize realistic facial animated images using computer graphics. Because of its capability in creating facial animated images using a few amount of information, facial animation using feature vectors was extensively studied in recent years. In general, this method is considered as one of the performance-driven facial animation base methods. In this regard, facial feature vectors, which reflect rigid and non-rigid movements of the face, are extracted using facial analysis methods. These feature vectors are then used for transferring the facial movements to a graphical model. Our approach in this paper is merging keyframing and facial animation using feature vectors. Using this method, the amount of the submitted information is decreased to about 30%; while the synthesized sequences have 4.95% mean squared error and 0.000629 difference of correlation relative to input sequences. This error is negligible compared to the error of synthesized sequences without using interpolation.