High demand on the productivity of the animation industry in Indonesia requires a change in the existing production process. Motion capture technology is the implementation of a computer vision principle to adopt the human eye senses to understand the phenomenon of motion results from a camera and mapped the virtual movement patterns. This paper will discuss a method for tracking marker features in the human face to obtain information about facial expressions. The tracking technique is using implementation of temporal coherence principle. This research assumes that using temporal coherence approach, the tracking process in sequential images can be simplified by calculating similarity on markers in each frame. The results show this feature-tracking process have reliable results to handle a lot of frames. Computation used very efficiently and cheap because it does not require a learning process in advance. The precision accuracy of tracking parameters generated a database of good visual expression.