Having looked at a variety of hardware and software based facial animation systems, it was now the task to view each one and derive a suitable methodology with which to animate from.  To begin with it was obvious to see that hardware motion capture was out of the question.  The hardware involved was far too expensive and prohibitive to use.  A software based approach using Maya 1.0 had to be the most flexible and viable solution.
                    Each software package had an advantage over the other.  FAMOUS's most innovative feature was the 3d cluster paint.  Maya employs a system whereby the user can deform geometry using a 3d paint interface.  Unfortunately 'Maya Artisan', as it is known, was not available for use.  'Realistic Human Faces' used a revolutionary 2d painting system to deform the base geometry from photographs, which ultimately produces the same effect as Artisan.  Maya can also implement 'Realistic Human Face's 'regional blending function'.  This is particularly attractive because of the range of facial expressions it can generate.

                    From studying the available options it became apparent that through using a regional blend; it is theoretically possible to generate any facial expression.  By employing a set of morph targets that represented every muscle used in the human face, the animator could have absolute flexibility and control over the expressions generated.  The next step was to research into exactly how and which facial muscles are used.  From the research it was possible break the muscle system down into the bare components necessary for facial animation.  It was clear however, that while the muscles for the top of a human face were not that extensive, the bottom half would be too impractical to implement.  Out of the eleven facial muscles that contribute to the human visage, seven directly apply to the cheeks and below.  Therefore, the bottom half was going to need re-thinking.  The next step was to decide upon and record the dialogue.  Once the sentences had been recorded, the next step was to record some live action footage of myself acting out the dialogue.  This was useful to have as a reference into how the muscles reacted to each other.  (see accompanying video tape for footage)
                    Based on the aforementioned research the following muscle scheme was implemented in the final product.  The next and final step was to begin animating.