For example, if motion patterns 600 are all possible motion patterns for control point 427 with equal probability assigned to each motion pattern, then when an animation is generated, the associated computer model will be used to generate an output video with the boundaries 428 moving in conjunction with control point 427 with the motion from motion patterns 600 assigned to control point 427 at a given time. The motion assigned for a given time can change based on the probability assigned for a given motion state and a motion pattern of a particular motion state. 
As described herein, an animation is assigned a motion state for a given time or time period. The motion state is selected from the state-space of all possible motion states for the computer animation model. Different computer animation models can have different frameworks (e.g. combinations of variables or data structures) for the associated state-space of the model. In one example embodiment, an animation framework associated with a computer animation model comprises A control points (e.g. control point 527 or control points associated with eyebrow areas 450, 451), B motion patterns (e.g. motion patterns 610-620), and C tempo harmonics (e.g. harmonic multiples of tempo 504 determined from beats 502) such that a number of motion states N for the computer animation model within the animation framework is: N=(B×C)A??(1)