Most recent work in the area of human figure animation has concentrated on the editing of motion capture data. In these methods individual poses in the captured sequence can be edited, or goals such as footstep locations can be altered. In other cases, captured motion segments are stitched together to produce the desired animation.
Hodgins et.al. [16] apply an interpolation process to simulated motions to adapt them to different characters. The original motion is scaled to fit the new character in both limb measurements and mass and torque properties.
Popovic and Witkin [17] apply nonlinear optimization techniques to simplified versions of the original figure to reduce computation. A spacetime motion fitting technique is applied to the simplified character and the new motion is then mapped back onto the original figure. This method of motion editing is unique in that it takes dynamics into consideration when generating the new motion.
Lee and Shin [18] represent the motion through spacetime with multilevel B-splines which allow smooth interpolation between different edited keyframes of a captured motion. They compute a displacement map from the original motion to the desired one. The user is allowed to repose a key frame in the middle of the motion and the algorithm takes care of repositioning the figure in nearby frames. This method also allows the dynamic change of a character's physical attributes. This holds promise as a technique of introducing squash and stretch into the figure's constituent links.
Rose, Guenter, Bodenheimer and Cohen [19] discuss methods of producing smooth motion transitions between different sequences of captured motion. Their approach is once again physically based in that it attempts to preserve the figure's joint constraints during the transition by minimizing the torque required to achieve the transition. They also introduce a fast dynamics formulation to allow the solution of many degrees of freedom using constrained optimization.
The capture of real human motion has been used as necessary to accelerate the process of animation production. It falls short as a means of artistic expression for anyone but the actor whose motions are captured. Traditional animation has been an interpretation of real world motion. In the interpretation lies the art. Motion capture removes the interpretation of the animator in favor of hyper realistic motion that is constrained by real world physics. That is not to say that the editing of motion capture data cannot be considered an art, it is just difficult to consider it animation in the classical sense.
Table of Contents | < Physically Based Motion Generation | The Animator Program > |