Egocentric Mapping of Body Surface Constraints

Area of Science:

• Computer Graphics
• Human-Computer Interaction
• Animation

Background:

• Traditional performance animation struggles with mapping self-body contact postures across characters of varying proportions.
• Accurate representation of human body part relationships is crucial for conveying actions, intentions, and emotions.

Purpose of the Study:

• To develop a novel method for consistent performance animation of self-body contact postures.
• To enable accurate retargeting of complex human motions onto diverse digital characters.

Main Methods:

• Proposing an egocentric normalization of body-part relative distances.
• Utilizing character-independent egocentric coordinates to maintain posture space continuity.
• Employing a low-cost constraint relaxation technique with analytic inverse kinematics for online animation.

Main Results:

• The egocentric normalization method successfully preserves the spatial relationships of interacting limb segments.
• The approach ensures continuity of motion, even with self-contacts.
• Demonstrated superior performance in online retargeting compared to state-of-the-art methods, particularly for characters with significantly different proportions.

Conclusions:

• The proposed egocentric normalization method offers a robust solution for performance animation of self-body contacts.
• This technique enhances the accuracy and consistency of motion retargeting across a wide range of character designs.
• The method shows significant improvements over existing approaches, especially in challenging retargeting scenarios.