A Reusable Model for Emotional Biped Walk-Cycle Animation with Implicit Retargeting

This paper presents a reusable model for rapid animation of the walkcycle of virtual biped characters, with implicit retargeting of motion capture to any character, regardless of dimensions. Despite modern software continuously improving the quality of automatic assistance, the process of animating a biped character still requires substantial manual intervention. Our research contributes to this field by creating a theoretical model for emotional character walking, defining a series of proportional variables which can be changed to create different emotional walk cycles. We used motion capture data to assign real-world values to these variables, which are then used to procedurally create ‘emotional’ walk cycles. Due to the fact that we avoid fixed values and work solely with proportions, the system implicitly retargets the data to any biped body shape, regardless of the size and structure of the skeleton.