Motion Captured Data Driven Skinning and Animation of 3D Virtual Human
Download as PDF
DOI: 10.25236/mmmce.2020.025
Corresponding Author
Chen Xiaoman
Abstract
The motion simulation of 3D virtual human is becoming a hot research topic in recent years in both academia and industry. However, the current mainstream skin deformation algorithm, that is, linear blending skinning (LBS) algorithm will cause some distinct distortion problems such as skin collapsion or contortion. To void these distortions, this paper proposed a dual quaternions blending skinning (DQBS) algorithm for skin deformation that is driven by the motion captured data. Firstly, the skeleton model of the 3D virtual human is extract from the motion captured data files, and further binding to the skin model built by the 3D modeling software. Then the DQBS algorithm is introduced to deform the skin and its algorithm flow chart is proposed. Finally, a simulation software system is developed and both LBS and DQBS algorithms are simulated and compared in the system. The result shows that the proposed DQBS algorithm can improve the distortion caused by LBS.
Keywords
3d virtual human, Skin deformation algorithm, Dual quaternions blending skinning, Motion captured data