• DocumentCode
    3031242
  • Title

    Automatic skeleton generation and character skinning

  • Author

    Cheng, Wang ; Cheng, Ren ; Xiaoyong, Lei ; Shuling, Dai

  • Author_Institution
    State Key Lab. of Virtual Reality & Syst., Beihang Univ., Beijing, China
  • fYear
    2011
  • fDate
    19-20 March 2011
  • Firstpage
    299
  • Lastpage
    304
  • Abstract
    An articulated character skinning usually requires manual skeleton embedding and vertex weight painting. We propose a fast and automatic method for skeleton generation and character skinning. First, we segment the given character mesh by the sequential steps of NCV(normal characteristic value) computation, segment points refinement, and principal component analysis of segment clusters. Then, two types of joints and a skeleton are generated based on the mesh segmentation result. Furthermore, we compute weights of vertices influenced by skeleton automatically and then skinning the character mesh by skeleton driven and muscle pushing algorithm. Experimental results show that our method achieves both high visual quality and fast speed. It could be used in character animation and VR real time applications.
  • Keywords
    computer animation; mesh generation; pattern clustering; principal component analysis; NCV computation; articulated character skinning; automatic skeleton generation; mesh segmentation; muscle pushing algorithm; normal characteristic value; principal component analysis; segment points refinement; segmented cluster; skeleton embedding; vertex weight painting; Bones; Joints; Muscles; Principal component analysis; Skin; Surface morphology; Character Animation; Character Skinning; Segmentation; Skeleton Generation; Virtual Reality;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    VR Innovation (ISVRI), 2011 IEEE International Symposium on
  • Conference_Location
    Singapore
  • Print_ISBN
    978-1-4577-0055-2
  • Electronic_ISBN
    978-1-4577-0054-5
  • Type

    conf

  • DOI
    10.1109/ISVRI.2011.5759655
  • Filename
    5759655