Adaptive Wisp Tree - a multiresolution control structure for simulating dynamic clustering in hair motion

Florence Bertails, Tae-Yong Kim, Marie-Paule Cani, Ulrich Neumann
ACM-SIGGRAPH/EG Symposium on Computer Animation (SCA) - July 2003
Download the publication : sca03.pdf [1.7Mo]  
Realistic animation of long human hair is difficult due to the number of hair strands and to the complexity of their interactions. Existing methods remain limited to smooth, uniform, and relatively simple hair motion. We present a powerful adaptive approach to modeling dynamic clustering behavior that characterizes complex long-hair motion. The Adaptive Wisp Tree (AWT) is a novel control structure that approximates the large-scale coherent motion of hair clusters as well as small-scaled variation of individual hair strands. The AWT also aids computation efficiency by identifying regions where visible hair motions are likely to occur. The AWT is coupled with a multiresolution geometry used to define the initial hair model. This combined system produces stable animations that exhibit the natural effects of clustering and mutual hair interaction. Our results show that the method is applicable to a wide variety of hair styles.

Images and movies

 

BibTex references

@InProceedings\{BKCN03,
  author       = "Bertails, Florence and Kim, Tae-Yong and Cani, Marie-Paule and Neumann, Ulrich",
  title        = "Adaptive Wisp Tree - a multiresolution control structure for simulating dynamic clustering in hair motion",
  booktitle    = "ACM-SIGGRAPH/EG Symposium on Computer Animation (SCA)",
  month        = "July",
  year         = "2003",
  keywords     = "hair animation, natural phenomena, physical model, multiresolution",
  url          = "http://www-evasion.imag.fr/Publications/2003/BKCN03"
}

Other publications in the database

» Florence Bertails : in lab LJK base , in team EVASION base
» Tae-Yong Kim : in lab LJK base , in team EVASION base
» Marie-Paule Cani : in lab LJK base , in team EVASION base
in lab LJK base , in team EVASION base