Collada skinning

I hope you help me. My problem is with the collada skin equation:

v += {[(v * BSM) * IBMi * JMi] * JW} 
  • n : number of joints that affect vertex v
  • BSM : Binding Matrix
  • IBMi : inverse matrix of binding to the position of the joint i
  • JMi : JMi Matrix i
  • JW : The weight of the joint effect I'm on top of v

Each vertex "v" must be calculated (that is, through a "for" heap). But, is it not very, very slow if I have a grid of 10,000 vertices or more? Does it ever need to be counted in real time? There is no other way to count "v"?

Thank you very much.:-)

+4
source share
1 answer

You can probably use the threshold value JW - for each vertex v , you can skip further calculation for any connection i on v , where JWi below a certain threshold.

You could also IBMi * JMi for each joint once, right?

+3
source

Source: https://habr.com/ru/post/1438063/


All Articles