Hey, guys! How are you, sorry for disturbing here...
I am currently using the skeleton data captured by a Kinect to animate a human mesh (which is also captured by moving the kinect around the model), but I have some problems in doing this. Now I am asking for help, since I am new here, so sorry for disturbing you all again.
I get the mesh and using the approach proposed in siggraph 2006 to automaticlly get the skeleton[namely "SkelA"] fiting to the mesh. And when someone is acting in front of the Kinect, I can get the motion skeleton data by using GetSkeletonCap[namely this skeleton "SkelB"].
Now I am transfering the motion of "SkelB" to "SkelA" to animate the mesh. Thus, I should compute every joint's position and orientation, my problem is how to get the relatively accurate orientation data according to "SkelB".
My current method is as follow:
Assume joints A, B which makes a bone, their original position is P1, P2, the current position is P1', P2', then I can compute the rotation between P1-P2 and P1' - P2', using this rotation data to compute the orientation. I know it's not sufficient because of the only 1 DOF, So I am asking for help here...
Thanks a lot here...