I have a similar problem to this topic :
I followed the skinning formula from COLLADA 1.4.1 document and used it for a simple animated cylinder.
My problem is that all rotations of joints are applied from the origin of the world instead of the origin of each bone. So I think that I have a problem of WorldSpace/LocalSpace coordinates but I cannot figure out where since I followed the formula and my matrices seem correct.
I followed the waZim’s tutorial : http://www.wazim.com/Collada_Tutorial_2.htm#Second_Part so I did (vBSM) and (IBMiJMi) then (vBSM)(IBMi*JMi) (but I don’t think that’s important)
If someone has any idea where I did a mistake it would really help me.
I precise that my .dae is correct since the ASSIMP viewer can load the animation correctly.
Thank you for your time.