We decided to try the “hollywood” method to retarget. I mean not to build the skeleton for the character, but to film mocap animation, to build the skeleton and then to fit the character for the skeleton…

So we start the character “stage” and here I understand that the retargeting in wrong! I took the skeleton out of MoBu of the mocap data’s actor with the accurate pivot points. And guess what? Mocap animation isn’t fitting it’s own skeleton. The tick “match source” doesn’t help in this case. I have already written my thoughts on the issue of wrong retargeting in the MoBu, but I couldn’t assume that the animation wouldn’t fit it’s own skeleton. It seems that I will have to find another solution, as it usually is in every project.

Thinking logically, It should be right to transform bones without any additional offsets and layers, because the skeleton and an mocap data actor is identical. The only thing which can dispel the idea to think that the task is simple is that the skeleton should be built symmetrical. In MoBu an actor for the optical data is symmetrical, so theoretically there should not be any problems.

Does anyone have an experience of such a reverse approach ?

MoCap animation and the character, reverse approach
Tagged on:         

Leave a Reply

Your email address will not be published. Required fields are marked *