This might be a long story, but I'll try to keep it short. I just got a motion capture kit called Perception Neuron. It's pretty good, but now I actually need to figure out how to make it work with Source Filmmaker and its very disproportionate rigs from TF2.
Perception Neuron is using its own rigs, with normal proportions.
So I hit the record button (I do not record fingers, only main bones), do my little tap dance and export it as 3ds max biped (*.bvh).
Then I open my latest 3ds Max, and follow all the steps from this guide
. And as you can see, the model in the end of the video is messed up, due to shoulders being not as wide, hands not as long etc. I tried all kinds of retargeting - it's wrong every time. Even if I move the rotation and position of the mesh so it matches the mocap pose as close as possible - it's just not right.
This is the example of what it looks like if I do the Automap thing. Not only that it looks like something out of Silence of the Lambs, but the main problem as you can see is the CATRig is smaller and limbs have very different proportions compared to the Heavy model.
So I wonder if there is a way of rescaling the CATRig so it matches the Heavy from TF2. Or maybe there's a way of applying mocap data directly on to the Heavy rig, not via mapping from CATRig.
At the end of the day, messing around with a bone positioning on a Heavy mesh all day, and manual CATRig rescaling (that doesn't even work properly) is not practical, because I plan on importing a lot of animations like this for all TF2 classes. Doing it for every mocap file would take more time than animating them manually from SFM.
I include the *.bvh motion capture file I created from Perception Neuron in case some of you guys want to try it. I do the T-pose sometime at the frame 700.
Any help is appreciated! Thanks!
Update: So using this video
I managed to figure out how to fix the height of the CATRig problem, so it doesn't look as freaky. But the lenght and proportions of the limbs is still an issue. So any tips on that are still appreciated.