Using character solver with Mixamo rig


Hi - I’m trying to work out how to use the new character solver with mocap data from Mixamo. Seems simple but there must be a step I’m missing. First - get a T-posed rig as the source, using the Mixamo site to ensure everything is named and ordered the same way (I’m going to be attaching my character to this rig).

Then download the mocap data as joints only:

So far so good, it all matches of course. Then I get two character definition tags and use the ‘extract skeleton’, both for the t-pose and the animated skeletons. 66 joints each, all match. Then set reference and create a solver on the t-pose, point it to the mocap data, and I get this weirdness…

From my understanding, the solver is meant to use object names to map joint rotations between skeletons, but it looks like it’s ignoring the intial offset and just adding the differences from the start frame onto the T-Pose. Any clues…? Thanks in advance! Chris


OK found the answer - the t-pose and mocap skeletons need to have the same start position, and as they’re the same exact hierarchy that can be forced by just offsetting the mocap data a few frames and copying over the keyframe 0 data from the T-Pose onto the mocap stuff. There’s a fair amount of foot slide here but that’s my error, the (actual) skeleton’s limb positions were only roughly matched to the joints.

Here there’s no weights/deformation - the character’s limbs are actually embedded in the rig ie the skull is a child of the ‘head’ joint, the leg bones are children of the leg joints etc - and the character definition setup is smart enough to go through and pick out the right bones to connect to the mocap ones. Grabbed a couple of mixamo moves, blend them as motion clips and it’s boogie time :wink:


Yes you need to put both rigs into T-pose. It’s great system. I wish we had some kind of feet fixation tool as well.

In contrast I find the Mixamo IK/FK Control Rig super fragile to set up.