Need rigging help. On the left is MoCap FBX I purchased on TurboSquid. On the Right is my Robot I rigged using Mixamo. [I’ve used it with other data no problem]. How in Gods Green earth do I get them to work together. I just cant find the right tutorial and using the normal Character Definition tag to extract the skeleton ain’t working right. Be grateful for any leads or help. I’m halfway through a job and now I’m stuck! All the best, Brian
There is no real T pose for the Bip01 rig. Do you have any access to it?
The Character definition needs to be done manually. If you change the Include terms, you can save it as Preset then.
The rig Bip01 has three differences in the hierarchies. The Bip01 is below the pelves, the spine has a different amount of joints, and the clavicles are also not attached to the same spine. Those missing/extra spines’ rotation can’t be just ignored, as they are at different positions. This would be a request for an in-depth discussion tutorial in the Tutorial Request Forum. My short answer would be to skip the middle joint by shorting it and use the rotation than for the longer end of the spine. Example: https://www.amazon.com/clouddrive/share/BlFOE67LMo7mmOdltMzBsWJ0pWfoNfQXaN7pIz88qMQ
I leave the Dropbox open if you have a T-pose; I get an automatic email when a new file arrives, no actions are needed.
If you like to do more with the different rigs, since the robot looks soft with its current weighting, perhaps bind the robot with the T-Pose and have the rig matching from the start.
I have placed a tiny Joint below the Null Object and the first joint of the Mixamo Rig (no need to “weight” that at any time).
This allowed me to get most of the hierarchy “motion solved”.
Since you bought that file, I hesitate to share it here. But I think it is safe to share the changed Mixamo file with the transferred motion.
First off I’m just blown away by all your help! I have to do paying normal work this morning, editing all my renders from last night then I’ll go through all of this and get back to you on how Im getting on. I can’t thank you enough! Its for a short Im working on about hacking robots called botching.
So TurboSquid finally got back to me with a t-Pose at the top of the animation. If you get a chance to have a look its attached. All the best for now, Brian!
Thanks for the file, but… did you notice the differences between the FBX and the BVH file?
I’m not sure the author and I have the same idea about a T-Pose.
(…) edited
This is pretty much what I had set up on my own, a messy T-Pose, noting that the rigs will NOT match very well.
I made some more tests, and setting extra joints to zero (spine, neck) works so far. I added two joints on top to match the source rig, so all rotations will be taken. However, I haven’t gotten the Pelvis/hips gotten to work, meaning that the Mixamo is not allowing me to receive the given hierarchy. I think there must be an option, but I’m not seeing it for now.
I checked into a private user forum, but after I notice that all “heroes” have seen my question, but no one downloaded my example, I deleted it. Perhaps a sign that this is not the way it should be set up. I stressed it before, these rigs are not matching, and that is a big problem. To have not a clean T-pose is not helping either.
The BVH T-Pose has only 24 objects. The Mixamo has 68 items, each time including the Parent Null. Fingers and toes are not the only things missing; they do not count for the difference alone.
Fewer objects and a different structure will (!) lead to a motion that will not match. Even with some tricks, I was not able to get closer.
I used the afternoon to go carefully through the manual again. It states clearly that different hierarchies can be connected. It is not mentioned that they must have the same amount of elements or how to handle a different amount, as mentioned in earlier posts, that will lead to limitations. I assume that it seemed so clear to the authors that it was not even worth defining. Yet, I like to explore always options to work around limitations. As I see your request as a day-by-day problem. Please request a tutorial about it in the Tutorial suggestion Forum. Thanks.
For now, I will not dive deeper into it, as I believe that the time invested in making this fast option work takes longer than to constrain a different target rig to a source rig.
P.S.: A little tip, as I found that setting up both rigs lead to a mistake at one point (pulling the left joint instead of the right one).
An easy way to troubleshoot if the definitions are set up in a working way is to apply the data to itself (a copy). This will indicate only parts that don’t provide information at all. So, a rotation of a joint is not available. So one knows where to search first.
There are options to expand or reduce one joint, per sub-hierarchy, with some loss, but to start to reproduce the results of a joint-chain for a single joint with Constraints renders the idea of a quick transfer useless. It is thinkable if that is a day-by-day need.
As a side note, I found it a good practice to delete the Solver while doing any changes. Otherwise, I notice that the receiving rig might get parts of it set to scale zero. If the reference pose can’t be called up, it is a laborious fix.