I want to apply a motion clip that was created with a Mixamo human figure animation to a robot model. The robot is not a single mesh but a collection of parented subunits. The robot components were parented to a T Pose rig from Mixamo, so it’s the same hierarchy as the walking figure. The assumption is that joints are joints, and so long as the hierarchy is maintained, I should be able to mix and match the motion clips from different characters. But I’m missing something because when I apply the motion clip to the robot rig, it doesn’t move - everything is frozen in the T Pose. Is there a setting that will unfreeze the T Pose and apply the motion clip?
The post was moved to the Q&A, as the Tutorial Forum is only for question regarding a tutorial and should be always created with the Help button next to the tutorial. This helps to keep things organized and everyone can find the wanted content faster.
If it is not working, then my best guess would be, the information is not connected with the target. Since guessing is not really a professional way to solve cases, a scene file is needed.
You will see that the “Robot” rig is first matched in terms of orientation of each axis, then the clip is applied. I hope this will supports your target.
If not, please share the scene file in question, perhaps as dropbox file and send me a PM here if you wish to keep it private. I can also send an upload link. New and uncommon cloud services, please don’t, as the risk is too high, thanks for understanding. I will not use those anymore.
I think I have a error in fundamentals: I assumed that once a Mixamo rig was saved as a motion source, it could be applied to other Mixamo rigs of the same type (i.e. bipedal with fingers). But clearly I am missing something important.
I am in the process of making a robot character of separate parts, parented to a Mixamo T-Pose rig. I wanted to test the approach and discovered that the parameters of the T-Pose are retained even when a motion clip is applied. I followed your advice and zeroed the orientations of the components so they matched the rig (at least I think I did). But it still won’t walk.
This is a foundational step working towards a large project. I want to build the robot to optimize Mixamo’s motion capture resources. If I can’t do it with balls and spheres, then a more complex model is out of the question.
Edit: I just saw the post you have send a minute before mine, that is the one I got as PM /edit
Hi mford610,
Thanks for the file. The Motion System works based on hierarchy. If one places new objects into an existing hierarchy, the information flow is broken.
Two ways you can work here:
The first one is based on recreating the hierarchy with objects, whereby the axis of each object has to we where the joint is. It is a separate hierarchy and not mixed with the “original” Mixamo set up.
The second one uses the Mixamo, but works with Constrains to transfer the data, this enables the rig then to work more freely.
Yes, sorry. I wanted to open it up in the spirit of sharing.
I see that the fundamental error is in thinking I could parent objects to specific joints within a hierarchy - and so breaking the information flow. I have parented objects to Mixamo rigs before, but as I look back, these were always at the end of a chain - such as to a foot, hand or to the head. So the information flow was not affected. I am now trying to use the constraint tag to connect the individual robot components to specific joints, and it appears to be working very well.
Would Xpresso work as well using Global positioning?
Very nice of you, mford610 to share it. It certainly makes this thread more valuable with your scene file. Thanks a lot.
Exactly! On the end absolute end of any hierarchy “Branch”, if really nothing comes after it, that works, but is also really limited.
If you like to use XPresso for the rig transfer, The best way to do so is via iteration, so a quick set up and done. Please have a look at the screen capture and the resulting file: https://www.amazon.com/clouddrive/share/NtrAORHjBvkeId9E2FbTWj7F3rb3eJzFwb4fGJfJefL
(You can download this clip for a better viewing experience. I mention this as tutorials are not for downloads, but these clips are. )
P.S.: Please have a look here as well, just to discuss most options. https://www.cineversity.com/vidplaytut/siggraph_2015_rewind_bret_bays_creating_a_character_rigging_pipeline
@02:09
This creates a Preview Rig, which excludes the need to connect things with XPresso, but Joints, have a limited texture preview in the editor view, hence my caution as “Preview”. Other side effects are, it will not have all the options a real polygon object will have. So: Preview quality only! It doesn’t even render as instances in Cloner objects.
The copy/paste position is crucial. If you take just an object from anywhere, it will be pasted on that same position, but it can be selected and moved later.
If the Joints are not set properly to Polygon, it crashes C4D here. So: safe early, safe often. Auto-safe should be on of course. This shows, that I never reply just from memory, except I say “I guess” or “brainstorming”, things change, and answering from memory seems rather non-pro to me.
Ghost Rig, well, this is certainly not an official name, but that was the one I had in mind when I worked on it.
The target here was, that you can have a exact copy of the rig that worked with the joints. However, the Joint replacements with the Instance object allows for an quick supply of parts to that Ghost-Rig. To have more control about it, place each object or even group under a Null Object first, then place the Null into the field of the Instance. Since this is so far only a shortly tested idea, I hope I do not miss a limitation of it. Perhaps your target will show it.
The nice part is, at least I hope so, that you can even replace objects on the fly, as the Instance field to place objects in is animatable. To fill a group with dummy objects, select all Instance objects in questions and drag the dummy object into the field.
Yes, these are very helpful! I had not understood that a Cineversity forum would result in such quick and useful responses. But you’ve completely changed my perspective. I hope to make use of this resource more consistently instead of wandering around Youtube hoping for a clue.
You’re very welcome, mford610, thanks as well as for the nice feedback.
The real magic starts always by combining options, functions, etc., in C4D. Which means that it creates a nearly infinite number of possibilities. Each question can point to an existent creation or triggers an exploration toward new combinations. I’m happy to look into this, and share the findings here. Of course the more precisely I know what the target is, the more fun it is