Download

Which skeleton is the most accurate and easy to retarget with different animations?

Which of the currently available skeletons/rigs would be the best and most future-proof starting point worth of investing time and possibly even money?

The long story.

I have seen different posts here about people having issues with some animations, and some answers are that the skeleton of their choice is not the best option for specific kinds of animations. I have heard critics about UE mannequin, Blender’s Rigify and other skeletons for various reasons (issues with twist bones, fingers, root motion etc.). I have also looked at AllrightRig and it looks awesome but it’s still beta and I’m not sure if it would be free of all the issues of the other skeletons/rigs I mentioned.

From UE4 docs, I found this to be really helpful starting point:

This approach fixes most of the issues, but not all of them - some complex animations might still look not right and cannot be fixed because of the skeleton design.

At some point I might want to try out all the latest (UE 4.14) features (IKs, montages, retargeting, blending, Pose Drivers, Virtual Bones) and have a future-proof and UE compatible skeleton which can be used with freely available mocap animations (the famous Carnegie-Mellon University BVHs, which have been converted to FBX by cMonkeys).

Besides all the usual walk/run/idle animations I will need good finger animations and facial animations. For faces, most probably will go for morphs; bone animations are too cumbersome to get right. But I might also look into Cluster Shapes (if I find some way to do it with free software tools). My project is more a simulation than a game, therefore it is important to have decent quality of animations because of lots of close-up views, but I’m not aiming for AAA or something like that - it’s just a one-man hobby project.

While I’m ready to tweak some animations manually, I don’t want to start with something that later on turns out to be inappropriate for too many animations - I don’t want to re-rig everything from scratch.

I’m ready to pay a reasonable one-time price (let’s say, under 50 USD), but it is important to be assured that the rig will be maintained for the future UE versions.

BTW, I tried to add a poll to the thread, but always received the following error:
be63806d63f5ec20d0d93965803a5fff1c71746b.png

So I added the poll as a Google docs form here:

I suggest you stick with the default mannequin. Any other skeleton will require you to invest quite some resources into maintenance and development.
The UE4 default skel has already a ridiculous amount of time spent in development for games. Animations from marketplace are 100% compatible. And 3rd party stuff could easily be retargeted.

On the other hand tho, you mentioned you are doing more of a simulation with lots of close-ups (whatever that means), facials and detailed finger animations. Not quite sure to what level of detail you are aiming to get, but yep maybe the simple ue4 skeleton might not be enough. However any other skeleton you mentioned would be tough too and you might need to use a lot of corrective morph targets to make things look right using so little bones.
Im currently also working on a project that is a sort of simulation (interactive adult vr) and we are aiming for AAA quality and interactivity. This means we need a very sophisticated skeleton/rig that allows our characters to reach a wide variety of extreme poses while still looking anatomically accurate. Im in charge of the rigging and parts of the animation system design for the project so im developing a custom skeleton system that can allow us to reach all these targets. Detailed skin interaction, muscle action, fast facial pipeline from 3d scanned people, anatomical deformation, etc. etc. So if you are aiming for something similar, Id suggest you spend resources on a custom development, else you would have to stick to the default skeleton and try making the most of it - still a legit option, especially since you mentioned you are not aiming for AAA.

PS: can you send a link of the fbx conversion from cMonkeys? Did a 10 second search and didnt see it.

Of the selection listed I checked Genesis3 by Daz3d.

Here is why.

Overall though the complaints relate more to the inability to properly edit animations as to requirements at the source level. Trying to animate in UE4 would be like trying to paint a picture in UE4. Sure it can be done but not the most ideal tool for the job.

Thanks for the responses. I guess I’ll stick to UE4 then. I’m only afraid that I might hit some dead end at some point, but I hope that it will not be too complicated to modify UE4 skeleton and not mess it up. But I have watched a UE4 stream video where they are using Blender to work with UE4 skeleton, and the workflow looked pretty good. It’s great to see that UE4 team is paying attention also to folks who do not have options to use UE4 Maya ART tools.

About the cMonkey animations - I found them in Unity forums:
https://forum.unity3d.com/threads/huge-free-fbx-mocap-library-released.261401/

It seems you have to install Unity first to access their asset store. They are also available in some mega.nz share (mentioned in the Unity thread), but when I tried to download them from Mega, it reported that it will zip them for 23 days and then timed out after 10 minutes; so I guess the best way is to use Unity Asset Store.

I just hope that the Unity skeleton (which most probably was used as a target in those FBX animations) does not differ too much from the UE4 skeleton - in which case those FBX will need major work, anyway.

About animating directly in Unreal - I have high hopes for Allright Rig: https://forums.unrealengine.com/showthread.php?116258-Allright-Rig-For-Unreal-Engine-4 , it is great.

So I use a combination of both. I currently do custom skeletons for really important characters and animate myself, but as one-man team, the marketplace resources save a lot of time for NPC’s and the such. I haven’t had much luck retargetting with different skeletons with acceptable results, and the time it takes to do so didn’t seem to be worth the effort…