Adding bones to a MetaHuman skeletal mesh breaks the skin clusters of the lower LODs.

Development Character & Animation
, ,
1

On the MetaHumans’ body skeletal meshes the LOD1+ have their skin cluster broken. (Seems to flip the _L and _R bone skinning)

This cannot be repro’d with Manny’s skeletal mesh.

The repro is not 100%, closer to 75%, once in a while the MetaHuman does update properly.

[Image Removed]

[Attachment Removed]

2

Steps to Reproduce

  1. Open a 5.7 or 5.8 MetaHuman project which has the Core Metahuman installed to the engine.
  2. Create and Export or find a pre-exported MetaHuman.
  3. Either duplicate or edit the body skeletal mesh directly.
  4. Add a bone under the root bone.
  5. Save the skeletal mesh. (Or save all)
  6. Drop the edited skeletal mesh in a map.
  7. Assign any animation to the skeletal mesh. (Issue not visible in A-Pose)
  8. Back the camera away or force to LOD1+.

[Attachment Removed]

3

Hi, yeah, the issue here appears to be that the skinning data is modified when the new bone is added, even though that bone should be unskinned by default. With a non-metahuman mesh, the solution here would be to regenerate the mesh LODs after adding the new bone. But that isn’t a valid workflow with a MH mesh as it breaks the DNA. I need to speak to the team about this to see what the options are, I’ll follow up once I’ve done that.

It’d also be useful to get more info about the use-case here for adding the new bone? Would this bone ever be skinned or will it always be unskinned?

[Attachment Removed]

4

Thank you for the answer, that does make a lot of sense.

In our case, the bones added were unskinned and would remain unskinned. (The IK helpers and item attachment bones.)

[Attachment Removed]

5

Hi, I’ve spoken to the team and they confirmed that this kind of operation wouldn’t be supported on a MH body mesh from within the editor since it invalidates the DNA as I mentioned previously. The only currently supported workflow is to export from the editor and then use the MetaHuman for Maya plugin to make modifications. The Maya plugin also contains the DNACalib2 library, which you can use to make lower-level modifications than you can via the UI. We haven’t documented that API yet unfortunately, but it is similar to the original DNACalib library, which is documented here.

Having said all of that, I don’t see any support for adding joints rather than just removing them so let me confirm with the team whether adding this kind of helper bone would be possible or not.

[Attachment Removed]

6

We don’t need these bones to be driven by the DNA. These are mostly bones for the IK Retargeter which don’t come natively with MetaHuman. (Similar to what Manny skel has)

That being said, yes we could use a workaround of adding these bones in Maya and exporting the skeletal mesh from there. Would be faster to not do the DCC roundabout to add IK retargeter required bones, however the limitations are understood.

[Attachment Removed]

7

Hi, sorry for the delay following up, we had a long weekend here so I’m in the process of catching up.

I spoke to the dev team about this and adding these IK helper bones should be supported, but it has to be done via editing the DNA. The DNA is tightly coupled to the mesh topology, joint hierarchy, etc, so any changes to those have to go via the DNA for it to remain valid for the updated mesh. There’s no API for this in DNACalib so you would have to go to a lower level and manipulate the DNA directly. But it should be possible via the API that is available through MH for Maya.

If you want to go in this direction and would like a sample script just let me know and I’ll try to put something together.

[Attachment Removed]

8

We haven’t had any issues adding the bones to the skeletal mesh only. Rig logic still seems to be behaving properly. That being said we have already moved to adding these via DCC from now on to avoid breaking the LOD skinning.

Alternatively, if we add these helper IK bones in a pose editor (MH for Maya) skeleton hierarchy and export the DNA out, the DNA will now include the bones. We haven’t thoroughly tested the stability of this however as we have not seen any issues with keeping the bones at a skeletal mesh only level.

[Attachment Removed]

9

Ok coming back here, we’ve finally hit some issue with the bones being added in the skeletal mesh only.

It is breaking in the IK Retargeter’s additive poses.

We’ve opted for a ‘parent skeleton’ that holds these bones and the body follows it as a quick fix.

However… now we’re realizing that we can no longer preview the ik/attachment bones in the animation sequence editor.

I don’t trust my pose editor method enough, so a sample script of the DNACalib would be appreciated.

Thank you very much for the time and effort so far.

[Attachment Removed]

10

Hi, sorry for the delay getting back to you after your message from last week. I’ve been discussing the expectations further with the dev team, and there’s been some back and forth there that has slowed things down. It was a misunderstanding on my part in terms of what would, or wouldn’t, invalidate the DNA. In the original setup that you mentioned, regenerating the mesh LODs in the editor (as I’d described as a workaround with the initial skinning issue) would invalidate the DNA. But if you add the unskinned bone to the mesh directly in the DCC and reimport, that should be fine. As should be adding it via Pose Editor, and then exporting/importing the DNA back into the editor, as you’ve been doing. Sorry for any confusion there.

So I expect that the issue you’re running into with the Additive Pose op in the IK Retargeter is unrelated to any changes to the DNA data. The retargeter just deals with transform data for the bones that are being operated on, so assuming that the IK bone is positioned correctly in the input pose being fed to the retargeter, it’s likely something is going wrong with that op. We have had a number of issues with the Additive Pose op in the past.

Can you give me more specifics about what’s going wrong with the retargeter’s output? Anything else that you could share like screenshots, etc, would be helpful.

[Attachment Removed]

11

Hello Euan,

Firstly, I would like to apologize for the very delayed response. I must have missed the original email that notified me of your response.

Glad to hear that the bones should be safe to add via maya. We try that again later, however there’s still the concern of the additive pose breaking.

The issue we were having when pushing an additive pose on the retargeter to go from Metahuman Performance Skeleton to the MetaHuman skeleton our MetaHuman skeletal mesh body mesh with the extra bones was that the upper arm or lower arm rbfs would break. (upperarm rbfs no longer attached to upperarm, etc) When trying this on a out of the box metahuman without those extra unskinned bones, it worked well. When removing those extra unskinned bones it had no more issues.

[Image Removed]

[Attachment Removed]

12

Hi, no problem on the delay, the notifications are easy to miss. I’m going to spend some time tomorrow looking at this to see if I can get a repro and then work out what’s going wrong. Is there any chance that you could share the assets? That would likely make the process much quicker.

[Attachment Removed]

13

Hi, sorry for the delay following up again. I needed to clear some time to try and repro this. I’ve gone through the process of adding extra bones to the skeleton in a setup similar to yours, but unfortunately, I haven’t had any luck seeing the same broken behaviour. So I have a few questions on your setup that might help narrow down the issue, assuming you’re not able to share the assets.

  • Are you using the newer MH setup where the correctives are applied within rbfs are applied via rig logic?
  • Do you have any retarget chains set up on the IK bones? It looks like you do based on the warning in the log, but I wanted to double-check
  • Are you using the pin bone op to fix up the IK bones’ locations?
  • Are any other bones constrained to the IK bones in some way?
  • Can you share what your post process anim bp looks like for the mesh?

It’d also be interesting to know if you go through and disable the other retarget ops one by one, whether you see a change in behaviour with the broken transforms on the upper arms? That would narrow it down to an issue with a specific retarget op at least.

Thanks,

Euan

[Attachment Removed]

14

Hello, apologies for the delayed response, we were trying out the adding of the bones to the DNA using Pose Editor in Metahuman for Maya and so far it’s working really well.

This allows us to preview the deformers and extra bones in animation sequence editor using the collections and we also created a combined mesh with a new DNA based off the one we’re using for the game to let us preview these on IK retargeting.

If you’d still like to have a vanilla repro of the original issue we can find some time to try and get this repro, however we’re currently unblocked by using the Pose Editor in Metahuman for Maya.

Thank you for the time and effort so far. Let us know if you’d like further information or if we can consider this ticket closed.

Tania

[Attachment Removed]

15

Hi Tania, no problem, and it’s good to hear that using the Pose Editor setup is working for you. Since it sounds related to the specifics of your setup, I think we’re probably good to close out this ticket. But I’ll keep an eye out for anything similar to this happening again with other licensees. And if anything comes up again with this on your end, you’ll have a few weeks when you’ll be able to reopen the ticket to discuss things further.

Thanks,

Euan

[Attachment Removed]