Daz Orientations causes faulty animations

[Maneki]

When I’ve Daz Orientations activated in the global settings, parts of the body don’t seem to be correct when I import a animation.
I imported a animation made for G9 on G9. I did screenshots of Daz, Blender and Blender without Daz Orientations. I don’t know if that’s a limitation or a bug but if it’s a limitation you maybe could explain to me what exact use cases Daz Orientation has.

‌

‌

[Alessandro Padovani]

As for the case use of daz orientation, it is to import daz poses and animations on prebended rigs. This needs both daz orientation and subtract rest pose. Unfortunately for some reason subtract rest pose doesn’t work fine with the standard orientation. But as you noted daz orientation has issues too so Thomas uses some heuristics to deal with that.

If we find a way to make the standard orientation work fine with subtract rest pose, then daz orientation is no more needed.

https://diffeomorphic.blogspot.com/2023/01/

https://bitbucket.org/Diffeomorphic/import_daz/issues/1445/

[Maneki]

Here’s a test I did with 8.1 now, it doesn’t seem to have the same issues with Daz Orientations active as G9:

‌

[Alessandro Padovani]

You may want to upload the G9 daz pose for Thomas to check. Otherwise it is unlikely that he can fix it just by looking at your pictures.

[Maneki]

I attached the frame I used as reference for these examples. What’s weird is that on G8.1 the same animation but optimized for 8.1 works well with Daz Orientations on but when I turn it off it has some really weird glitches inbetween where the character does a 180 flip occassionally on some frames.

Edit: Those glitches only appear apparently because I merged a eyelash rig with the main rig.

‌

[Alessandro Padovani]

As for glitches, merging rigs shouldn’t cause any. If there’s flipping in animation it means the rotation order and/or angle values are not the same so gimbal locks may arise. The standard orientation uses different values than daz studio so this is more likely to happen for FK animations. The daz orientation uses the same values as daz studio. In both cases using quaternions is likely to resolve the flipping.

Again, you may want to upload the flipping animation for us to check. So may be it can be useful to find bugs.

[Maneki]

@Alessandro Padovani How am I supposed to upload the flipping issue since it’s very specific. It’s in combination with a eyelash product which uses a separate rig.

[Alessandro Padovani]

As usual, you can upload a test scene as simple as possible in duf format. In this case also provide a link to the web page of the specific eyelash product, if you can’t reproduce the issue without it.

[Alessandro Padovani]

update. FIX FOR DAZ ORIENTATION

Commit 5986450.

@Maneki I can confirm that your test pose G9_pose.duf works fine with quaternions. This is not a fix but rather a workaround. Anyway consider that quaternions is practically mandatory for animation and you should always use it in blender. Unless you want a faithful FK daz rig for some reason.

Quaternions may also fix your problem with the flipping daz animation.

steps:

1. enable daz orientation and quaternions in global settings
2. test poses and animations

‌

note for Thomas. Since quaternions is mandatory for animation to avoid gimbal locks, and also resolves issues with daz orientation, we may want to move the quaternions option in the debugging section, default ON, meaning quaternions is required for daz/blender poses and animations to work, especially with IK rigs.

additional note. important. Personally I’d use quaternions also for hands, that are practically free joints even if more limited than upper arms and thighs.

[Thomas2]

The pose comes in right now, except for the hip which has rotation = -360 degrees. The arm bones are treated specially, but that basic reason for that is that the roll angles have be a certain way to work with mhx and rigify. The special cases are handled more consistently now.

Quaternions doesn’t resolve the issue with daz orientation, since for quaternions the plugin always uses the old way of matching poses in world space, irrespective of the daz orientation setting. Also, I think that jcms don’t work for quaternion bones, but that may be fixable.

‌

[Alessandro Padovani]

Nope, it seems that jcms work fine here with quaternions.

‌

[Alessandro Padovani]

Commit 8019137 works fine here, thank you Thomas for the fix. If Maneki has nothing to add we may close as resolved.

[Thomas Larsson]

You may forget about my comment about the hip. I imported the pose with and without daz orientation enabled at different frames, and then the character rotated at intermediate frames. But now I notice that the hip Y rotation in the duf file is actually 360.1906, so that is expected.