Armature bone limitation on import

[Ada Radius]

The SL/opensimulator armature currently has 159 bones, but viewers limit mesh imports to 110 bones - the number of bones a mesh attachment can be rigged to. If that number is exceeded, “Include skin weight” in Upload Options is greyed out.

The pre-Bento default Fitted “BaseMale” and “BaseFemale” avatars, available at http://wiki.secondlife.com/wiki/Mesh/Rigging_Fitted_Mesh in a variety of formats, are in one piece, with separate objects for the eyes - normal for game rigging - and they replicated the inworld default system avatar in 2014.

Since ‘Bento’, SL has not released a replica of the inworld avatar - same number of vertices, same rigging. They issued Maya-based dae files at http://wiki.secondlife.com/wiki/Project_Bento_Testing in 2016 - the body in three pieces and not rigged to all the human mBones nor any of the collision “fitted” bones, such that Appearance sliders are limited. The SL Wiki labels these files as “broken”, no exaggeration, though I’ve been able to repair them so that they’re usable, if limited, in Blender. (by combining the dae data with avatar_skeleton.xml data from the viewer Character folder)

The post-‘Bento’ workaround so far has been to use that divided mesh, which creates seams and increases the vertex count. If they’re rejoined, the vertex count and arrangement is identical to the pre-Bento BaseMale and BaseFemale, only taller and with longer arms. The inworld system avatar is clearly in one piece and is attached to a rig that has all 159 bones. This can be viewed by wearing a black skin and in the Dev menu, Avatar > “Show Bones” and Rendering > Wireframe.

I’m not a programmer. I make rigged costumes and avatars in Blender, Marvelous Designer, etc. and volunteer with the RuthAndRoth project when I have time. I have been working, off and on for the last few months, on making a replica of the current inworld default system avatars in Blender 2.8x, so we can use them for testing. The two major roadblocks are converting the Maya-based files available from SLWiki to Blender’s armature system for the collision bones (they bork on export/import), and the viewer import limitation (grey out import).

I’ll keep working on rigging and re-weighting the avatar in Blender - digging information out of Blender and SL dev discussions for open source information that, my opinion, should have been made available years ago. I was hoping someone on your team would take an interest in increasing the import limitation from 110 bones to 159 bones. Or removing the limitation altogether. I don’t know what the programming issues are.

Thank you! Please let me know what files you might need from me. I’ve been blogging workflows at my blog at https://adaradius.com and expect to have more up today or tomorrow. Cross fingers.

‌

[Lisa Laxton]

Thanks for your submission! I reclassified this as an enhancement. Did you attach an IMA White Paper?

[Seth Nygard]

It is my understanding that by design there is a limitation of 110 joints for any single mesh object. This would be where viewer imposed limitation on importing comes from. Multiple meshes would need to be used to allow higher numbers to be imported. This however can then result in the visible seams issue due to the handling of vertices and normals.

[Ada Radius]

I remember the notice https://community.secondlife.com/knowledgebase/english/enhanced-skeleton-project-bento-r773/, but Jeremy’s explanation doesn’t hold water. The Bento test avatar they issued (and have not updated since) does not include weighting to collision bones nor mFoot and mToe, so the Appearance sliders for butt, boobs, muscles and fat don’t work. A nice rig for high heels or mesh feet doesn’t work either, leading to all the designers who have to make them with a lumpy seam at the ankle. The inworld system avatar does not have seams and is weighted to collision bones as well as mBones - as far as I can tell it’s using 124 (female) or 125 (male) bones, as it doesn’t use the wings, tail and hindlimb bones. So…why the 110 bone limitation on import?

[Lisa Laxton]

Thanks for your input on this! We need to determine what is feasible as a team. I am putting it on hold until a working paper (IMA White Paper) is submitted. We are looking for developers who would like to assist evaluating this as a new feature.