Export: write custom attribute arrays for non-skinning vgroups #1515
Conversation
We can't remove it, it will lead to miss cache in some situation |
| #vertex_groups = None | ||
| modifiers = None | ||
| else: | ||
| # Check if there is an armature modidier | ||
| if len([mod for mod in blender_object.modifiers if mod.type == "ARMATURE"]) == 0: | ||
| vertex_groups = None # Not needed if no armature, avoid a cache miss | ||
| #vertex_groups = None # Not needed if no armature, avoid a cache miss |
There was a problem hiding this comment.
This can lead to miss cache
|
Ok, cool, so this PR should wait until your work there on custom caches is done? If I understand #1245 is the ticket to watch. |
|
It does not require a custom cache. It will also not cause a cache miss: if there are non-skinning vgroups, this branch always exports them, so since the vgroups are always used, it is correct that they are always part of the cache key. But I don't think it is good that they are always exported. If it's desirable to merge these features into mainline, I think they should be behind an export option, cf #1232 (comment). |
|
BTW, some other custom attribute can be exported, as requested by some blender devs. See #1505 Anyway, now that vertex groups are stored on mesh, no more on object, all this code needs (probably) to be refactored. |
Oh, that's right. |
|
Plus one to the idea of adding this feature to the official exporter (as an export option)! |
@julienduroure Are there any plans to do so? |
I've combined the facemap exports and vertex group exports from @scurest's work last year, and tested that this works on 3.0.0.
I haven't followed all of the changes and improvements since Nov 2020, however, so I'm not sure if this is still the ideal way to do things.
There are also a couple of places where a line of code has been commented out (e.g. gltf2_blender_gather_nodes.py has
#vertex_groups = Nonein a couple of places). We should probably decide what we want to happen--either remove it, or improve it?#1232