Until you implemented full control over material merging in the export pipeline can you please hard code some more logical settings for instaLOD, using UV coverage, texture size or just prioritize body and face textures over tongue and teeth etc. to get a more logical texel density distribution.

As you can see on the image here the teeth and tongue and other "stuff" takes up a lot of space in texture and the face that is most important in many cases takes up minimal space, this is not optimal...

In the future as I already described in detail before, we need full control over this, which material to merge, image size per merge, naming etc.

But for now, can we in some way make this stuff usable?