>>2774697I recommend compiling more coherent and complex UV maps (so you can do something like this:
https://video.twimg.com/ext_tw_video/1648304023781376001/pu/vid/1280x720/5r_v9gkntS_yxU9k.mp4?tag=12 ).
The primary MMD approach to skirts is to create a ring of an arbitrary number of vertical chains of bone-attached rigids that are connected with joints only within the chain. The "conventional" naming pattern—i.e. something you're more likely to see saved in motion data (for example, in motions with a sitting start where the creator tried to help skirt-wearing models)—would be [スカート_RowNumber_ColumnNumber pattern, with column numbers incrementing counter-clockwise relative to the model's orientation and 0 corresponding to the front (0 degrees). It's what the PMXE skirt plugin uses. Additionally, you can start with a static rigid to be able to move a part of the skirt even with physics on.
Now, to prevent clipping, people ensure there are no gaps in the skirt rigids and make thigh rigids fatter & match them to the thigh's full length plus knee (which your model doesn't have). The depth or radius of skirt rigids can also be increased, with potentially negative consequences for the skirt's behavior if there's little movement room.
I wondered about a physically connected rigid mesh here before. If I'm not mistaken, you've answered that thought: there are no real advantages.
Regarding double-sided materials: it depends. They have some big cons, such as not working well with reflections and some shaders. But you can use them for shadow play too.
P.S. BOOBA.
P.P.S. Btw, regarding arm IK chains, I didn't realize it when we were talking about it, but the actual non-retarded method of elbow control was simply having the IK arm (腕) bones visible/saveable and rotating them. Sorry. The ASK model above shows a way to make it convenient too, but I believe an IK chain should target the palm of the hand to properly keep the grip on elongated things.
