>>92686736It should just work, e.g. this image prompt is "shishiro botan, hoshimachi suisei, scenery"
Make sure you use an Illustrious based model and put the character names toward the beginning of the prompt. You will still occasionally get a third person, or an extra head or something, but it's miles better than it used to be.
There's no great way to make each character do or wear something specific, though. You can try putting tags near the right name and using BREAK between character descriptions, but it's a crapshoot.
