Deltarune XL + 1.5 LoRA

SDXL:
Trained on Animagine and Pony models. This model is for 14 characters from the game Deltarune + their variations:
Kris (both light world and dark world), Noelle (both light world and dark world), Susie (both light world and dark world), Jevil, Ralsei (chapter 1 and chapter 2 appearances), Berdly (both light world and dark world), Spade King, Lancer, Cyber Queen, Rouxls, Spamton (+ Neo form), Swatchling, Tasque Manager, Rudolph

It is recommended that trigger words be placed at the beginning of the prompt. Unless it is Pony, which was trained with shuffling.

In legacy models, they aren't really capable of interacting with each other, unless they are sufficiently similar. And stylistically they can influence the output, even if not by much.

I tried to make the prompting as simple as possible. Even so, additional details do help, like for some reason lightkris doesn't have his type of sweater a lot of times unless you prompt it. Be careful with "masterpiece, best quality" of animagine models, it might make it more sexual than needed or change gender.

About rework: please read the "about this version" of the version.

Pony:

At the time, it is a bit undertrained, but good enough for most characters.

SD1.5:
Feels a bit worse than XL in outputs, obviously, but it did learn the characters. I don't really have a lot of 1.5 models to compare the generations with, though, I think it just requires more prompting than SDXL.