I would prefer if you guys don't use this model for Commercial Use.
I don't recommend using token merging (tome). The result get worse. If you want to use token merging than try to lower the effect of it in settings.
the base model I trained this lora on was anything 4.5.
I used screenshots from the anime and fanarts to train the lora.
check out my pixiv: https://www.pixiv.net/en/users/34413372
v1.1 require good prompts and you will struggle to get meredy original clothes, but this lora is more flexible and you can generate more stuff with it compared with v1.0.
v1.0 doesn't need long prompt if you want to get the character clothes compared with v1.1, but it is more restricted in what it can create. I recommend 0.6-0.8 value on the lora model (anime style).
I use xformers so the result will look different if you don't use xformers.
use these keywords. It will help you with creating the character: red dress, dress, yellow ribbon, ribbon, ears muffs, headphones, pony tails, pink hair, aqua eyes or pink eyes, detached sleeves.