Kakegurui style LoRA
Making models can be expensive. Do you like what I do? Consider supporting me on Patreon š æļø or feel free to buy me a coffee ā
Should work fine at around weight 1 on AnyLoRA ckpt. Use kakegurui style to increase fidelity.
Remember this is a style LoRA. To make characters accurately you need other networks too (but it can get pretty close with accurate descriptions and character names).
How to use LoRA's in auto1111:
Update webui (use
git pulllike here or redownload it)Copy the file to
stable-diffusion-webui/models/loraSelect your LoRA like in this video
Make sure to change the weight (by default it's
:1which is usually too high)
Description
FAQ
Comments (9)
the glowing eyes/eye glint on the first image is amazing, can never get them to work or look good
How do you make loras of styles? Any tutorials?
Same as making character loras that are non "one word".
Just with a dataset that's one style only. Plus you have to train on descendants instead of precursors.
@LykonĀ what does it mean "descendants" instead of precursors? And thanks for replying!
@joule32Ā NAI is a precursor mode, AnyLoRA is a descendant model.
@joule32 you don't "have" to train on precursors, its just recommended by some peopleĀ
@AnnawnĀ yeah, even tho, if I had to guess, most just do it because they learned it from some guide and they find it difficult to adapt to descendant training.
No game no Life, Oda Nobunaga no Yabou, Madan no Ou Vanadis, Strike the Blood style please
Please make one for the style of Suzuhito Yasuda