v2 version (04.05.2026): bigger dataset.

This LoRA replicates a style I liked. Some dudes on RisuAI did this by mixing a bunch of artists together (Imamura Ryou, Kezime, Wagashi (dagashiya), sparrowl and others) in NovelAI image generation. Looks and works great, except NovelAI's pricing and payment processing is abysmal garbage (they still don't accept crypto!), and I don't want to pay. So, here you go.

Trigger word is "stylemix", weight is whatever, the bigger - the better. I use anywhere from 0.9 to 1.5, I suggest you experiment and find what suits you best.