This was my first time ever training any LORA so please be patience with any updates. Recommend Strength: 0.9 (still works great at 1.0, but less flexible) I barely did any tag cleaning on the dataset, so it has pick up some issues that can make it unreliable for some compositions. I have tried to see if natural language was impacted since I didn't add any to the set, but it seems to work fine. This hopefully should serve as a "good enough" for a style that isn't baked into Anima.

Yelftea Style [Anima]

==First Training Run== 229 images, minimal cleaning

Files with SHA256: f04749192778c988214263d1525854977a85cb9fdb4b34293f2f127c9e095e73