I made a Lora of her some time ago and was disappointed that the generated images were very blurry. After rewatching season 3 I noticed season 3 has much cleaner images so I decided to try to train a Lora on season 3 specifically.No outfit trigger words because of the high number of outfits in the dataset.

Files with SHA256: 866d126a24769cc27a900333b798516e8c6eda51ba31c71b6ad846a5b0f579b8