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: 0dc0f617da627ea6965741cb484f35ec20781a62dfd510ec47584991edbb38ea