Trained on a small dataset but performed better than other version trained on larger datasets. No other additional LoRAs used when generating these images.

Used strength ranging from .85 - 1 for the best results.