Something something 15% or more. This LoRA came out surprisingly well and was easy to train on captionless images. Trained on BB95 and works well with a lot of models and style LoRAs.
Training data here https://mega.nz/folder/RawywbBS#IqHlgaZmFEhj7WVS7EsnAw
Trigger with "martingecko"
Description
Initial release