A cinching before and after concept LoRa, trained on a very small dataset.
I recommend weight of 0.4 - 0.7, anything more will muddy your results if youre using multiple LoRa's.
This was trained on Pony so I do not know its performance on other SDXL based models.
buy me a coffee! help support what i love to do
https://ko-fi.com/diffusionarchive