LoRA based on the style of D.I.S.H.
Description
Recommended weight 0.8-1.0.
FAQ
Comments (2)
As an aside, training the lora while retaining the artist's imageboard tag in the dataset means that your lora is largely drawing on a lot of latent knowledge, not just the material you used for training.
I recommend either going tagless or using a tag different from the artist's name.
Cheers
Yeah, just kinda doing the old Pony ones the same way for Illustrious idk why, I should probably change it, but then I feel like I'd have to remake the pony ones as well... Have been changing the tags for the newer Pony ones ever since I realised that though. Thanks for the heads up.