SDXL LoRA based on the wonderfully monstrous art of Neurodyne!
This is my first attempt at training a LoRA externally with Civit, as I've never gotten a trainer to work locally on my mid-tier hardware (my PC already struggles just as a local generator). Unfortunately, I don't have much Buzz to experiment with, so it may be a while before this gets updated or something different is published.
Version 2 is now available after months of accruing Buzz. I finally had enough to train this new version on a larger dataset with higher parameters & a custom checkpoint!
Description
Version 2: Uses a dataset almost twice the size of V1 (120 images), trained to 15 Epochs at 1024px resolution on the Juggernaut XL Ragnarok Checkpoint (see suggested resources below).