Just want to say thanks...
Thanks for a incredible work, best anime sdxl public model right now! =)
I just want to second that. I don't want to diminish others' efforts in this space, as really a lot of us are just too computationally-constrained to train XL nearly as well as we can train 1.x or 2.x, but the main anime model I've seen recommended by others is an ungodly lora merge-pile. I hope this project sees nothing but success, and that others in the space create their own uniquely interesting models so we have both a good selection of quality finetunes and many possibly useful merges to come from those.
I want to third that. This is the best anime model I have ever used. I didn't even need lora so far.
A milestone for the community. Good job!
Thank you for the warm message, I'm glad the model reached everyone.
It makes us motivated to train more models.
Ahem, Senpai, I Much agree with the rest of them - and i'm aware my anime SDXL ones are probably the ones they call LORA SQUISH TRASH :P
But knowing how long SDXL takes to train, and the KNOWLEDGE and COST it takes to train, it's an absolute godsend someone has taken the time to make a stable enough SDXL model :)
V1 was a great model, V3 is gonna be even better <3