This is my second attempt on Agawa Ryo's style. I like how it worked out.
Hope you guys like it too.
All the images were generated using the LORA at 0.55 - 0.60 weight
Trained over AbyssOrangeMix 2 Hard
320 x 320 Highres x2 Latent DMP++ 2M Karras at around 0.6 denoise
Special thanks to @Jeager122 for the dataset and idea :D
Description
(dont ask for V1, RIP)
FAQ
Comments (4)
Thank you so much, this will be fun to test out. :)
Hey.
I still cant understand the lora method . Do i have to pick my own images and train them with this model or i can just drop this to lora folder then select it as a checkpoint merge and generate images ?
you use a LORA as you would use an embedding, you drop it into the models/Lora folder and then select it from the "extra networks" window inside your webUI. It should bring up the style in pretty much any images you generate with it enabled
@ROOD Now i get it and it works. Ty for feedback