when you give enough tries there is just one that works :^( I don't even know if I can recreate the same one again :')Train on many different overwatch CGI images, many of which I used to train 1.5 LoraAble to make very overwatch-style faces.Still a test, this is my first time making an sdxl Lora, So I don't think this is good enough yet :-?Feel free to give me some feedback here or in Discord, I will keep improving this LoraAll train images are on 10241024using settings other than 10241024 may give a funny resulthighly recommend using ADetailerPost images used high res fix + AdetailerTrained mostly for the face, Style is adding to the model, but somehow didn't change the background enough which makes it kinda look like an oil painting (I think this is even better :3 )Spend all that time waiting for SD3, it just takes too long so I gave sdxl a try :)Share some posts :)Join Discord to chat :D
