V1 for manga and anime
V2 = V1-anime and some photoreal lora merged
for the photoreal nsfw
SD 1.5 的一個大問題是在跑NSFW的場景時常常出現一堆很不正常的特定部位
這個LORA是用來盡可能的修正的
One of the big problems with SD 1.5 is that when running NSFW scenes there are often a bunch of very unusual specific parts (P***Y) of the scene that this LORA is designed to fix as much as possible.
some example use lora blockweight and hires-fix
1:lbw=act is a blockweight of lora act=1,0,0,0,0,0,0.2,1,1,1,0,0,0,0,0,0,0
Description
FAQ
Comments (4)
Is this mostly intended for anime & cartoons?
yes, because there are trained by anime and mixed strokes of the LORA, will cause the photoreal type base model output of the shadow is too heavy, need to reduce the lora weight or the use of lora blockweight adjustment.
@oosayam Thanks for the photoreal version! I noticed in the downloaded filename, it has "asian" in the filename - will this still work for non-asian subjects, like european women for example? If not, would you consider adding a version that supports more ethnicities? Thank you!
@kurokuro
Asian face imafe training out, but in fact can also be used to produce other races of the results by the Prompt, if not , you can use my previous uploaded