Posts
maybe this was a too hard task, or i need classes on prompt engineering, but only 1 UNDERRATED model was able to deliver the image I wanted (read prompt,) Even base model couldn't give me that. Then I tried with img2img, some models barely changed anything (base once again.) other did some improvement, but imo it was add Ciro's LoRA that finally aided me to get one step closer (I know complex stuff we rarely get on our 1st tries... especially without manual edit and stuff).
But in this case I WILL NOT BLAME ANY MODEL, but the loras. Loads of loras people use just because, for many (not all, lets be fair) don't do any friggling thing! Even if I wanted the usual girl, I tried many popular ones, they end up ruining my pics and if I hadd to set them very low, they occupy a slot, hence rather remove them. Where are the brilliant LoRA from 1.5? Is it so hard to create loras for xl? If not, I feel I becoming tempted to do something not caring if folks heart it or not.So far XL is a mere 2.1
