This LoRa takes Z-Image model to next level.
shaper and higher quality images.
Masterpiece artworks: Z-image is weak in many styles. This lora gives you higher level styles like illustration, surreal, abstract.
Weights: with 0.5, 0.8, 1, you will get different art style paintings.
humans: more Caucasian look.
Best Settings:
steps = 10-15
CFG = 1.5
weight = 0.8 (0.5-1)
sampling = (euler + ddim_uniform) or (heunpp2 + beta)
Description
FAQ
Comments (4)
Without the lora looks a lot better. The characters look natural,t he lighting is more natural, the skin has actual human skin texture and doesn't look like a photoshop job instagram filter. Was this trained using AI images?
I don't understand. When you say, 'Without the lora looks a lot better', do you mean don't use this lora? That images look better when you don't use this lora?
Well, seems like this lora isn't for you then, because it clearly states it's for different art styles which means probably less photorealistic.
Does what is says -> takes Z-Image model to next level :)