final result <-Original prompt These took like 2 to 3 hours each. I'm getting the process down. I generate images, try to get the prompt down to something that's difficult, like here, I couldn't get the stable diffusion to render a woman shuffling cards, so I just had her playing on a controller, then I traced some hands, did a shitty job shading it, then had diffusion paint over my shading work. So far, I haven't trained my own Loras yet, which is where you train the AI with the images you provide. So I could have definitely trained the stable diffusion what shuffling cards looks like, or hell, even how to draw my faces so I don't have to do it every time. I think for the shuffling cards instance, I think I should have probably made a lora for it. Going forward, there are a lot of "shuffling" type pictures, since it is a Sexy Casino game, after all. It really is case by case basis. Standing poses are really easy for stable diffusion. Anything else