This is a DAZ3D render (the background is a free image from Pixabay) + Stable Diffusion's ImgtoImg. I purposely chose a pose with rather complicated hand position to see what Stable Diffusion would do with it. And to my surprise, there was almost no needs to fix the hands. I really like what it did to her pants as well. The face turned out really nice and needed no postwork. No dead eyes, no blank expression! And I also love how it interpreted the hair. I have other generation based on the same render with different seeds, which gave different girl each time. Some needed a little bit more work, but overall, this is so much closer to getting what I want from the first time. The second image is a straightforward comparison of my Jeannie image's 3D render and SD postwork. Low denoising settings allow you to keep the original posing and clothing, and it is also uses less GPU power.