Hi Folks,

My first attempts at AI rendering. I'm using sexy.ai with Realistic Vision 5.1. prompts like
"beautiful woman, gorgeous face, cute, pinup style, dayglow rubber latex jacket, bbw, nsfw, big ginger hair, cheeky smile, lying face down in deep mud, fun, long gloves, cleavage"

I am beginning to learn the limitations of the model so have started using EasyDiffusion (GUI for local Stable Diffusion.) with the same model since EasyDiffusion has more features. I'm at the stage where I'm trying to understand LoRA and how to train using my own images to get more mud coverage.

Anyone has any suggestions I'd be grateful and yes, I do have a type

if you can use the same model you could do it in stages of being clean, kinda dirty then caked in mud. or does it change the outfit every time ?

Hi - it depends on the AI program and some settings. If you use sexy.ai and put the prompt words in, it has a random seed, so different every time. If you use EasyDiffusion it allows you to use the same seed each time, so the same prompt words would generate the same image I think. You could then add more prompt words like mud etc but I think the image would be exactly the same, but just getting dirtier in stages, so it wouldn't really look right. I haven't learnt enough to be any good at this but I think you could use what is known as a control frame to move the models limbs each time. I'll have to experiment.