So the program is
You must be registered to see the links
(It uses standard Stable Diffusion checkpoints)
Everything I do is local on my machine. In this case I even trained an AI module called a LORA
View attachment 2940128
Usually you will run a batch of images and use the best ones, here is an example of one that might be usable but if it was cleaned up or you let the AI repaint it.
View attachment 2940134
As we add more photo realistic details the AI "knows" (Has been trained on 100k images or more) humans don't have fur, or fox ears so it turns it into more of a cosplay.
View attachment 2940143
Not to say fantastical photo real elements aren't possible just harder to train.
View attachment 2940153