I like a lot of the results but it kept turning an img2img of a woman with the word woman in the prompts into a man. But results were good, just an unexpected quirk. Same base image and nearly the same prompt for all, difference being trying it without specified style or including it.