Going to bump this up as we've gotten some more resources for AI image generation since then!
I've been messing around with Stable Diffusion and figured I'd take this opportunity to share something. I've been using a combination of input images, inpainting and traditional editing to get Game Over/Peril-esque scenes. Haven't spent too much time with multiple characters interacting with each other yet but I hope to at some point, as I have no real direction over what I want to do with this tool.
Original scene for reference was from Captain Planet
I basically took the original image, added in shoddy pink slime/blobs over Captain Planet and the logo and let AI take liberties with the background. I then added the character, retextured the skin so it matched the lighting relative to the rest of the image and then added in the trademark costume, hair color, and slime over some of the body parts.(think crude paint style textures) Then let the AI run some more over the character and the rest of the image to kind of help blend them in together to get something that I was pretty content with(and willing to share).
The hard part is fighting with AI sometimes as it likes to interpret things in unexpected ways, body parts coming out of adjacent areas where it might make sense, interpreting objects as something completely different(see background changes)
There's a bit of work in putting this together but it was significantly faster than anything I'd be able to churn out personally both looks wise and just composition wise. I'm basically using it as an advanced form of photo editing/manipulation to be able to create scenes I've always just imagined in my head. It's wild.