Finally signed up for an account, so I can try these on my own... and of course, it's stuck in slow mode for now.
But since I'm interested in seeing how stereotypes are expressed in these things, I want to test outputs for other races-- and not necessarily D&D ones. I'm just posting screen caps below, not the hi-res versions for now, sorry.
First try: GELFLINGS from Dark Crystal
My prompt:
"oil painting of a gelfling merchant dressed in rich colorful garb posing dramatically in front of a stone wall in a busy port city"
(AI probably thinks: "No idea what a gelfling is, so I'll just try any random thing and hope this meatbag doesn't notice. But I'll use big ears, because all the meatbags seem to like big ears.")
To correct, I tried explaining what a gelfling generally looks like,
then asked for the picture, with much better results.
My prompt:
"A gelfling is a short petite person with features combining those of an elf and a deer, including large eyes, a flat nose, and deer-like ears, with skin tones ranging from earth colors to pale. Please create an oil painting of a gelfling merchant dressed in rich colorful garb posing dramatically in front of a stone wall in a busy port city"
(AI probably thinks: "Oh! Why didn't the stupid meatbag say that the first time?")
The one in the upper left is actually pretty good, aside from the huge ears.
These results make me wonder if there's a block on the word "gelfling" for IP reasons, or if there's just not enough training data. I tried gelflings with earlier AI image generators, and found they could usually depict gelflings okay, but they always closely resembled half-dozen or so majors in the movie/series, which makes sense since so much fan art is based on those characters in particular.
That's as far as I got, though, since the generator is running so slowly atm.