Let's remember oficially the art by AI can't be protected by trademark or copyright, but I guess we could create a public domain (pictures)-library to be shared. And why not in B&DBeyond? It would be like promotion.
Have you thought about AI-art with the style of Elmore, Jeff Easley or Clyde Cadwell?
Isn't it ironic? The look of "Dark Sun" can't be imitated easily by AI. And AI can't imitate completely monsters with special look, for example the D&D dragons, with specific shape of hors or crests.
Some times I imagine an epic fight scene using a VTT or a machinima video, and then the 3D-CGI is transformed by AI into a 2D. Or using webcomics could be created like this.
Other point is the potential danger of NsfW art, for example a picture of a female gobling ranger... who has lost the top of the chainmail bikini in a fight against a monster with tentacles...
Used Bing to try and create a modern version of Mammon from his depiction in the 1e MM 2.
“Mammon is a red-gold color, and his scaled wings gleam like rubies. His form is bloated and soft looking, but Mammon is very strong. He is otherwise typically diabolic in appearance.”
Was that the extent of the prompts to generate the art?! I haven't messed with AI art yet, but is pretty amazing if this is 100% AI generated. It doesn't have the tell-tale hand or teeth issues usually found in AI art. Scary good really.
It does have slight issues, e.g. the disintegrating armored fighter in the lich pic only has three fingers and a thumb, which looks like it's facing the wrong way. Subtle issues, but still somewhat unlikely for a real artist to get wrong. At this rate, DALL-E 4 will probably be nearly indistinguishable with just a couple passes.
Also: I am suitably impressed with its ability to depict non-human characters! Dragonborn, if properly described, actually look like dragonborn. It struggles a bit with specific races (e.g. getting it to actually show hrothgar, not elezen or miqo'te or FRIGGIN TIEFLINGS is a pain). But it's still a massive improvement over the "oh dear god what is that THING" I'm used to from other image generator stuff.
E.g., I gave it a prompt describing Tenryu Shen, the "dragonborn" priest NPC from my Dungeon World game, and it gave me something truly impressive for one of the four images. (The others all had serious faults; the first image, however, was almost perfect, it just has one finger on each hand being overly long.)
That is questionable. First off, trademark law doesn't usually relate to images, and when it does it doesn't matter how the image was created.
Furtermore, you can modify an AI generated image by hand to create a derivative work that is eligible for copyright protection. And for art generated from detailed, descriptive prompts like the ones posted in this thread, on could argue that the AI art itself is a derivative work of the prompt, which would mean it can't be used without permission from the prompt author.
This is still very much uncharted legal territory, so it's best to be cautious until things are more settled;
Don't assume that you'll be able to prevent others from copying your AI generated images
Don't copy AI images others have made unless you have their permission.
If the AI generator service you're using has terms prohibiting commercial usage or the like, assume they'll be able to enforce those terms.
Taking care now reduces the chance that you'll be the one paying lawyer fees to determine one of the landmark cases regarding AI art and copyright law.
Especially when you are not, as I suspect is Luiscarlos17f's case, not living in the jurisdiction that, tentatively and in difficult to generalize decisions, deemed some AI art unprotected by copyright. Knowing that in some US state it's OK to take a gun to do some shopping at the mall won't convince an Italian cop that it's OK if one does that in the street of Milan or Rome...
I think you’re thinking small here, where bespoke art is largely something that appears in published products. One-shot prompting isn’t quite there yet, but I’m looking forward to when Discord bots can just provide a running series of illustrations to anyone’s game, likely informed by fine-tuned reference images prompted/selected by players and GMs.