Yes, though there should be some rules (like you can stop your content from being included, and checks to make sure the content is appropriate and not copying off anyone else, e.t.c…)
Callums Compass
•
Mar 29, 2025
No
Large Man Named E
•
Mar 2, 2025
This is an art program, not an AI program
O LOW
•
Apr 10, 2024
Nice idea. You could train a LoRA with a couple of dozen maps and use it with Stable Diffusion, e.g. via Nightcafe. However, Stable Diffusion can generate dungeon maps that look cool from a distance, but are actually fairly rubbish up close. Fine as art, but not much use for a game.
Procedural generation (rather than diffusion) is the way to go for maps, I think, because they have an essentially constrained tree-like structure from a data/model point of view, though diffusion could be used to prettify. :)
Log in to comment and vote
Comments5
Bloodillrigger
Aug 4, 2025
Yes, though there should be some rules (like you can stop your content from being included, and checks to make sure the content is appropriate and not copying off anyone else, e.t.c…)
Callums Compass
Mar 29, 2025
No
Large Man Named E
Mar 2, 2025
This is an art program, not an AI program
O LOW
Apr 10, 2024
Nice idea. You could train a LoRA with a couple of dozen maps and use it with Stable Diffusion, e.g. via Nightcafe. However, Stable Diffusion can generate dungeon maps that look cool from a distance, but are actually fairly rubbish up close. Fine as art, but not much use for a game.
E.g. https://creator.nightcafe.studio/creation/mfaj2J4KangofRvBWw6N
Procedural generation (rather than diffusion) is the way to go for maps, I think, because they have an essentially constrained tree-like structure from a data/model point of view, though diffusion could be used to prettify. :)
Mar 14, 2024