Fantasy Map Generator (2018)
azgaar.github.io
azgaar.github.io
There's a page at https://pixelforest.itch.io/flowscape which sort of shows it off - basically it's a Unity engine "game" which lets you make landscapes.
It came and was the full install, they were just giving a discount if you owned the previous version. It came with two additional CDs filled with DEMs. One with DEMs in the US, the other across the world.
Fun times, getting lost in the madness of creating.
If you occasionally check it and worry that there are no new entries, then be reassured that this happens from time to time. The author considers this his "Forever Project". For that reason, he's not forcing things so that he does not lose interest in it.
https://heredragonsabound.blogspot.com/2020/02/the-forever-p...
As well as also being open source, the page gives interactive demonstrations of each step that the process takes which is interesting.
(edit: clicking round a bit more, Azgaar's project lists the one at mewo2 as on if its inspirations/influences)
I have a World of Greyhawk enlarged and mounted on my wall.
Azgaar’s Fantasy Map Generator (2018) - https://news.ycombinator.com/item?id=21099344 - Sept 2019 (36 comments)
Azgaar's Fantasy Map Generator - https://news.ycombinator.com/item?id=17933765 - Sept 2018 (1 comment)
Otherwise very cool
What I'm more interested is making this a 3D world with procedural geometry, make cities, make it run as a multiplayer game... Not to mention you cannot store all that geometry, it's too large, you can only generate the place you're exploring on the fly and save it to a cache and then render it.
Then you would have to explain all your artists and game designers how to make all of this work. It's a lot of constraints. The multiplayer side won't be interesting unless the world is persistent, so you have to split the world into several servers to not overload servers, because players need to cross the path of other players for the game to feel alive.
AND THEN, you would have to adapt netcode strategies in very densely played areas like cities where you have a lot of players next to each other.
In the end, such game would be very popular, but I don't think you would need a map generator like this, the map would still be done by a human because it would yield better results. What matters is the high resolution procedural geometry, at a low scale: trees, terrain, buildings, etc.
So yes, that's essentially the story of developing world of warcraft, except world of warcraft has several world instances and phases, because their world is not very large when you look how many players they have, they can't have a single world, even when they use phases, which is not a cool feature.
Honestly, you only need some shared towns to be persistent. Everything else being ephemeral could be made a feature through lore (i.e. reality destroying mists which eats the world whenever nobody is watching, player characters try to fight against the neverending destruction by venturing out into the world, temporarily reclaiming parts for harvesting etc.)
The biggest issue would be to actually generate towns, npc story etc on the fly. I think you're really underestimating the size of that challenge. Generating assets like that is generally the most expensive part of game development with multiple people working on each one. If you automate that away, you'll be able to make a ton of money just by providing modules for unreal engine and unity.
(That whole series of videos is interesting for anyone who has an interest in procedural generation and simulation.)