For my RPG Project I'm working on procedural content for quests, NPCs and AI-- but not for levels / maps / environments.
If you've played games with procedurally generated worlds, then you've likely seen the opposite, where the maps themselves are generated. The rest are often static or cookie-cutter placements expected at targeted locations.
"Procedural Generation" means creating something dynamically through algorithms. That's a bit technobabble-ish, but you probably already grok this from the widely promoted No Man's Sky, or even the original Diablo. Here's my simplified example:
A program could be made to generate a classic jigsaw puzzle from any image. Not just as a simple overlay or even random pieces, but with formula to create the correct difficulty that seems random but is tuned to be fair or challenging. Weighted randomness. Jigsaws would be an optimum example, as good as hand-crafted puzzles.
(this puzzle wasn't generated, I just overlaid a jigsaw cutout. Sorry.)
Most examples of procedural generation are not as good as hand-crafted work.
I know a bunch of programmers that would love to challenge me on that statement, but I stand behind it. It's... situationally true. Beyond the qualifier of "most", hand-crafting anything implies a lot of time, work and artistry.
Diablo is a great example of map generation that's done well, to give you many combinations of dungeon layouts. Most action RPGs followed with similar systems. I particularly liked Torchlight and that's when I finally enjoyed the genre, because I'll admit the randomness of Diablo I & II felt less like adventuring to me and more dungeon-crawling. That's my take, but I cannot argue with the quality of the game. Sure, I could still say hand-designed dungeons would be better, but players would also burn through content quickly. Even an army of map designers wouldn't be able to keep up to the players' expectations of the endless loot piñata with fresh dungeons. Kudos to the designers of Diablo, they made a game that kicked off a genre spanning decades.
I'd like to do it differently: An RPG with finite maps, overworlds and underworlds. That hand-crafted feel. Yet I want replayability and I'm just a sole indie developer. So how do I do that? I'd like to mix it up with NPCs and AI that... wander... and form groups and maybe even build organizations of their own across the maps. Changing strategies and movements and patterns according to the players (it's a multiplayer game) and even according to other groups of AI. I'd say the changing landscape, but really I mean the changes on-top of the landscape. The terrain is immutable, but the rest will be dynamic. And the quests will plug in to those changes.
I've been planning this after a series of AI development talks by Dave Mark over the years, which culminated in a related talk for EverQuest Next as "Orcs by the road" or Emergent AI. That game was never released and frankly it was getting bogged down by a host of other mismatching features, but that stuck with me. I don't have any association with Dave Mark, but his work really inspires me.
(Dave Mark has a summary you can read about some of that EQNext history and his involvement.)
Summarizing the concept of "Orcs by the road": In a typical MMORPG you'd have a quest to defeat some Orcs that are waiting for you in ambush. Quests and mob placement are static in most MMOs, so players quickly learn (or lookup online) exactly where the Orcs are. Then either encounter the Orcs or avoid them with this knowledge of the quest at hand--
Imagine if the Orcs are actually smart and behave according to their needs and desires: They would adjust where to put their ambush according to where players currently travel. Or they might actually intentionally avoid the players if they feel hunted themselves. Procedural content via AI.
I'd like to take this idea into urban areas (as well as the forests) and create procedural political intrigue, where some nobleman might request that you steal a valuable item from a neighbor, or dig up dirt on a rival. As one group gains power, their opposing groups get hungry. I have a lot of thoughts and plans in this area and it will take a lot of work to achieve it. Factions, rivals and player-verbs beyond combat.
For quests and NPCs... I have procedural plans that still need to be fleshed out, but there are other developers on the forefront of research on procedural storytelling and related systems. I have a lot to learn and implement. I really love this stuff, so I'm excited as to where it may take me.
The maps however, are the maps. The land will be there, you can explore it, but it's a place and setting for adventures to occur within.
