I think they get a bad wrap due to how frequently they are used as a crutch to scale up content quantity without quality. Which isn't an unfair opinion to have given the fact that this is the case more often than not.
But at the end of the day ProcGen is a design tool like any other that, when in the hands of a passionate team using it with intent and creativity, can be an effective way to bring elements of surprise/randomness/chaos and/or remove tedious work from development to allow for more time to handcraft content where it can best be utilized.
Some games that show off how it can be an effective tool (not all specifically ProcGen Dungeons), Dwarf Fortress, Noita, Caves of Qud, Minecraft, Elite Dangerous, Deep Rock Galactic, a lot of 4x games (Civ, HoMM3, etc), also a lot of indie rougelike/lites (Rougle Legacy, Splunky, FTL, etc)
The important part, imho, is that the developers are investing the time to make it "good". Either by treating it like a core design mechanic with it's own unique/engaging qualities, or by treating it like a "quick rough draft" and going back to curate and hand craft quality content on top of it.
I'm sure they can be done right, but I've never seen them done right, so I'm not for them. Everything starts to feel the same very quickly.
Not even Diablo II? I think that's done pretty well.
Amazing when done right like Diablo II or Minecraft !
In my opinion it lacks the core essentials of game design.
But one ought to get used to it. With the A.I. boom, procedurally generated is no longer secluded to the dungeons and "rogue like" games, as the future in the mind of a lot of game devs these days is how it augments the possibilities of any given game. And while in theory it is true, in practice it translates into very bland gaming. Because it lacks the intention and precision in hitting whatever makes the contextual gameplay interesting and engaging in the first place.
But... to each their own, I'd say.
The proc gen maps in Diablo 1 worked pretty well IMO.
It's really interesting that one of the reasons Diablo 3 was a step down was the increase in handcrafted story elements that kept interrupting the procgen flow. Some games just do really well with it.
I feel like if AI is gonna be used in games in the future, they should implement it in the opposite way. Use it to make different stories and characters to populate a world someone actually created.
The Mantella mod for Skyrim is what I'd want out of AI video games in the future. That thing is kind of a beast imo, as you can use it to generate completely brand new stories and quests. This General Sam video where he plays through Skyrim with the mod, talking to any and all animals he comes across, really makes it feel like an entirely new type of game. I mean, he goes into a bear den after some town guards killed a different one, convinced one of the cave bears to follow him to town to get revenge for killing it, and even had a conversation with the bear over which healing spell "felt" better.
When done well it greatly expands the game’s replayability.
When done poorly it feels bland and boring.
I like Shattered Pixel Dungeon, The Binding of Isaac and Lethal Company, so sure they’re great when done right!
They can add a lot of replayability, but they can just as well very quickly make your game suck more than if it had purposefully made levels. (I think a prominent example of bad proc-gen in general is Skyrim’s radiant quests.)
well, I'm subscribed to every roguelike muni on lemmy, soooooo
I don't think I've ever seen "community" abbreviated as "muni"
It starts with us!
It puts all the weight of the game being fun on the gameplay mechanics/loop. For number-crunching genres like roguelikes it works. If a Zelda game did it, it would suck. I put a couple hundred hours in Deep Rock Galactic, so I believe it can work under the right circumstances.
If a Zelda game did it, it would suck.
While I don't find the idea particularly appealing personally, there are modified versions of Zelda games that randomize various aspects of the game (like what items are in which chests) and apparently a decent number of people do actually enjoy playing them. (Usually not on a first playthrough though!)
I enjoy A Link to the Past Randomizer, but primarily because it adds replayability to a game I'm already so familiar with. ALttPR becomes a puzzle of which chests/dungeons have the highest probability of containing progression items. Calculating that optimized routing in realtime while racing against a clock is fun. Also figuring out the best way to deal with a boss that you already know well, but now you have an unexpected equipment loadout is fun to me.
However. If I were to play a new game I didn't have any familiarity with and its item placement and/or map layout was procedurally generated, I don't think I would enjoy a first playthrough. I don't enjoy variety just for the sake of variety. The proc-gen would have to have some known parameters that allow me to strategize in how I approach it in order to not seem arbitrary. If I didn't enjoy the first playthrough of such a game, I might not be motivated to learn enough to enjoy future runs.
That's why I think I don't love Spelunky or Slay the Spire despite loving games that play similarly like Cave Story and Magic the Gathering respectively. I think I could love these games if I could reasonably plan ahead, but I feel those games have too much variance and the outcomes feel arbitrary as a result. Though that could just be my lack of dedication to understanding the bounds of the generated content.
Zelda 1 randomizer is popular, and it's all about procedural generation of dungeons and procedural assignment of items to locations. It's not as well designed as the original Zelda 1, not by a long shot, but it combines the familiar gameplay of Zelda 1 with the novelty of procedural generation.
If the game has good combat and good loot, rng dungeons means I can fight and play loot lottery forever. When encounteres are 100% scripted, with cutscenes or platforming in the mix, it gets old fast...
Depends on the game. In Deep Rock Galactic it's great, you always have to keep exploring, you never know what's waiting for you down there.
Edit: but it's not a classic "dungeon"
Functionally it is rooms connected by hallways, in terms of the generation. Its just that the hallways might be open(or winding) tunnels or might be dig-able terrain. But they do start the generation with placing "rooms" and then connecting them, which is similar to how other procedural generation methods work.
More info about DRG's caves specifically: https://steamcommunity.com/games/DeepRockGalactic/announcements/detail/4593196713081471259
I used to play DRG and with a driller and gunner you can get almost anywhere. With a full tank of drill fuel and a few zip lines it doesn't matter where the exit pod was, we could get to it sometimes before Molly got to it.
Hard to put a yes or no on this.
Yes, really really yes, if it's done right and it works well for the game.
No, really really no, if it's done wrong and it doesn't work well for games.
I think that like a great many game mechanics, the fact that it's been done badly many times doesn't mean that it can't be done well.
I'm pretty easy to please, as long as procgen fits within the overall vibe of the game.
My only hang-up is when there aren't enough unique/modular enough nodes and I can recognize walking through the same room several times in one dungeon. If that happens, I'm not playing your game for more than 20 hours no matter how much fun it is.
Procgen should increase novelty. If you're just rearranging the order of parts A, B, and C, that's bad.
i think they are great when they are able to provide variety in terms of gameplay, challenge etc, as opposed to just being different layouts. My first thought is roguelikes and roguelites - Some of my favorite games are Binding of Isaac and Spelunky, which do it really well, and I still play the original roguelikes like Angband sometimes.
Depends on the game.
I think that they have been used effectively in games like Starbound and Terraria or many roguelikes and roguelites.
I think that there have been some games where they do not work well.
Starfield has a beautiful terrain generator, but different terrain doesn't really change gameplay, nor does combat really scale up to making use of very large maps, so you have the ability to explore infinite expanses of planets, but it doesn't really provide much in gameplay terms. Aside from finding a cluster of useful resources near each other for an outpost, which isn't that interesting from a gameplay standpoint and doesn't need most of the terrain generator's functionality, it's mostly just cosmetic.
I think that they work best where how you play the game changes substantially based on the mix of features of the dungeon. Then throwing a new mix each time at the player helps keep things interesting.
I hate playing on a sponge. Handcrafted all day.
I find them really boring, especially in RPG contexts. The difference is night and day when you walk into a handcrafted dungeon that has situational storytelling, creative direction, and ambiance that conveys a specific feeling. Bethesda games do this exceptionally well, for example.
Handcrafting a world also gives meaning to exploration; when I explore a procedurally generated world I feel like I'm just walking around aimlessly, looking for just another treasure chest or enemy to fight. But in a hand-crafted world, there are specific things to look for, situational stories to be told, or even secrets to find that the creator hid. That's a lot more fun to explore than walking around in a glorified geometric algorithm
I‘m fine with it if the actual layout doesn’t matter, like in Hades where it’s just room after room.
love em. i'm actually working as a programmer for a game right now where procedural level generation is one of my main foci. sure, like any game system, it can be done poorly, and it's not everyone's cuppa tea, but i like it.
Absolutely essential for the full experience in a retro roguelike.
Usually repetitive and boring in 3D.
There's no shortage of games that have been successfully built around it. This is the whole basis for Rogue (1980). Games that are designed to be replayable would be far too repetitive if you only had a handful of premade levels.
But there also no shortage of games that didn't feel like they used it well, games that could've been significantly improved with hand-crafted levels. I played Persona 5 before 4, and this stuck out to me like a sore thumb when I went back to 4.
Generally, if level design is a core focus of the game, then the levels gotta be designed. But if maps are just a backdrop for a more mechanics-focused game, randomization may help those mechanics shine.
