5 ms·

I love this sort of thing, but the far more interesting challenge to procedural games is engaging the player emotionally and intellectually, and giving them a s

• AnIdiotOnTheNet 8y ago
  The difficulty is that it is relatively easy to generate patterns that follow rules because computers are really good at it, but it is incredibly difficult to make an algorithm be creative. What we really want when we say we want an infinity to explore and interact with is that we want infinitely many new and interesting experiences, which procedural generation has been, thus far, unable to provide. This is not surprising, even most human beings aren't very good at it.
  • dstick 8y ago
    But such a thing exists, life ;-)
    • swift532 8y ago
      Yeah, only in life we are bound by various constraints. Whether it's that you wouldn't want to be a criminal in real life, you don't want to risk your life in extreme sports or battle, or you were born too early to explore the universe.
      • thedirt0115 8y ago
        +1. Alternatively, maybe you were born with or acquired some disability that makes it difficult to do what you want in real life. VR games especially can be amazing here for people with mobility issues. See this, for example: https://www.youtube.com/watch?v=qs0QPkR3BcY https://www.youtube.com/watch?v=qs0QPkR3BcY ("Virtual reality opens new worlds to a man with cerebral palsy")
  • indigochill 8y ago
    >What we really want when we say we want an infinity to explore and interact with is that we want infinitely many new and interesting experiences Agree. > which procedural generation has been, thus far, unable to provide Disagree. I think Dwarf Fortress (and to a lesser extent, EVE Online) has shown how to do procedural generation that does result in interesting interactions. There are four important factors in successful procedural games: 1. Granular modulation (you don't just punch a dwarf, you punch out a dwarf's left eye). 2. Diverse interactions (lava can be used to smith weapons. Or burn invaders. Or mixed with water.) 3. Persistent state as a result of the above. In a game like GTA, you can stage a one-man stand against overwhelming military force, but when you return to the site of your last stand, not a trace is left. In Dwarf Fortress, even if you lose you leave behind traces that can be revisited in future games. This is also the magic of EVE Online's alliance endgame (establishing player structures in contested space and seeing your alliance people flying there), without which I think the game loses its luster. 4. Environments small enough to build up a history. You can remember sites of memorable conflicts in Dwarf Fortress because your play space is constrained to a single tiny piece of the larger world. In Minecraft, Starbound, or No Man's Sky, your play area is so large that you never build up memories of any particular place (Starbound offsets this somewhat by having set pieces, but loses that sense of location outside of those pre-crafted zones). In EVE Online alliances settle for months or longer at a time in a particular region of space so that they develop an emotional attachment to it.
    • Retric 8y ago
      EVE Online gains most of it's value from PVP interactions even if they are remote. The random world generation is kind of meaningless backdrop to the human drama involved.
      • Ntrails 8y ago
        The only thing that makes eve online engaging is players interacting with players. Especially at scale. The market/economy is incredible, the high-consequence pvp is both boring as hell and occasionally exhilarating. The metagame is nuts. It's a cool as hell set of concepts which would never make it into a modern game imo, but that's the strength of it and what keeps it ticking
• moomin 8y ago
  Skyrim and Oblivion are a good example here. Oblivion heavily used procedural generation to create the overworld. It resulted in a pretty samey experience. Skyrim, on the other hand, hand-crafted the entire overworld and paid attention to making sure that things looked spectacular from where the player was standing.
  • gameswithgo 8y ago
    Are you sure this is correct? I Feel like both were hand built overworlds, but Oblivion just had less decoration because of hardware limitations at the time.
    • moomin 8y ago
      I think I’ve misremembered this rather badly. It was used for things like tree placement, but not gross physical features like mountains. I’m sure I read something at the time that said what I’m describing though.
  • doytch 8y ago
    This isn't exactly correct. Oblivion's overworld and most dungeons were hand-built. The only procedurally-generated parts were unnamed caves/dungeons.
    • whamlastxmas 8y ago
      It's been forever since I read about this but the map itself was generated randomly. It was then manicured by hand. The buildings and cities and dungeons were all hand made, yes. But the dungeons were all using such similar art assets and there was a real lack of variety, so they all rather felt the same. Not that I'm complaining, it was an amazing game and probably one of my favorite gaming experiences of my life. It was way ahead of its time.
      • thom 8y ago
        It's weird to think of Oblivion as in any way procedurally generated, coming as the sequel to Arena and Daggerfall that really felt like RPG versions of Elite.
• jpmoral 8y ago
  This is why IMO a well-run pen-and-paper RPG will offer a far superior experience to any computer game for the foreseeable future.
  • gameswithgo 8y ago
    mmo?
    • jpmoral 8y ago
      Are you asking if mmo beats pen-and-paper? IMO, no. There are no programming limits to how 'alive' a pen-and-paper game world can be. MMOs have no persistence and NPCs are little more than cardboard cutouts.
• knolan 8y ago
  A problem I have with a lot of modern games is the fake sense of scale. In the Mass Effect games the Citadel is supposed to be some vast city floating in space but you’re limited to a few rooms with billboard textures to show the rest of it. On the other hand the game Fuel delivered a massive sprawling (mostly empty) wasteland to chew up in your buggy and was non-random per user. I would love to see this kind of procedural generation to deliver an actual Citadel type world where the developers could then sprinkle landmarks across it. You could just ignore it and take your flying car to the waypoint or you could drop down into the streets and trade, craft or just soak up the scale.
  • smogcutter 8y ago
    IMO games like the Witcher 3 and elder scrolls do a much better job of that. Mass Effect tries to trick you into feeling like you're in a big world with freedom to move around, but actually it's a lot more like a Final Fantasy type game: each area is a small "town" zone with the same couple shops, adjacent to a dungeon you need to clear to advance the plot. I like the Witcher style much better. Here's hoping their new game Cyberpunk 2077 lives up to it.
    • knolan 8y ago
      The Witcher indeed does a great job, and to a somewhat lesser degree Horizon Zero Dawn does well too. However these are essentially worlds with small villages and towns. Novigrad is pretty impressive but isn’t much bigger than a typical European small town. What I’ve love to experience is getting lost in the sprawl and scale of a metropolis. Oddly enough L.A. Noir did a pretty decent job of this.
  • setr 8y ago
    I think its more an issue with game design than it is with how big it actually is. Like you don't need infinitely generated worlds to produce a sense of scale, you just need it to take time to explore, and a good reason to explore. And this is ME's greatest failure I think: You can't interact with anything. You skip around space instantaneously, so the largest thing in the game is the citadel.. it's the only object that takes any amount of time to actually traverse (that, and planets, which you don't really want to explore). Yet you jump around that too, and are only actually interacting with preset objects. The problem I think is that there's nothing to interact with. And there's no interesting ways to interact with the things you can interact with. Like star control 2 is technically a small game, and a small world; it just feels big, because it actually requires time to traverse, and there's actually things to find. Zelda games tend to be even smaller (primarily a sequence of small rooms), and yet feel much larger. Again, time, secrets and interest play a large role in why it feels that way. people try to use procedural generation as a crutch, to avoid actual game design (see: no man's sky, minecraft; note the latter is popular by accident, adventure mode was intended as the primary, but loses its appeal quickly. Creative mode became the primary appeal by accident.) but its simply not sufficient without giving a compelling reason to actually explore and interact with the map.
• virtualritz 8y ago
  There are many different games types that can benefit from procedural content. I've lately been playing Dead Cells a lot, er too much. It's a roguelike-metroidvania hybrid with a bit of Prince of Persia sprinkled atop. All levels are procedural. Every time you die you start from level one again. But since it looks different every time, it doesn't bother one much or at all. Curiously, there is no narrative or story line at all. The Indie studio behind it, Motion Twin, simply doesn't have the resources or budget for doing this 'properly'. The typical approach is then to just use text -- just because you have to have a story. But they opted not to even do that. For me Dead Cells is a great example of on-the-fly procedural content done exactly right -- w/o any narrative.
• LargeWu 8y ago
  This seems to be geared more at tabletop RPG. When you're running a game, it's helpful to have these sorts of artifacts to enhance players' understanding, but it's not the sort of thing that GM's really want to spend a lot of time on, unless there's some specific thing you need. Having this just be generated potentially saves a lot of time.
  • Fomite 8y ago
    One of my interests for it is miniature wargaming campaigns. Where I really don't care how the city was built - just what it looks like before someone sacks it.
• mikepurvis 8y ago
  IMO the best is a hybrid approach, where the world and places are hand-designed, but procedural generation tools are used to amplify the creative potential of the person doing the design. I believe that the developers of Horizon Zero Dawn used this approach, and it shows— the world is lush and appears very organic, but is also thoughtfully laid out and not at all same-y.
  • wlesieutre 8y ago
    That's how Star Citizen is doing their planets; different procedural biomes that an artist can take and brush around to quickly build a whole world with details that you can walk around, but they can also zoom in and move rocks, make lakes, place structures, and whatever else. It's a much bigger scale than an artist could design by hand, but we're not at a point where an algorithm can fill a universe with memorable landmarks.
• ianbicking 8y ago
  In one article about Dwarf Fortress they talked about how they would write stories, then pull apart the stories to figure out how they are made and then how to make them randomly. It's a bit different from physical/realistic inspiration, and likely to lead to more fun in the results. If I vaguely remember, there's this sense of major plot points, and then everything else is connective tissue. You can imagine the same thing in a city design – what are the parts of a city that make it distinctive, or fun, or notable? You randomly choose some distinctive design elements, and then fill in. This generator has a bit of that. I like how the rivers or an ocean can dramatically redefine a city in this generator. But there could certainly be more. Interesting topology. Highways, instead of a couple equally-sized roads. A scattering of distinctive building forms, like a military base, convent, marketplace.
  • siavosh 8y ago
    The game "Caves of Qud"'s creators have given some GDC talks on procedurally generated histories [1] which I thought was pretty interesting. But it all seems very limited and hacked since NLP research is still very far from creating any useful text generation like good summaries let alone fiction. [1] https://www.youtube.com/watch?v=H0sLa1y3BW4 https://www.youtube.com/watch?v=H0sLa1y3BW4
• Merg 8y ago
  One game which for me is closest to Dwarf Fortress and which for me, tend to beat it in few areas ( graphics, while still being text only, structure of the world ) is Ultima Ratio Regum[0]. It is still in very deep beta state, but this doesn't stop richness and complexity of worlds generated in it. It is seriously underrated project, and whole blog if worth a read [0] http://www.ultimaratioregum.co.uk/game/ http://www.ultimaratioregum.co.uk/game/
• sandworm101 8y ago
  Why story? The breakthough of SimCity was the concept of software as a toy. The user creates their own fun. In a more modern context, open things to the community by allowing them to write and share their own stories. Anyone familiar with the history of Kerbal Space Program knows this issue. KSP's developers releases a "contracts" system that was essentially a random story teller. A few variables were randomly selected to generate a task. Of course many of the tasks ended up being ridiculous and they had to back-fill with a series of rules, but it is still rather silly. Then in a more recent update they allowed users to create "missions" and share them via steam (yawn). But this was layer atop the previous contracts scheme, resulting in a grinding mess that many/most players simply ignore. They instead play according to self-created story arcs and self-imposed rules. SimCity did it best. Give the players the tools to play, but don't tell them how to play. Build a world full of towns and potential then let the players loose to explore and play how they want. Don't try to randomly build quests. Create the interface so that players can write and share their own storylines. I envision something more like OpenOblivion, aka Elder Scrolls with community-created quests/factions/stories.
  • mywittyname 8y ago
    I wonder if a modern SimCity could use some ML algorithms to emulate how humans build cities by mining data from real players.