Weekly Topic Open World Game Design

This past week we’ve seen two ‘open world’ AAA titles launch, with both Horizon Zero Dawn and Zelda: Breath of the Wild. Open world games have had a surge in popularity in the past few years, so we want to ask you “What makes a ‘good’ open world game?”


Where do you think open world games succeed? Where have you seen them fall flat? What are some of the user pain points you’ve discovered in these games?


Enjoy some food for thought!
Insights from Todd Howard (Bethesda) | Q&A with CD Projekt Red Quest Designers Pawel Sasko & Mateusz Tomaszkiewicz


-----
(To see past discussions, go to this index thread)

 

Fun topic! But why isn't it posted in the Game Design forum? It seems spot-on for that forum to me...

 

Non linear gameplay?
Resource gathering solely for crafting purposes and crafting in general is getting a little tiring imo.
Anybody have a good recent example of crafting done well in an open world game?

I'm bustin butt on a tiny puzzler - I can't imagine developing an open world game - at this time.
If I was doing game dev full time and had an army of devs at my disposal - maybe -

 

In the first 6 seconds of meeting someone you already have a good idea of their personality. I feel like this has a similar impact on games, but in this case, an open world game. All the games I really enjoy playing are games that capture my curiosity/attention from the start.

A great open world game makes you "feel" like you are actually in the world. It has a living environment that tells a story. It has depth. A place that enables your mind to run free; where you can discover and explore.

I think where open world games fall flat is when they try to be too much like another game. You can't re-create a first time experience for gamers. It just doesn't happen. Companies constantly threw out games with systems similar to World of Warcraft. Some of them did okay, but World of Warcraft still stands at the top, and those others are fading away.

It's hard for me to watch companies dump millions of dollars and thousands of hours into these open world games, but most don't even come close to what people really want. We want another mind-blowing first-time experience, with features we've never seen before. We want endless discoveries, unique system's and terrain. And finally, we want a well-written story that connects all these pieces together.

Open world games were dead for years. Seeing Horizon, and Zelda, Witcher 3, really gives me hope for this genre. I'm excited to see what comes of this.

 

I think this is a really good insight.

On the other hand, there are lots of first-person shooters doing much better than Quake. So, I think it depends.

To make that Quake reference more on-topic: a game like that wouldn't be fun as an open-world game, because the players would spread out too much and not encounter each other. I'm not sure open-world works for multiplayer competitive games. They work great for single-player or co-op games, though, because it doesn't matter if you wander off... if it's designed properly, there will always be content for you to interact with.

 

Best open worlds are those that are enjoyable to explore. Things like repeating the same structures or topography over and over feel bland and not fun. It takes a lot of work to create a truly interesting large open world.

 

Warning: this will be very, very long. I am fascinated by and a huge fan of the concept of open-world games.

Excellent, excellent talk which probably influences a few of my views:

In my opinion, open-world games are the zenith of video games--at least video games that try to tell a cohesive story and put you "in" a world. Everything outside of the main story itself serves to reinforce the fiction of the world, and gives the story (main and side stories) that much more impact.

I think the first "open-world" type game I ever played was American McGee Presents Scrapland.

Spoiler: Wall o' text about a game I love

fantastic quirky and funny story about a naïve robot in a very cynical world. From just about the beginning of the game you were open to do whatever you wanted. There wasn't a whole lot to do, but what was there was fun. You had around 6 or so different public venues you could go in (the police station, the newspaper office, the town hall, etc.) and four different "outside" districts connecting all of them. You traveled outside by spaceships. There were maybe a dozen or two different types, but they were all modular--engines, weapons, and hull were all separate components and could be mixed-matched as one chose.

It handled death one of the best ways I've seen in a game: you were a robot, so they uploaded your "mind" to a big computer (they called it "The Great Database," and it was controlled by "Bishops" who had a penchant for...creating work) and thus you could respawn, as long as you had the money and the "lives." If you didn't, you woke up in jail, where you could escape.

Apart from your character, there were maybe a dozen different archetypes, as well as around half a dozen other unique characters. Each character had a special ability: Staplers had a double jump, Nurses could bash someone over the head with a hammer, "functionaries" could speed up time, and your character could "dash" through any other character(destroying them).

The main thing that got me though, was the fact that the game world was so alive. Outside there were dozens of ships zipping about. Bankers and cops darting about shooting each other. Smaller generic ships following "paths" or roads through the city. And inside, it was the same way. Dozens of characters in a building. And each one could act out their ability at will. Small cops would go up to someone and demand money. If the character refused, the cop would turn into a giant cop and start shooting at them. Other nearby cops would pick up the scent and you'd have these massive shootouts in the middle of a public space, with bystanders getting vaporized.

It was a small world compared to many games out today. But the...I can't think of the word, but the "ability to act independently" of the agents was something that isn't replicated in most open worlds today. A bandit in The Witcher will never spontaneously attack a nonviolent citizen NPC. Bethesda games have NPC schedules, but there's no variability. In Scrapland, you truly didn't know if the banker across the room was going to steal money from a bishop, who would then attack the banker with his laser weapon, who would then be attacked by the police, who would then be attacked by other bankers nearby...I can't recommend this game enough, if that isn't obvious.

My next experience with an open world was Oblivion. And as much as I loved Scrapland, and consider it a game ahead of its time, Oblivion opened my eyes to the potential of a video game world to be a "real" place. Just the feeling of walking along a road in the middle of nowhere, and gradually seeing a city appear, that you could approach and then go into. It was just an incredible experience, and it made the world feel far more real than almost any game where you jump between random locations all the time.

I remember one time I was walking along the road (I think it was near Skingrad) in the middle of the night. I encountered the Count of Skingrad. Now, I don't remember exactly how I knew this (I think it was the Dark Brotherhood questline), but I knew he was a vampire. Immediately I got a very uneasy feeling, suspicious of his nightly prowl. I don't recall what happened afterwards ( I don't think he did anything special), but by encountering this character in a different context, and then in the open world, I was able to experience something "immersively" that would never happen in a paint-by-numbers linear game.

More recent open-world games (outside of Bethesda experiences) have been more...shallow experiences, I would say, because of the distance between the player character and the characters in the game. I think this is a consequence of the (largely) mindless push towards larger and larger spaces. I don't mind larger spaces, but in such places for some reason the dependence on meaningful interactions seems to have decreased. In The Witcher I can speak to random NPCs, but all they do is give their one or two-line barks. And they don't do anything on their own. In Ubisoft games I can't even talk to them (somewhat understandable since they don't make RPGs). They have a type of systematic interaction, but it's very shallow. The random thug I kill to stop him from killing someone else, the random child/thief running away...the game generated these moments ago, and they'll disappear the moment I get too far away.

And part of it is that these interactions are created specifically for the player to intervene in. Far Cry (and the later AC games) did a little bit better with this one, where you would get fights between factions beyond the players influence.

That isn't to say that I don't enjoy these games, but they aren't as immersive in my opinion because of these things.

I suppose there are a few different criteria (in my mind) to create the optimal "open world game." Now, let me preface by saying that this set of criteria (and the "optimal open world game") is inherently subjective. Different people are looking for different things when they play these types of games. But in my mind this group of things define the most meaningful world for a player to experience ("experience" is critical--it should be like going to a different country, not like going to a theme park where everything is directly for the customers).

Story. This seems backwards: an open world frees the team and player to focus on things other than the plot. However, context is everything. Open worlds with no kind of designed impetus upon the player require the player to come up with their own meaning. And while that does work (look at Minecraft), at that point you're not crafting a world. You're merely creating a space for a player to inhabit. The optimal open world game in my mind is a crafted world which the player can experience. And to convey that world you need "story" of some kind. That doesn't necessarily mean a main plot driving you through bombastic events, but a context through which to view the world and your character.

Agents and Systems. As I mentioned a lot of open-world games today are almost totally scripted, where the NPCs are actors more than agents. The Witcher is an example. But the best open-world games have agents and systems working in concert. They can act on their own volition, and the world responds to them. These together create a world that is alive and real. Assassin's Creed has systems, but it doesn't have agents. It is "alive," but it is not "real." By having systems, I know the world has its own logic and functions on its own. By having agents, I know the world's logic and functioning is meaningful - it impacts "real" people or beings. These are typically separated out (like AC), but for an optimal world they should be together.

Furthermore, these things don't just define how a player observes the world, but also how they interact with the world: I don't really care if I run over a randomly spawned character in Watch Dogs or Ghost Recon Wildlands, but I'll be far more circumspect in my actions towards persistent characters in a Bethesda game (though they still have the randomly created enemies, they do have the agents as well).

Memory or 4th dimensionality. Many games drop you into the middle of a world which for all intents and purposes was just created. Everything happens in the moment, and is gone the next moment (or within a few moments). There are two aspects to this--history and current events. A world with history is one where things have happened before this. Many games pay lip service to this, but games such as (again) TES have detailed fictions describing the rise and fall of empires. This is not a momentary sandbox which the player jumps in and out of. This is a world with a past, and that past informs the player's experience.

The second component is current events, or the world's response to the player. Most games really don't have this down--probably the closest would be MMOs because of how they work. But anyway, this refers to how the world "remembers" what's going on right as the player plays. If the player kills all of the sheep in an area--if they save and reload, are the sheep still dead? If the player leads the police or "law enforcement" forces on a wild chase which they eventually escape, what happens when they walk up to those law enforcement characters a few minutes later? Reputation systems like (yes, again) TES has do this to a point. Metal Gear Solid V had this in an interesting way (or at least in theory--I never saw it in practice), where enemy characters would counter the way the player played--if they sniped characters frequently, the enemies would begin to wear helmets (or so I heard--I never saw it in practice). I would like to see this in more impactful ways, however, where the agents respond to such things in a permanent (or semi-permanent) way.

Economy. This is something that has not been done really much at all, but I think it's critical to create a living world. In the real world, nothing is infinite--neither goods nor capital. Games, on the other hand, regularly create and destroy things at will, creating a disconnect with reality. To really feel the impact of something, economy needs to be implemented. In Skyrim, the player could chop wood and sell it. This was an infinite source of money. But imagine if doing so actually removed trees from the world. If the characters you sold the wood to actually did something with it. It would change the way we view the hundreds of items we craft in game, the many "junk" items we pick up as loot from bandits. I think (emphasis on that part) this is where an "optimal open world game" needs to go.

There are probably more criteria, but these are the ones that really pop out to me.

I'm only one person, but I am trying to make an open-world game. I'm doing it through the "power" of procedural generation, and the concepts of agents and systems. If I can build a simple world, and then put in agents with (relatively) simple rules, their interactions with one another will create something more than the sum of its parts.

I need three or four main things to create the "base" of my open world game:

1. Environment. A procedurally-generated world within which everything takes place.

2. Interactions. Interactions between agents, and between agents and the environment. Things like cutting down a tree, tilling the ground, attacking another person, hunting an animal, building a house, etc.

3. Agents. Both human and animal agents to do the interactions above. Not really talking 3D models, but AI. Each one needs a semi-randomized group of variables which define their interactions or actions.

4. Objects used by agents. Trees, plants, buildings, tools, weapons, etc.

With these, the game world should be able to "play itself," which is what I am most interested in. And while it's definitely going to be a ton of work--I think it's more than achievable. Ultimately it's about crafting systems which can reproduce themselves, rather than constructing everything oneself with a team of 1000 people (hello Ubisoft).

All of the ideas behind this game were founded two years ago, upon reading the book Prey, reading about Oblivion's Radiant AI promises (which didn't come to fruition), and discovering cellular automata. The idea of complexity, of interaction, of the whole being more than the sum of its parts, grabbed my attention--and still has it (not just in games, but in my career as well).

Unfortunately, most of what I plan for the game only resides in my head at the moment. However, through TerrainComposer I've managed to create an algorithm which can give pretty varied terrains with different seeds. The next step is to figure out all of the possible interactions, and then build an AI system which incorporates those types of interactions.

 

Just a query-Where should I post promotional poster for unity 3D games,from my company?I mean is here any particular category in Unity forum?

 

Work in progress would probably fit that best, however bear in mind that this is a technical forum for developers, good place for feedback and criticism, not so good to market to.

 

In regard to the pacing for open world games, traversal becomes the main downbeat. Every upbeat: event, quest, ect. Starts with the player in traversal.

For this reason it's important that traversal be varied- far cry lets you chain different forms of traversal- walking, to driving, to gliding. And breaks it up with small firefights of varied type.

Traversal types also sets the scale of the world. Shadow of mordor traversal involves parkor-ing around the landscape, which is broken up by killing or avoiding gangs of orcs. As a result, mordor is many sizes smaller than kyrat or skyrim ect.

 

I think this is a big problem. Everyone thinks they want huge worlds but then they use techniques to make creating the large world easier. In doing this, you lose much of the magic of an open world. It is boring to walk through "miles" of repeating terrain with nothing in sight. It is tedious to have to hunt over boring terrain for NPC or other players.

Of course, it is cheaper if everything is automated.

 

I really like how they ride that line in the new Zelda. The world feels absolutely HUGE. You can climb to the top of a mountain and look far out into the horizon and it just feels overwhelming (like you do in real life when you see things from that perspective). But the size simultaneously feels manageable -- you know that you can reach every location you see, and the various methods you have for traversal keep things varied like @LMan mentioned. It's a vast world (and I've only uncovered maybe 1/4 of the full map so far) but each area seems built with purpose, the villages are in places that make sense w/the surrounding terrain, and NPCs feel like they actually "live" there, with their own little stories and routines that play out over the day/night cycle. I don't know what it will be like when I've uncovered the entire map, but for now at least I get the impression that they spent a LOT of time making sure that all the terrain looks unique enough to navigate through, even without using the map, just based off identifying features in the environment, visually guiding you to places of interest. And there are enough collectables, enemies, hidden easter eggs to give you a bit of reward for giving into your curiosity when you see that single rock at the top of a hill. I suppose this is a benefit of a hand designed world over a procedurally generated one.

 

My first real experience with open world games was Asheron's Call. I could explore for hours on end. I never was big on just camping out in a dungeon and farm XP, I wanted to just run and fight and run some more. Same with World of Warcraft, although not as large it was more densely packed. When I play campaign style games I tend to explore every inch of every map until there is no uncovered parts. Just my personality I guess.

As pointed out by others I think a good variety of topography is vital as well as a few Easter eggs here and there. I actually saw one of the rare Tardis spawns in Asheron's Call once. Almost died trying to get close enough to get a good screen shot of it.

And as Ch33rios pointed out, so much of it is that first time experience. Can't recreate that.

 

Absolutely. I love hand crafted worlds so much more than procedural. It is a lot more work and much more expensive but worth the effort if possible. I rather miss the older games where ever discovery was a treasure, a hidden spring, a small shrine in the forest, a clearing among the trees full of wildflowers, a small village, etc. I see the benefits of procedural but I much prefer getting my hands dirty when making my terrains.

I watched a video about the making of Fire Watch and how players didn't mind traversing the distance if there were interesting places strategically placed along the way, spaced out just enough so before they were bored, they discovered something special.

That is advice I plan to follow.

P.S. Silly me typed Firefox instead of Fire Watch.

 

So... who's made a large* open world game in Unity? How did it go? From a tools perspective, what worked well and what didn't?

* Such that floating point imprecision becomes an issue.

 

Will Unity bring open world management editing tools supporting terrain tiles also, and ready to use streaming ?
This is what matters.

 

I definitely agree with this, however I don't see a way around procedural generation for a really open openworld game, for a small team or solo dev.

I think with a lot of procedural generation though, people miss the mark with what counts in terms of making an environment interesting and valuable to the player. Having spent a lot of time thinking about proc gen and how to do it properly, I've come to the conclusion that the difference between handcrafted and a sort of 'good but slightly generic' design ultimately doesn't have too much of an impact on whether something is interesting or not. When you're in a game, I think as a player you don't really notice the environment all that much, except if it's missing some threshold quality that you feel should be there. But after you take a good look around it mainly just sits in the background largely unnoticed.

In fact, the problem as I see it is that normally handcrafted levels come with handcrafted gameplay, and the latter is what ultimately makes it interesting or not. When a game uses procedural generation, it often generates lackluster gameplay (a limited and superficial set of options for the player) to go with it, according to some simple formula that is unlikely to ever break down from being incompatible with the generation rules. No Mans Sky was a good example of this imo - the world generation rules became so complex that ultimately it was difficult to spawn any kind of interesting gameplay that would not contradict something about the world it operated in.

The thing is that although an openworld-size amount of handcrafted content is hard to come by for indies, it's much easier to spend a lot of time on the gameplay formula, and there's much more opportunity to not only generate interesting gameplay, but also merge it with unformulaic gameplay in particular contexts if that's what it takes to make things more fun. Even if the gameplay 'formula' is a state machine with 1000 states, it's far easier than handcrafting content.

Aside from gameplay, the biggest fumble I see with procedural generation is a lack of understanding of how much of an impact context can have on the experience of something 'generic'. For example, with Skyrim, although I got sick of the same-old dungeons after a while, at one point I reached the 'north pole' of the map where there was just ice and water everywhere. That looked and felt fantastic, and once I crossed some lake on my horse and found a dungeon in the middle of the ice, even though it was pretty much the same as any other dungeon, the fact that it was set in the middle of a strange and unexplored 'icy desert' landscape made it feel totally different, like it had all these symbolic properties and background stories that the other ones didn't.

Another point about procedural generation (or generic openworld content) is that I think it's extremely important to control the overall relationship of each point on the map to eachother. In skyrim again, when I was near the main city, I felt like 'OK, there's a big open plain here, a forest there, and the city on the hill', and as I moved around near the city, it felt very 'correct' how each area related to the next.
However, as soon as I got some ways away from the city, particularly in some areas of the map, there seemed to be little relationship between dungeons and villages - they all just merged into a boring array of sameness.

Basically I think that for an openworld game, you need a focal point of course, and then everything else has to flow out from it with relatively high gradients of whatever properties differentiate it from that focal point. That way, not only do you know where you are and what you can expect to find, but those gradients offer a lot of potential hooks for anchoring procedural gameplay, as well as opportunities for driving interesting interactions between cities or clans, based on what those gradients provoke them to do in terms of for example war and trade, as well as their identities in the form of their culture. And by spending a lot of effort here I think it's possible to have a pretty generic appearance, that comes to life in a way that the player finds engaging when they interact with those locations.

 

WARNING: RANT AHEAD

I think that the popularity of open world games has little to do with story or production. Just look at Minecraft, which disproves either of those as being important.

No, the REAL reason that open world is popular is because it is open. It's right there in the name.

People are so sick and tired of the slog that modern games have become. It's downright painful.

Here's a meme that I'm sure everyone has seen and can relate to:


It's funny, but also kind of sad.

You get a game and you pop it.

You install it, which for some reason is inexplicably over 50GB.

Then you get a day one patch that is around 10GB to download.

Then you install the patch.

Then you load up the game and sit through a 5:00 minute long intro movie.

Then you start the game.

Then you need to sign up for the company's mandatory online service that no one is interested in because it doesn't do anything that Steam hasn't been doing for over a decade now.

After going through about 50 names such as "IH8EA", "OriginIsTrash", or "SeriouslyUbi?" and finally resign to "uh123ry8fhp98fjpoi" because you're sick and tired of dealing with it.

Then the account is activated.

Then you get hit with a popup about features.

Then you finally start a game.

Then it asks you what difficulty you want, even though it can be changed at any time.

Then it makes a new save file, which is also inexplicably huge.

Then you sit through another opening which lasts about 20 minutes.

Then you get to move around for about 10 seconds and another cut scene starts.

Then 10 more seconds of moving followed by another cut scene.

It does this for about 2 hours.

Then you get to the hub area and it pops up a survey.

Then it asks you why you gave "Plants versus Zombies 2" such a low score.

Then it lets you move exactly 10 feet before telling you something you can do.

Then half way there you get hit by another window that pops up and tells you another thing you can do.

Then Navi says "Hey listen!" and tells you how to do a barrel roll and every other action in the game.

And if you keep at it maybe, MAYBE, you'll get to start making some choices around the 3 hour mark.



Every single action, mechanic, and section is painfully explained to you and the whole thing is on rails with a giant "THIS WAY STUPID!" sign every five feet.

Rather than making new area, the game just uses wave after wave of enemy everywhere you go.

Rather than making areas that are varying in difficulty, they just make everything rubber band.

And rather than make an area for the player to play in, they force you through the specific experience they had in mind.

There is no more discovery, experimentation, or wonderment left in the experience.

It's just awful.



So when a game comes along that lets you just play the damn thing and figure it out for yourself (Minecraft, BotW, Dark Souls) it's overwhelmingly refreshing.

 

Minecraft is popular because it is a sandbox. Sandbox is not the same as open world, although, Minecraft is an open world sandbox.

But...it was done right with lots of discoveries, surprises and things to do. Sandboxes must have lots of things to do or they wouldn't be any fun.

Oblivion and Witcher are open world games but not sandboxes. If Minecraft did not have the sandbox part and was an open world game, story driven, procedural, it would not be the amazing game it is now.

 

Can you give some examples of "procedural gameplay" that you think work?

Imagine if the terrain dictated what kinds of animals were on the map. And the interaction of those animals with one another (predator vs. prey, etc.) determined the overall ecology. And human agents in that world acted on those animals, on the environment (with stuff like farming or mining), and on each other.

If a player was placed into that situation, and could contribute to any part of that, do you think it would be meaningful gameplay?

Additionally, can you expound a bit more on controllling the relationship of things to one another? I don't quite understand how a dungeon and a village's location might be changed to be "better." How might you do that (both in that specific example, and in a broad sense)?

I can appreciate your frustration, but I think you're exaggerating things. Further, those things are tailored to specific experiences. There are linear FPS games, and there are open FPS games. There are linear third-person action-adventure games (Uncharted, probably one of the most acclaimed series in the world...), and there are open third-person action-adventure games (AC, also very famous). There's no one type of game--and honestly, only barely any "one type" of AAA game.

 

Honestly, the procedural is the level design, making it easy to create.

Not sure what the difference would be if you placed spawners by hands in the world in the locations where those creatures would most naturally live. SWG did this very well, different creatures in different environments...and that was years ago. It is nothing new.

Ryzom had an ecological system as have other games. Farming and mining is common in games now.

Of course it is meaningful game play but nothing new.

 

I'm of the probably dubious opinion that anything a human can do can (in theory) be replicated with computers, as long as you put in the proper inputs. For example:

Build a town only in areas with a certain amount of flat open space. This would be automatic, so towns spring up in a relatively random set of areas away from each other.

Then, build roads between the towns. The "cost" of the road would be both by overall distance and by the grade of individual segments, So a road might curve around a steep hill rather than go directly up it.

Stuff like that, extended vastly across the whole game, so that it's totally unique based on the initial terrain, but it's also meaningful in its design.

 

I am of the opinion that you need to go back in and add the small handcrafted features that makes people want to explore. Having played a number of games with big worlds, the ones that I remember the most fondly are the ones that take time and effort to create special places within that world.

I think we want to convince ourselves that 100% procedural is the way to go because it is easier for us as developers.Level design is not easy, takes time, and an eye for it all. We want to believe it doesn't matter, but as a gamer, in my experience it does matter. When I talk to people about games, we remember those places, we notice them. We also notice the same villages and npcs and dialogue that is procedurally generated.....and not fondly. Witcher, a stunningly beautiful game was a good example of too much procedural.

Right now, I am using Gaia top procedurally generate my trees and grasses. But...I still had to go in by hand and paint in some areas because it need special attention. I prefer the results after I give an area a special touch. Yes, it is time consuming, but people love what they see.

 

Thanks for the information.

 

I love open world games. But they have a tendency to all fall into the Farcry pattern of towers to climb or camps to be liberated, fetch/kill sidequests and randomly dotted loot, with a few story missions thrown in for good measure. After a while no matter how pretty or "living" that world is if this is all that's on offer it starts to feel like play one, you've played them all.

On the story side one major failing is that they have a tendency to feel like breadcrumb trails, or even worse curated exhibits where you're just told where to go, what to do, and you really feel like it may as well as linear. Very few open world games give you the feeling that you're in charge of your destiny. That you're the one sussing out where to go next or that events are dragging you along. You don't end up on a train because you followed a character that then brings you to a location where you need to do something, maybe a mansion that should be explored, that then gives you the next place to be. Instead you nearly always have a guide character telling you what to do often without any connection to what's just happened, they just happen to know more than you do suddenly, or your own character narrates the linear path you must follow, again furnishing the logic with information you as a player weren't privy to. It's like really bad cheap storytelling.

IMO To make a great experience your character shouldn't know the world you're in any better than you do, you should discover it together, find the locations of things or people from NPC's, exploration or detective work. The pacing needs to be fundamentally slower than a linear story. The greatest moments in GTA, or the Witcher are when you're cruising around and something happens that drags you in to the action, a cahracter "just happens" to be passing by doing something crazy and you realize you need to follow, sometimes it's the sidequests that shine.

What extends out an open world for longer play is how alive it feels, and how much chaos there is. It's not always possible to do, but it feels great if an NPC doesn't just sit there doing an endlessly repeated animation or bit of dialog. If random things just happen in the background, NPC's make errors while driving and get into accidents and end up in fights that in the right circumstances escalate into crazy explosions, ambulances, fire trucks, police hunts. It feels even better if actions have a lasting impact on the world, if that brawl and chase is reported on the news on tv's in a shop that you go by etc (an easy thing to do if there's no dialog just an areal video shown of events that trigger the police reaction).

Great openworld games to me also to have to look good. They are showcases for what's possible. Kinda like old 16bit demos on the Amiga. They should be full of tech handling tiling, procedural fill in for non-detailed regions to allow outrageous sized maps, strong style or beautiful artwork for the regions that need detail (cities, landmarks etc), beautiful shaders, volumetric lighting, fog, foliage, weathering and weather, time of day skies etc.

Also fundamentally gameplay needs to be fun. If you're meant to shoot then shooting should feel fantastic, not an afterthought. If you need to drive then driving should feel great, not a chore. You're going to be doing these actions a lot, so everything you can do needs to be really refined. If it's nto fun then no-one is going to bother exploring your world.

 

Not really. Actually I think this is the biggest burden of procedural generation in general. I know Skyrim isn't procedural as such, but I think it offers the right sort of anatomy for a successful procedurally generation game.

The problem is that AAA games don't really do a lot of procedural generation, because they don't need to. And when indies do it, they typically attempt to use it for absolutely everything (i.e. No Mans Sky, or Limit Theory). Nobody seems to be able to restrain themselves and simply make a really big game that relies on procedural generation to fill out the difference in work output between a few indies, and a 200-strong art studio. I don't know if they feel some kind of guilt for using proc gen, and can only justify it by producing an infinite amount of content from it, or what.

All I can say is this: if you want to make an open world where each part of the map is interesting in it's own right, you have to keep the gradients high all the way around the map (and non-local). These gradients can be anything:

• Hunting vs agriculture;
• City vs nomads;
• Type of animals e.g. (domesticated -> wild)
• Environment - e.g. mountains -> rivers -> coast
• Friendly -> tough
• Cultural values, e.g. conquest -> community

Now this is already pretty obvious. But it's not nearly enough. You can't just step outside of the city and from here to the far side of the world, everybody you meet just asks you to fetch some cabbage. The fetch quests etc should be integrated into these gradients. For example:

• Just outside the city, there might be a lot of quests to steal something from inside the city, due to the high gradient falloff of material wealth;

• Still near the city, but further from it, the quests might be similar (involve a focus on relative material weath) but be more along the lines of "I'm trying to move to the city, can you help me establish myself there?";

• Further away from the city, the people might be nomadic and have a distrust of it. To offset this lack of focus on the focal point of the map (the city) you could use the high positive gradient of danger from wild animals, creatures such as dragons, other nomadic tribes and even a harsh environment itself, to increase the gameplay value there.

You can see that no matter where you go here, you feel like you know where you are. And the player's identity can also drive some interesting interactions with different peoples depending on how they view them.

Now there's still a problem, which is that most of these gradients fall off in all directions at the same rate. It's possible to tweak them locally, but it wouldn't make sense to have nomadic tribes for example vary a lot in terms of their basic culture.

And this is where context comes in. If on one end of the map you have a coastline, and on the other end mountains, the context provides a huge sense of uniqueness. At each end you might have the same position on the gradient curves, but down south might be a tropical paradise and north you have snowy mountains, and maybe east you have some forest people living up in the trees. And this alone can drive some cultural values (e.g. general happiness and approachability) but the environmental context itself may be enough to give the place enough uniqueness.

In the end, if the entire world is working together this way like clockwork, does it really matter if you have some 'generic' trees, rocks and even layouts to the villages? I think not. It's all about giving each place a strong 'personality'. If someone looks boring but has an interesting character, you're not going to identify them by their features.

So to sum up, with the focal point of the openworld map in the center, the gradients that drive everything in the game should look like this:



rather than this:

 

Clearly he's exaggerating to make a point. And it was very funny to read. There's a lot of truth in his rant. Many modern games do feel like a guided and pre-scripted theme park tour. And far too many games make you go through a lot of hoops before you actually get to play.

As for the topic of open world design, Breath of the Wild is a stellar example of how to do it right. In addition to everything that @aliceingameland mentioned, a major piece of this is how well thought out all of the various sub-systems are, including how they interact with one another. That holy grail of "emergent gameplay" can be seen in this title.

Not only do you have near complete freedom in where to go, you'll often have multiple possible solutions to overcoming obstacles in your path.

If you don't mind mild gameplay spoilers and NSFW language, find the Zelda: Donkey Breath videos on YouTube to see some of the more offbeat, creative stuff you can do in the game.

 

I suppose it depends on what you're looking for. I like very scripted experiences, as contrary as that sounds to my comments about procedural generation and open-world games. At the same time I can enjoy more open "survival" experiences like No Man's Sky (been playing Survival Mode a bit since the Path Finder update - it's pretty fun) or Space Engineers or Kenshi. I don't feel like I'm "wanting" for either type of experience.

 

It all comes down to the buzzword "agency."

Open world games succeed where they let the player feel in control. I still go back to "The Amazing Spider-Man 2" because it is freakin' sweet to web-swing around virtual Manhattan looking for petty crime. I have literally never seen a game so perfectly express how I always imagined Peter Parker's existence as Spider-Man really worked. Apparently, the developers read the same comics I did and had the same ideas I did about what being Spider-Man meant.

It has not escaped me that it plays an awful lot like the "Batman: Arkham Whatever" series, but being Spider-Man is so much more fun than being Batman. Spider-Man is just more like a gamer. Batman is super-cereal about everything and spent years training himself and has oodles of money. Spider-Man is an awkward young guy with no money who lives with his aunt while working a crap job where his boss hates him for no reason. More abstractly, Batman is awesome whether he has the costume on or not; Bruce Wayne is a hot genius billionaire. Meanwhile, Peter Parker is just some guy, you know? It's a lot easier to identify with that.

Where open worlds fail is by taking that control away from the player, even in small ways, because the player has become accustomed to having so much of it. Back to ASM2, whenever you do a mission, it literally means getting locked into this little box where you can only do certain things. When you enter a gang hideout, you are literally forced to play the whole thing as a stealth exercise, but you also can't just go steal the tech. The one and only path to completion in a gang hideout is to eliminate all the gang members by stealth. It would make sense, given that (a) your goal is to steal the tech and (b) you can often get to it without alerting anyone, to run grab the tech and just scarper without anyone even knowing you were there.

Similarly, there are all these boss battles where you have to do exactly this one thing to beat the boss or you don't get to play anymore. Right now, I have Spidey stuck fighting the Kingpin, because the targeting system is garbage and it's really hard to target and eliminate the second wave of armoured guard reinforcements while Wilson Fisk is smacking me around. You have to shoot them with ionic webbing, then beat them up. I keep targeting the wrong guard, and when I try to switch targets the game "helpfully" targets nobody at all. This is not fun, so I quit the game and go play something else. Eventually, I delete the save and start over, because I really like the rest of the game.

And that's the major pain point: by making it possible to do anything, these games frequently make players fight the interface to do the one obvious thing they have to do. Like in "Dragon's Dogma," when you have to run toward the camera, using the knowledge gained from previous failures to avoid obstacles you can't see yet.

This is a challenge, not because the player is learning a skill (which is fun!), but because it is literally impossible. The game design makes it possible solely by restarting you at the beginning over and over until you get it right by memorising the locations of everything you might run into, after making it impossible by forcing you not to look where you're going. Seriously, major "Brave Fencer Musashi" flashbacks there. That's unfair af y'all. And then, on top of this, the game refuses to let you skip the unfair sequence.

Welcome to the part of the game where you do the same thing over and over until you get it exactly right! We read about something like this in Dante's Inferno once, and thought it sounded like a Really Good Idea.

You know what open world game almost never fails? Minecraft. Every single thing that happens to you in that game feels like either your choice or your fault. If something is hard, it's not because of the interface, and it's not because the game has locked you out of the tools and abilities you want to use. It's because you don't know something, or you didn't bring something, or you can't find something. But whatever something it is, you could know it or bring it or find it. You just didn't.

Although seriously, creepers suck.

 

I just saw this gif circulating on Twitter and Holy Emergent Gameplay Batman!
https://twitter.com/Beever/status/840132081824890883

 

I think I know what you're talking about, and it's something I've been thinking about a lot in relation to my own game. Basically, without missions and story, a game feels boring, but as soon as I'm on a mission and taking part in the story, and I get 'that feeling', I want to take it with me off the treadmill of the linear gameplay and go exploring or something.

In Homeworld 2, there's no doubt that the missions and story were what made it special for me, but I wanted to explore, spontaneously find bits and pieces that added more to the story, maybe interact with some alien civilizations in a non-combat context. And when I felt that way, the linear story structure felt like a limit.

I think an open world game done right, with a lot of procedurally-driven emergent gameplay, would give the right experience. The backbone of the world should be static, but everything in it should be able to morph into very different states, driven by procedural story arcs that are generated when you set something in motion.

For example, let's say you spend a lot of time fighting against clan B for any reason (even just your own enjoyment). Now let's say clan A is in a stable yet competitive relationship with them. The game state could read your behaviour and generate an NPC in clan A, that wants to go further, and assassinate some leader in clan B and dominate that clan. So if you do that, your next assignment is to destroy their towns/villages or something. And so on until this clan is challenging the king.

But in the meantime, the bigger that clan B gets relative to everyone else, the game state might generate more animosity toward them from surrounding clans, and provoke alliances based on this shared animosity that grow quickly.

This sort of thing is what really interests me about proceduralism - I'm not interested in 'infinite worlds of boredom', but rather a large yet limited fish-tank in which anything can change according to what the player does. I've had some ideas about a game set on a very large space station that depends primarily on social dynamics, but no time for that yet!

 

I think open world game should have some way to let player travel fast between place without affecting combat or teleporting.

A few example are Shadow of Mordor where you run faster every time you vault over an obstacle, or in Batman series where you can use your grapple to fly really fast.

It not only really fun to move extremely fast, but also give the player the opportunity to discover things hidden around the world, not to mention you got to avoid staring at the loading screen whileteleport.

 

I agree. When I discovered fast travel in skyrim, my enjoyment plummeted although I couldn't justify not using it. Before that, I had found a way to enjoy everything that I did even if it took a long time.

 

I always feel like the main story is a trap. Like if I get too far into it, it's going to lock up sections of the world that I can't play anymore. So I tend to run like hell from the main storyline and do every sidequest I can find before I even bother going to talk to the Jarl of Whiterun or whatever.

And in that particular context, Skyrim's actually quite a different game when the dragons are not spawned into the game yet. Just like in Oblivion, until you get to a particular part of the main story, there are no Oblivion gates. And in both cases, the world before these things enter it... is a pretty damn good place to hang out.

Then you open up the main storyline, and the looming Big Bad makes a big old mess, until you finish up and have to deal with the aftermath of events you couldn't control that made the world worse. You go from good to bad to not quite as bad.

This has infected other games for me, too. Like I start Rebel Galaxy and I'm supposed to dock and talk to some guy about my grandmother and LOLNOPE off I go to side-mission my way through the rest of the galaxy. That's another bothersome one, because you have this one place you can't go until you reach a pretty deep point in the main story... and then when you can go there you have to stop playing the main story, or you won't be able to go there anymore.

Same pattern, agency. If I do more than one or two main story missions, you'll mess up the world and I don't like that. Then, even if I do everything you wanted, you won't put it back like it was! How is that fair? I mean, at least make it better.

 

Definitely, this is ridiculous. There's a certain length you can go to to guarantee the integrity of the game path, but locking in and out of entire sections of a map is not one of them. It seems to me to be lazy, because you could theoretically handle it (even somewhat illogically, such as pausing the intent of the game's antagonists, or simply making them disappear from that area until the player has arrived in the main story context) in a way that doesn't simply put a barrier to the player's ability to move around there.

 

I love how 'Shadow of the Colossus' puts the player in a isolated world. there is nothing to do but kill some giant monsters, but the game creates an incredible atmosphere. This is what an open world game needs to do: Put the player in the world. Is not just have a thousand of missions and secrets, and have an soul inside the game is essential.

 

I hadn't thought of that before! How interesting that they gave the traversal of a linear game an open aesthetic.

 

Plot twist, horizon zero dawn is using procedural generation:

BTW the open world of witcher 3 is painted by hand, but the npc cinematics is pcg (mostly, there is still thing it doesn't do that must be added by hand, the rules has n't been discovered).

Assassin's creed do use pcg as a tool.

The gdc of the new zelda has been released on youtuve for the curious, they made a 2d prototype of the game and devised a "chemistry engine" with element to material interaction.

The problem of pcg is that most people have stayed to world generation (think spawn and placement of things), not aesthetics composition, not gameplay composition and not story generation. You don't have procedure to lay down line of sight and its interactions with path for example (weak attempt by academia but nothing substantial). Key word is interplay, the gradient system proposed by someone would be boring because it does not create emphasis, only plausibility, most pcg has no notion of emphasis. However shadow of mordor (and its sequel) seems to have a basic framework for story pcg with emphasis (pcg npc reaction that tell a simplistic but coherant story) that beg to be expended.

No man's sky has a limited PCG color composition and it allows it to distinguish greatly from similar product, and is successful at being a screenshot generator, mostly because there so few elements aesthetics collision are bound to happen, supported by some choice like having arch and a very deliberate visual layering, having random craft flying above your head was kind of genius. When after 10h of harsh survival you emerged from a cave and the accidental framing of a valley, giant animal at a distance with some ship flying over and two planet at the horizon can be breathtaking. The thing is that this pcg is driven by artist with a specific kind of emphasis in mind (70s cover) and they succeed at making these accident happen. Too bad care about other aspects (world, gameplay, story) didn't measure to this.

When teila said that she need to go back add interest to gaia generated, its because gaia is gear toward world generation plausibility, not aesthetics generation at all.

Spelunky has pcg level design that some rank on the same level of hand made generation, mostly because they have a pcg rules that implement design rules of progression through a level, something most pcg game just don't have (Beyond simplistic later level pick harder monster and cooler items).

PCG is a complex topic only due to the myth and misinformation around it.

Open world can benefit from pcg if thought are actually put into it.

 

When you think about it, the real world is apparently pcg, regardless of whether you think anyone wrote the code or not. I mean, the wind blows for a reason, and the mountains are where they are for a reason, and the rivers flow the way they flow for a reason... animals, including humans, settle where they do and have the populations they have for a reason... whether you say the eagle was created or evolved, there is a procedure that dictates where eagles are and how many there are.

The more completely you understand that procedure, the more accurately you can duplicate it, for better or worse. You can say "deer might appear here" by creating a standard encounter table and sticking "deer" into it, or by having a bunch of parameters on the deer that let the game figure out whether a deer appears in a given place or not.

Look at Mount Rushmore. The cliff into which it's carved is pcg, but then you have direct interference to produce the final result. And when you think about it, the interference is also pcg - go over to Egypt, and you have the manual sphinx in a pcg desert, because if you give us a canvas we're going to paint. That cliff made us build Rushmore, and that desert made us build the sphinx. It might have been something else, certainly, but it was going to be something no matter what.

We have a sort of holy grail in our heads most of the time where we imagine building a world that we don't have to touch. Where we click a button and an open world pops out of it which is automatically populated with people and animals and plants and an apparent history. We can't do that with 100% pcg yet, and maybe we never will. But holy crap does pcg make it easier to focus human attention only on the parts it can't do.

 

Even in your description you still stop at world pcg, I have been talking about aesthetic pcg too, but remember pcg HAS to be designed, it's not just pressing a button and it magically understand what I want, you have to put work into explaining what to do to the pcg, most people don't do that, they use adhoc technique instead of building an aesthetics language for their pcg.

For example take photography, it's based on a certain number of rule set, if you are entry level you know about the rules of third, at higher level you start thinking at the semantic level (looking for direction, contrast, framing to tell something, etc all of which are still rules base), it's based on seeking out, in a highly chaotic world, heighten point of view. We can define art as loosely a collection of heighten point of view (pov), world pcg give you the plausible chaos, but games are art they are a collection of heighten POV. Fantasy world have the most density of heighten pov beyond plausibility, realistic game like unchartred have heighten pov that remain in the realm of plausibility. A great aesthetics PCG would filter world pcg (really just generate directly) to create a collection of plausible heighten POV.

You example with the mount rushmore is quite easy to create, it's all about emphasis, specifically in this case, line of sight, scale, etc ... that's rule base, you can generate a path around a point of interest (or POI around the path), spawn things that create elevation and spawn relevant face on it, trace line from the poi to the path and spawn framing elements along that line (tree, arch, etc ...) and suddenly you have something that look deliberate instead of just plausible.

The way to understand pcg is to understand the power of grammar, grammar don't make for great novel alone, it's how you used it, all novel still follow the basic rules of grammar and use a finite set of words, but the endless combination is what gives us the expressiveness of the human mind. And combination isn't enough without INTENT. You need to generate intent to break from chaos to deliberate, and most pcg implementation aren't there yet. They are treated as random boxes, when in fact we should look at them as reverse parser, aka language generator!

But to come back to this assertion:

That's true for a very simple raison, not even a human can 100% replicate what is in his mind nor someone else mind, not even the best artist, we get at best good approximation. Internal life and experience are surprisingly difficult to fully express and render, but also incredibly personal. In general we value piece of art expressive power by how close they can bring that to us, and then it's are to find perfect rendition (which prompted a culture of scoring artistic impact). Because experience is personal, a machine can never reach that level, but so can't a human, pcg isn't a replacement for existing, breathing, feeling, having opinion, standard and preference, pcg can't be you, nobody can, but YOU can use pcg to express yourself. I mean complain about skyrim is that their dungeon are to similar and they were human made, it's not specific to pcg, good pcg is indistinguishable from a human hand anyway (nobody realized, until the gdc, that horizon was pcg).

And the idea behind pcg has been around in art too, it's call art direction and style guide, creating pcg algorithm is equivalent to doing that, but instead of assistant doing the grunt works, it's the machine that implement the designer vision, the quality of the directing will decide of the quality of the execution and its coherence among many artist with different style or the machine.


http://dblakelyfuller.blogspot.com/2008/02/more-style-sheets.html
http://dblakelyfuller.blogspot.com/2007/11/style-sheets-for-production.html
http://dblakelyfuller.blogspot.com/2009/01/when-obsessive-compulsive-friends-come.html

 

Good points neoshaman, I think it's not emphasized enough that pcg is not just about formularizing natural processes, but also (even much more importantly) formularizing the ideal human experience of it. And that ultimately means, at least in terms of bringing pcg to the level or human artistic skill, formularizing the artistic rules behind what humans see as aesthetically pleasing.

Something very interesting about the human experience of anything, is how much we filter out when we experience reality. All of the useless noise, boring and chaotic feedback from the environment, does not even reach the level of conscious awareness. But we know what to filter out in reality. In a game setting though, because it is to some extent artificial and contrived, we do not instinctively know what to filter out, nor do we expect to have to filter out anything since it's meant to be a distilled experience, it's meant to be artistic. So when some of that unaesthetic or chaotic stuff gets through in a game environment through unsupervised pcg, it really sticks out.

 

I don't understand the difference. When you generate a world, don't you inherently generate the world's aesthetics?

Um... yes. Software has no features by default. Each and every one of them has to be written by someone.

It is, however, an example. I am not talking about "make Mount Rushmore." I am talking about "make a procedure that generates wonders." That procedure would need to be capable of making both Mount Rushmore and the sphinx, but extraordinarily unlikely to generate either. It would need to produce the occasional Colossus of Rhodes, the occasional Parthenon, the occasional library at Alexandria... and it would need to never duplicate any of them. That's an intractable problem right now.

You might also want to explain how this is not aesthetic.

...because it is. Which is the whole problem. The procedure you just outlined will generate variations of Mount Rushmore and nothing else. A world employing this procedure will have dozens of Mount Rushmores, which makes them rather unlike Mount Rushmore at all.

 

The thing is that you have an understanding of aesthetics as "theme" or "object" not as "composition" I'm drawing attention to composition, for example my example wasn't about "mount rushmore" as an object, it was about emphasis as a composition.

You don't give enough data as to why it cannot create wonder, as I understand them those are objects with meaning, I don't see that as intractable. The library of Alexandria was the biggest (attribute) library containing all the knowledge of the world (meaning) and it was certainly place in a way that reflects its importance (emphasis). In order to create a wonder, you need to generate the meaning first (story building) then use other rule to convey that importance. Collosus of rhode was just a bigger statue, place at a meaningful place, representing something meaningful for the culture that create it, it's composition with intent.

 

Don't tell me what my understanding is. I keep trying to explain that the problem as I see it is that we can't make a pcg system that convincingly generates believable great works of art.

That is an idea. Composition is only one little part of that idea. So is theme. So is object. And you cannot generate a great work of art by copying anything.

When your pcg algorithm generates something with meaning, what does it mean?

If you don't know, that's effectively the same as not meaning anything.

 

But....no one wants to believe that. It is like the folks telling me that VR does not make people sick.

 

I don't know why anyone would hear somebody say "this makes me sick" and respond "no it doesn't."

Honestly, it buggers the mind.

 

Because they don't want to believe it. Like many things that might interfere with what they want to do, they just deny the existence of such things.

Not to say one can't make a perfectly fine terrain that is fully and only procedural. I say let it become the norm, then those of us who still handcraft will be special. lol

 

But that's irrelevant. What you want to believe doesn't matter, things just are what they are. How does anyone manage to live an adult life without understanding that?

I mean, belief against belief, sure. That's a debate. But when it's belief against reality, belief loses. End of. If you feel sick, you feel sick, and what someone else believes doesn't enter into it.

It can certainly be argued that it's possible to make VR that doesn't make people sick (or to make people who don't get sick from VR), and to bring things back on topic that it's possible to make procedural content that qualifies as great art. You might even disagree on whether we've already done those things or not. But to say "that VR does not make people sick" or "that pcg makes great art," these are testable. There is a reality in the question, so your belief doesn't matter.

 

@cdarklock I see you are going to go for the semantic and philosophical debate

I didn't think of the sphinx as an element of "Art" that pcg must rival, but as element of world design that is place in a way that make sense.

The problem with rivaling art is that it's so easy to move the goal post. When we point at excellent work made by machine and code, people are always trying to find a case where it does not apply. First it's not like human, then it's not like great artist (because of course all human are great artist) then finally it's not creative because it follow a procedure and then it's not meaningful because it's using randomness, we are very past decent production by that point and trying to rival god like figure It's testable and david cope have tested it! And you must be aware about what is happening in the AI world who made great stride into artificial imagination too (keyword, vector space).

And anyone who are familiar with debate about creativity know this is a debate that hasn't been settled philosophically, so mere human we invoke it as a win state to any argument because it's so floaty and unknowable no one can argue against it, no one know or agree about what it is!

Let's say I want to make an arbitrary fantasy story as a human, why would I want to do that? maybe because I like adventure and escapism? why? because life is boring or I aspire to the impossible. why? etc ... this is the problem of source it's unknowable because we can try to find it ad finitum and it will involve what it mean to be human and nobody will agree, it will be metaphysics, and that's the goalpost people are chasing.

We can settle with good enough. Algorithm have to start somewhere, and this somewhere is arbitrary, that's where people try to attack, after all we human start from experience right, which has value right? Well that's what you are supposed to code into pcg to begin first, my proposition around line of sight and the difference between world and aesthetics is grounded in my art education, I come from this place it colors my judgement of pcg.

But in a philosophical sense, we start from arbitrary point too, before experience kick in, we are influence by where we are born, you don't come to similar conclusion based on your genetic make up, upbringing, wealth, luck, aka a lot of chaotic parameter we emulate using structure (the actual design) and randomness in pcg. And btw many people don't believe in creativity because everything is remix, they will show you how all great works are derivative (so call influences).

And there is always teh argument that's it's not pcg who is doing the real work anyway, it's whoever design the algorithm, it's like saying brush can't make great art, and pointing at a Monet, and then saying it's not the brush it's Monet who did the work! Yep pcg is literrally nothing less than a very effective brush, it does not make great art if it's not used for that

I also have a definitive personal and clear definition of meaning, but that's long enough already

 

There is no debate. You are just not interested in the same problem I am. That's not because either of us is bad or wrong, it's because we're different people doing different things.

The problem that interests me is how to procedurally simulate multiple believably human-like cultures in an open world game. It's a problem I have been thinking about and sporadically working on for almost twenty years.

The problem you are interested in has absolutely no bearing on that. I am not trying and failing to be interested in your problem because I don't understand it. I am interested in a different problem.