Designing Open Worlds

What would you consider to be “good” open world game design and “bad” open world game design?

Okay, so what are the benefits of open worlds? The benefits are, you have this huge amount of content available at one time, much like a metroidvania, except instead of the levels being hallways, it’s one big continuous space. This means that players can approach any given area from a variety of angles, and they are capable of engaging with any of the content open to them across the entire world instead of engaging with the content in a set sequence. Progression is defined by event flags instead of through the areas open to the player, or the player’s progress through specific areas.

Drawback: Big continuous spaces are bad level design. There need to be walls, pillars, and other barriers that . And because every point is connected to every other point, you can usually go around any sort of roadblocks in your way. In MGSV this meant you could run around the outskirts of almost every encampment and not engage in the core stealth gameplay. A lot of afghanistan had big pillar cliffs organizing the map into giant “hallways” as a result. Far Cry 3 blood dragon had a fair amount of success enclosing every camp in giant walls, limiting the entry points so the inside could have real level design.

The solution of most AAA game devs is to make missions that only concern a small area, which they alter to have real level design for the duration of that mission. Or they make areas segmented off from the rest of the map that have real level design.

So here’s my solution, build the world in chunks. Don’t do the whole thing at once. Build one part, make sure it’s good, then build another chunk next to it that is also good. Open world games have a focus on quantity instead of quality. Make many small chunks that work, and connect them, then build outwards from there. Test each chunk from multiple approach angles to make sure they work. Rather than starting with a giant world and trying to populate it, start with a small world and keep making it bigger as you’re sure each chunk is sound. Treat that as its own style of level design. Consider using Z-action and bridges as another way of binding levels together.

Beyond that, think of ways to play through segments of the open world that are interesting. The most common approach to this is scavenger hunts for collectibles across the world or again the mission structure. Alternatively there’s fetch quests which exist to make you run through the content that’s there, but traversal usually isn’t that interesting in open world games, because the ground is all flat, so that doesn’t work out so well. A smart idea might be triggering enemy migration and dispersal patterns as you move across the world, and selectively opening and closing different paths out of sync with each other so it changes up what’s happening all the time. A rondel might be a good metaphor for how to do this, a couple of them rotating out of sync with each other to rotate level elements as you pass through.

Alternatively, go with a gameplay style that isn’t really dependent on good level design, like Nier Automata did. Most stylish action games work on entirely flat planes and only have the pitfall that you can run past enemies rather easily.

9 thoughts on “Designing Open Worlds

  1. Space April 27, 2017 / 5:29 pm

    I’ve been wondering how do people design so that there can be fluid combos that players can uncover in fighting and hack and slash games? Do you just somehow make moves ‘comboable?’ Also, how do designers allow for the existence of certain tech, like wavedashing and blade mode canceling? Do combos and tech have certain principles or is it a matter of luck?


    • Chris Wagar April 27, 2017 / 7:37 pm

      I go over this a fair amount in my other article on frame advantage:

      Basically, when you hit someone, you put them in a state called hitstun. In this state, the enemy cannot act. It lasts a certain number of frames. If you can hit them with another attack before that state ends, that’s a combo. If you can hit someone with a move, then keep hitting them with moves that put them into hitstun long enough for you to hit them with another move, then you can keep the combo going.

      So the basic principle for whether a move starts a combo is just, “Does this move stun the enemy long enough for me to hit with another attack?” This means you can easily find all the moves in the game that combo by just looking for which moves have the most frame advantage (hitstun time remaining after your move recovers) and which moves start up in less time than that. So if a move has +7 frame advantage, it will combo into every move with 7 or less frames of startup.

      Then there’s cancels, which let you inflict hitstun, and then immediately do another move without having to wait for your move to end, meaning you get a lot more frame advantage, you can use the whole stun time. In fighting games, you’re limited to canceling only into specific things. For example, anime fighters let you combo into things in chains, like Skullgirls which lets you cancel LP > LK > MP > MK > HP/HK. And you can start at any point in that sequence and skip from any earlier move in the sequence to any later move in the sequence (though light attacks might not have enough hitstun to combo into the slower startup heavy attacks)

      Combos consist of different moves because cancels are effectively a limited resource, you can only cancel into certain moves. Also because moves with a lot of hitstun usually also have more startup than they have hitstun, so they can’t combo into themselves, requiring you to go from a slower move and combo into a faster move. So combos either start with the higher frame advantage moves, moving into the lower ones, or they start with a move at the start of a chain and move through the chain to the end. There’s other things that create variation in combos, like there might be a high damaging move that has a fast startup, but you can’t combo out of it, so you want to stick that on at the end of the combo after you’ve fit in other moves before it.

      Beat em up games usually have way more flexible cancels, something that lets you cancel anything, so infinites are really common in beat em ups.

      Combos and tech are a matter of luck and a matter of design. You can deliberately build these things, but sometimes they just happen. Explaining how these come to be is really varied and complicated. It’s usually a matter of a state getting canceled and carrying something over into another state. BM canceling was just a matter of deciding, “Oh yeah, Beam Mode should just cancel everything.” Like, every time you press a button, you’re changing the state of the character. During each state there’s a list of what states you’re allowed to enter. When the animation for a state ends, it returns to the idle state. The idle state is allowed to enter any other state. If they want to make a certain cancel possible, they can just say, “Okay, this state transition is allowed during this animation.”

      Wavedashing is harder. Air dodges work by adding to your root velocity, then reducing your root velocity while you’re in the air. If you land on the ground during an airdodge, what happens? They had a few different choices of how they could have resolved that. They could have decided, “oh, they stay in the airdodge animation until it ends, then enter a normal falling state which will allow them to then enter the landing animation.” They could have decided, “Okay, you enter the landing state if you touch ground during airdodge, but all velocity is set to 0.” Instead they decided that you should enter the landing state, but keep all your X axis air velocity, meaning that you could transfer the velocity of the airdodge to the landing animation.

      Any more questions?


      • Space April 27, 2017 / 8:46 pm

        Yeah, I’ve always speculated that tech, like Blade Cancels and Wave Dashes, is usually the product of chance.

        I’m familiar with how the basics of combos, I guess what I wondering is how they manage to make combos that seem flexible and dynamic, while making sure they eventually come to the end (more so in fighting games), and how designers can make sure that your basic options aren’t too overpowered in terms of comboability (ie new DmC).

        The comment about ‘cancels being a limited resource’ is helpful though.


        • Chris Wagar April 27, 2017 / 10:26 pm

          It’s not always chance. Blade mode canceling was intentional, as was jump canceling in DMC3 and 4 and Block canceling in God of War. If you want to stick a cancel in your game, that’s extremely easy. You can just do it. Unintentional animation cancels in various games are usually only there because of developer carelessness about which states can transition into which. It’s harder code-wise to prevent animations from canceling each other than it is to allow everything to cancel into everything, so sometimes devs just overlook that they didn’t prevent the character state from transitioning to another one. A smart developer would probably have a check before each state transition to see if the state to be transitioned to is in the list of currently allowed state transitions, but not every developer is smart, and sometimes they allow state transitions on purpose, not realizing they’ll have a certain effect (being allowed to shoot or sprint during reload, canceling the reload animation, is allowed in most shooters, because devs don’t want you to get locked down during a reload animation, but this has the effect of allowing people to cancel the reload just after the ammo counter is updated, giving them a faster reload time).

          Combos in Street Fighter 2 are called a glitch by many people, but almost nobody knows how they actually came about. In Street Fighter 2, performing special moves was deemed by the developers to be really hard, so they made it so releasing the button would also trigger the special move, and so the first 5 frames of an attack could be canceled into special. This was intended so if someone pressed the button too soon after doing an input, hopefully they’d release the button and it would come out by canceling the normal move. A side effect of this was that if your normal move had hitboxes come out in 5 frames or less, then that cancel window extended past the point where you hit the opponent. On top of that, since there’s hitfreeze when you hit the opponent, it also pauses your character within that 5 frame window, effectively extending the window to be even wider. If you play SF2, you can try this out for yourself. It’s possible to cancel ANY normal grounded move into a special during the first 5 frames, and every normal move with 5 frames or less of startup can be canceled. This is obviously not true in any fighting game after SF2.

          Combos in fighting games are basically about a bunch of limited resources. Imagine a character has an attack that has 5 frames of startup and 5 frames of advantage, this means the attack combos into itself. Cammy has this on her crouching MP in SFV, Ryu has it on his standing MP. It’s not uncommon. What prevents this from comboing into itself forever? Pushback. Every time you hit them, they get pushed back. After 2 hits, they’re so far away, you can’t hit them anymore. SF4 and SF2 had high frame advantage on a lot of attacks, so you could combo almost any move into any move, but they keep pushing the opponent back, so you can only hit them so many times before they’re out of range. Pushback also means that you need to hit them with an attack that has enough range to actually hit them, your close range attacks won’t cut it anymore, and your closer range attacks are typically the ones with faster startup and more frame advantage.

          So what other resources are there? There’s pushback, pushing your opponent away. There’s your cancel chains, which go one way only, but allowing you to skip around during them (in some games like melty blood you can chain combo from later moves in the chain to earlier ones, called a reverse beat, but this is still limited by pushback as above). There’s Meter, because meter can be spent in many games to reset your chain or your pushback, either by empty canceling your move, being used to activate a move that can be combo’d into and out of, or moving you closer while inflicting enough hitstun to combo off of, resetting the pushback. For example, someone might combo into super and then be able to combo off their super, but they can’t repeat that, because they don’t have the meter to use a super again. Some older games have a feature called custom combo, which makes everything cancel into everything, and moves you forward the entire time it’s going on, so for the duration of this move you can do whatever combo you want. The best custom combo is almost always to just hit them with heavy attack over and over again, until damage scaling is at maximum, then you want to hit them with lights since that takes less time and you’re getting the same damage as heavies by that point. In that case the custom combo is limited by a timer and by meter. For juggles there’s Juggle Points and Progressive Gravity. SF3 uses something called Juggle Points, which decrease the length of time the opponent can be hit before turning invincible in the air, so that after 6-7 hits, they’ll have such a short vulnerability time that trying to hit them again will simply miss and they’ll fall out of the combo. Guilty Gear and Blazblue have progressive gravity, which makes gravity increase the longer an opponent is juggled. Marvel Versus Capcom 3 even has hitstun decay. Skullgirls meanwhile decided to make a game where infinites are extremely easy, you can reset your combo chains in a variety of ways, letting you repeat the same chains over and over, but they limit this by saying that the starter of each chain must be different, so eventually you run out of starters. Then they limited it again with the undizzy system which is basically just a meter that fills as the combo goes on, and once full the person being combo’d can burst out.

          Then on top of all this, there’s damage scaling. As you hit with more moves across a combo, each move’s attack power gets weaker, meaning if you want to inflict the most damage possible, you want to hit with your strongest attacks early in the combo where they’ll inflict their maximum damage.

          Then beyond all that, you’ll want to use different combos because each move has a different utility in the neutral game. Like you might want to combo off of light attacks because they’re your fastest attacks, you can rapidly cancel them into themselves, and this lets you easily confirm whether you’re hitting before you start your combo. You might want to use a slow move that sets them up for a big combo, like Hugo’s clap in SF3 and SF4. You might want to poke them from afar, then cancel into a special without confirming, because you can’t get close enough to use a better combo starter and you don’t want to risk using a slow move that might get interrupted. You’ll want to use different combos based on whether you have meter. You’ll want to use different combos because it ends in a knockdown or carries them to the corner, which set up for more damage later, even if it’s less damaging now. Cammy in SFV has this tradeoff, she can end every combo in either uppercut or spiral arrow, and the uppercut does more damage, but spiral arrow leaves you right on top of them for wakeup pressure. Some characters have unique combos that only work in the corner (because fireballs won’t pushback if the opponent is cornered, and neither will juggles). Sometimes you want to use combos that will do less damage if they hit, but put more pressure on opponents blocking (Guile in SFV has a powerful standing HP to crouching MP, which is unsafe if the opponent blocks the HP, or a weaker crouching LP to crouching MP, which will only combo if the opponent is counterhit). Oh yeah, and counterhits, some things will only combo if you interrupt the opponent, which grants you more frame advantage, so there’s unique combos for that too.

          A lot of tech in modern fighting games isn’t there by chance. If you look at enough tech, then eventually you’ll catch onto the patterns and see how to implement similar things intentionally.


          • Space April 28, 2017 / 6:04 am

            I think I’m starting to get it now. Two more questions I’m curious about:

            Is it possible for pushback to cause an opponent to pushed more, say, diagonally?

            Also, sometimes I see combos where the comboer jumps/dashes into the opponents face multiple times in the middle of the same combo, but it seems that the second time they jump in they have to use a different combination of moves from the first time they did, and they can’t infinitely reset the combos that way.
            Is it that pushback might increase over time, or is there some other factor that is involved?


            • Chris Wagar April 28, 2017 / 11:10 am

              Diagonally into the air? Are you still talking about 2d fighting games? Because yeah, diagonal pushing is possible (though not very utilized) in 3d games, but pushing diagonally in a 2d fighter is just a juggle.

              Which game are you talking about there? In SF4, there’s a mechanic called FADC which lets you spend meter to cancel into a dash. In Guilty Gear Xrd you can do similar with Roman Cancels. In other games, sometimes moves leave you with a lot of excess hitstun, so you can walk during that time to help extend your combo a little, like in this sakura combo video:

              I’ll need to know what game you’re talking about to really explain it though.


              • Space April 28, 2017 / 6:24 pm

                The last combo (:45) seems to be the best example, as it looks like the player is using the rising knee move (4C) to close the gap multiple times, but they can’t do so to make the combo last forever.


                • Chris Wagar April 28, 2017 / 11:21 pm

                  I don’t really play Melty Blood, so I’m not really familiar with that game, however the comments on the video explain that MB does have Progressive Gravity like Guilty Gear, so it’s not an infinite, it might just be a loop. Apparently the deal with corner juggles is, you want to keep them fairly low so the moves will connect, and at the end he launches the guy for a finisher air combo, probably because gravity was getting so high that he couldn’t continue the loop anymore.


  2. pingas May 6, 2017 / 8:55 pm

    Blood Dragon had level design?

    Always found your praise for that game to be quite strange since I remember it being one of the most boring and repetitive FPSes I’ve ever played (and I stuck to the main story instead of trying to clear the copypasted outposts and side missions)


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s