Designing Radically Non-Linear Single Player Levels

[Music] all right hi everyone i'm aubrey sir i've been doing level design for 10 years these days i run a small indie game company called 12 east games where i still do level design but i'm better known for my previous work at wolfire where we organized the first humble bundle and made games like overgrowth i also collaborate with other indies making art and levels my focus is on a specific type of level which is why i'm here to talk to you at gdc today i'm going to share with you my favorite kind of levels to design which i call radically non-linear levels my talk has three sections first i'll define exactly what i mean by radically non-linear levels how they're based around the idea of free movement and sandbox play and activating the space second i'll describe the basics on how to build them like layout options in typical dynamics and finally i'll explain some techniques i use to fix common issues that may come up in radically non-linear maps like the map feeling empty or not having enough structure then i'll wrap it up with a recap and final example but before that i want to mention an idea i think we as level designers probably agree with that idea is that all conventional levels feature some openness and some linearity in other words all levels have places you can go and places that are blocked these are some of our most fundamental level design tools paths and walls linear levels start with making paths often even when we talk about nonlinear levels what's described are levels with branching paths these kinds of levels can be great but an alternative kind of level is based on the opposite idea instead of building paths you start with an area of movable space and place walls to block the player the team behind hitman calls these types of levels non-linear sandboxes a radically non-linear level is a sandbox level that is designed to encourage unpredictable player movement and exploration of the space all sorts of single-player levels fit this paradigm which i'll give some examples of in a moment and there's also a lot of stuff in between even highly linear levels have open areas in them and there's an obvious overlap between this kind of level and most multiplayer arenas so i think these techniques apply to any space the player's free to move in so i hope some of these ideas can help you in whatever kind of level you might be working on despite that overlap i think this type of level is a distinct is distinct and interesting i didn't always feel that way though let me tell you about my first commercial level design job it was for a game called darkest of days i was hired as an artist but was immediately put to work on level design outside of a few excitebike tracks i made when i was eight i had almost no level design experience in fact this was the first major game project for most of the team we had our own custom engine and tools but they were lacking some basic functions for example the map editor had no undo function so in that way it was a lot like this talk i had to try to do everything perfectly the first try on the one hand it's nice to get paid to learn on the other hand it's a lot of pressure to care about your work and not know how to do it but don't worry it turned out okay as we were making these maps our lead engineer jeff kept mentioning that he thought his test maps were more fun than what we were making we were constantly running up against the problem we couldn't reconcile we wanted to make a cinematic game like call of duty modern warfare but we also wanted large dynamic battles and open spaces trying to solve this problem is how i became interested in the idea of radically non-linear maps worked on a lot of games with them and ultimately how they became my favorite kind of map with that out of the way here's part one i'll define exactly what i mean by radically nonlinear levels when i first started researching this question i wanted to find a really fundamental example of what i meant that everybody would be familiar with and the oldest clearest example i found was pac-man but we're going to talk about miss pac-man because it's a better game so why miss pac-man the map is open enough to offer free-form tactical choices like which path to take to avoid ghosts and large enough for strategic choices if you're gonna eat a power pellet or go for the bonus fruit the designer of miss pac-man mentioned that each ghost has a different behavior and in this way each ghost is a unique risk factor which the player player needs to consider and there's some other interesting things about miss pacman maps it's small enough so that the whole map is in play and i'm going to touch on why this is important later all the systems are legible in some sense complex enough to be interesting but transparent enough to be predictable there's no exit this is a common thing for this kind of map even in uh radically even if a radically non-linear map has an exit the player often ignores it until other objectives are met in that same way the starting point isn't very important either because you don't know how the player is going to move through the map you have very few tools for pacing in linear storytelling miss pac-man handles tension with dynamic systems where timing is important power pellets ghost responds these kinds of dynamic systems are also a reoccurring feature of these maps at this point you might be thinking hey aubry do you just mean open world maps there are a few big differences between this style of level and an open world before i mention the whole map being in play what if it wasn't if we imagine a miss pac-man level that stretches on indefinitely you can just run from the ghosts you can never eat all the pellets and the power pellets become an unlimited resource in other words it would have no set challenge and totally different gameplay another question is are these kinds of maps just the same as multiplayer competitive maps maps played with bots the answer to that is sort of they are similar but multiplayer competitive maps have a lot of additional design and gameplay constraints like symmetrical balance leg tolerance limited physics that change how you make the map and so they're really very different in practice if you feel familiar with mda theory the idea that any game can be categorized by these qualities i think you can see that this kind of level occupies a distinct mechanics dynamics aesthetic niche apart from multiplayer arenas miss pac-man was made in 1982 that's a long time ago is it even still relevant to modern level design here's why i think so if you look at doom 2016 you see that many of the levels have linear sections punctuated by non-linear arenas and last year's talk embracing push forward combat in doom they point out that these spaces were a byproduct of two of their core gameplay loop goals constant movement and improvisation of in navigation this causes 360 degree combat and resulted in maps that they felt had more in common with multiplayer levels than traditional linear levels doom 2016 exam levels are a great example of a hybrid approach and it's part of a cycle away from shooting gallery style design that peaked around 2008 and a return to less linear design like the original doom and i think this trend still has a lot of room to continue people are innovating in this area every year which is one of the reasons i wanted to give this talk as i was saying with the doom series and pac-man these kinds of levels have a long history including games like attention battle zone 98 and there's a lot of modern examples too like metal gear solid ground zeroes or hitman where they tried more linear levels in absolution and backed away from them with the more recent hitman games which we heard a little bit about earlier today the design techniques i'm going to talk about later can also be applied to games that use pre-designed tiles for randomized levels like spelunky which i'll show a little bit in my final example so what makes radically non-linear maps cool why do i like building and playing them there are a few reasons one of the things you notice about the games i've mentioned is that they emphasize gameplay over story remember how i said our programmer jeff preferred his test level it's because a more open map allows the gameplay systems more room to breathe some types of gameplay that really benefit from this are long range engagements using vehicles and stealth and gameplay based around controlling enemy movement this gives players a lot of choice with more agency to make strategic and tactical decisions map knowledge is important in this kind of level and so exploration is naturally rewarded because there's no need for fail-safe funneling of the player destructible terrain or killing any character become options they also emphasize dynamic and environmental storytelling because of fewer artificial barriers it's easier to feature realistic locations in architecture and as i was mentioning before aspects of radically non-linear design can be implemented in all sorts of maps that's a lot of stuff that i think makes these kinds of maps cool but the most important thing isn't it just fun to scout around come up with a plan and test it i love the kind of level that lets me come up with different solutions and that i can play twice with totally different outcomes it's just water part two here are some techniques on how to build these kinds of levels i'm going to borrow and adapt some theory from randy smith's 2006 gdc talk level building for stealth gameplay let me take a moment to explain this diagram here this is an overhead view of a map from thief and we see how smith used these safe areas and rewards shown with plus signs and enemies and exposure shown with negative signs to predict and control the player's path i want to use them in the opposite way so that it becomes impossible to predict the player's path this is what makes our level radically non-linear and also how we activate the entire space of the level for play the first thing i do when i start on a new map is to have a clear idea about the layout by picking a strong layout you'll be able to differentiate each map you make rather than ending up with a random hodgepodge of obstacles enemies and rewards there are a lot of different possibilities for layout but here are four solid solid examples symmetric we have a low tension center line with a left right choice this is an interesting layout because the player chooses which side to tackle first and then halfway through comes back to the center line that helps break up the tempo and creates pacing this is a great layout when the theme of the map is a major path like a highway concentric the outer area of the map is safer and rewards extensive scouting before committing to the center this layout works well for maps about assaulting a base or castle node these are a series of clearings that interconnect in a lot of ways this is the kind of this kind of map keeps up a high tempo of quickly alternating encounters with travel the player needs to resolve challenges carefully so that they don't bleed into other adjacent nodes in unpredictable ways this layout fits with ruins or slums field a large area of increasing risk this layout promotes alternating probing attacks and retreating and gives the player a lot of control over tempo this is a typical layout for something like a beach assault again the main point here isn't to know what the player will do but just to make sure that whatever they do it isn't boring one thing to note here is that it's also possible to use gameplay systems for the layout rather than building it into the map and i'll have an example of that using a field layout at the end of the talk as well as layout i like to have a clear grasp of how the gameplay is going to be unique for the level one method is to try to find an aspect of gameplay that's underutilized and focus on it this is hard to generalize since every game is different but here are some concrete examples from some maps i worked on this is brigadore an isometric shooter that was designed from the start to allow players to use a wide range of vehicles because of this it had some interesting design constraints for example within the interior of the map all the terrain is destructible in this game certain aspects are randomized which is one of the easiest ways to ensure players try different strategies on replated repeated playthroughs here the enemy spawns are hand placed and whether an enemy actually appears depends on several factors similarly the player will randomly enter the map from one of these two locations i think this technique of using hand play spots for enemies and rewards that are then controlled by systemic factors is really useful the gameplay idea for this map is that the center area is a trap it's tempting but if you go in there before picking off most of the enemies you'll probably fail as you can see from the main travel lines this map is roughly concentric the main hazards and rewards are concentrated in the middle to draw the players into the trap it's possible to clear the map by defeating enough enemies around the border so you don't need to go in the center but it's a lot faster to beat the map if you complete these central objectives in that way the concept of the map that it's a trap to lure the player in and the layout of the map that it's concentric relate to and support each other here's another example from brigadore the game has soft cover that blocks line of sight but doesn't offer much protection in the existing maps it was mostly combined with hardcover to add variety but i wanted to replace the hard cover with mud pits to give a really dangerous exposed feeling to this map this layout is one i haven't shown yet where the objectives are in the four corners this means the player has to cover a lot of distance which emphasizes even more how exposed the player is i think this shows how it's important to pick a concept and a layout that work well together here's a different game i worked on called raven field a single player game where you capture control points and shoot enemies in an effort to outscore the opposing army unlike similar multiplayer games the levels in raven field don't actually have to be balanced or fair on this map i was motivated by my frustration of the ai allies taking all the airplanes on the other maps so i made a map with a lot of airplanes in it during production i bumped the starting bases out out to the corners of the map to maximize playable space for the fast-moving airplanes but kept the roughly symmetrical layout unlike my previous example of this kind of layout this is an inversion where the center line is more risky and these side areas are a bit safer i think this works because in the aircraft you have to keep moving and naturally end up flying over the center line all of these examples benefited a lot from being for functional games with several existing maps in cases like this it's pretty easy to play with existing systems make tests and figure out something that works if however your game isn't very finished you may notice your map has serious problems and not know why okay welcome to the third and final part of this talk how to fix common issues or why is your map all busted the number one issue is that the map is too linear it's very tempting to want to guide the player and reflexively fall back on that skill set but if you're going to design a linear map or have linear map sections it's important to go all in on good pacing and have all of those existing techniques that make that style of map work otherwise open up more paths and space to move similar to the above but with gameplay you have one idea in mind with how the how to win the map this can stifle organic and robust gameplay systems it's important to avoid trying to restrict the gameplay too much trust that the core game plays fun on its own another possible issue is the map has no structure the map is too uniform so you end up doing the same things over and over hopefully by having a unique layout and concept you avoid this issue in a game with very dynamic systems sometimes the player can feel helpless the level should be balanced so it's reactive to the player and they act as the tipping point for the gameplay systems the gameplay can feel random or non-existent the player isn't able to see the cause and effect of the systems enough to understand why things happen for example using an invisible trigger to spawn enemies on the map is a lot worse than an alarm state in a stealth game alarm state gameplay still has triggers and still spawns enemies but gives the player a lot more information if you want to learn more about making systems legible to new players i very much suggest watching the gdc 2016 talk titled level design and hitman guiding players in a non-linear sandbox these previous issues can usually be a fixed assuming you have mechanical complexity to work with but if you don't then it may be something to bring up with your design team the sorts of systems these games usually have are things like stealth ai companions inventory looting destructible terrain dynamic enemy movement player tools for controlling the enemy sub-objectives like capturing areas or destroying enemy resources they have pacing systems and various movement modes like vehicles etc when discussing how these systems work having a map that feels like it's missing something to show the rest of the team can be very useful on the other hand if what seems to be missing is grand cinematic moments type pacing are very well defined challenges i'm not aware of any way to fix this yet it's part of the trade-off that comes with this kind of map which may not be the best choice for your project it isn't that dramatic scenes or clear challenges can't happen but they tend to result from the existing systems and so they're hard to predict i don't think this is an impossible problem though because there's been a lot of exciting progress in dynamic storytelling lately before i end my talk i just want to recap with an example that brings it all together and shows how a lot of this theory can work in a more linear game the game i'll use for this final example is one that i was in charge of level design on called receiver it has some aspects of linear design but still allows free and unpredictable movement the level was made from tiles each tile is roughly the size of a small office building and has three floors with doors on each side these act as lanes when these tiles are put end to end the lanes can combine or switch positions if i had just stuck these together i would say it was similar to a multi-path level kind of like this x but we did something a little bit different instead the tiles are randomly placed so the player often has multiple choices about how to progress in addition to the strategic choice of which lane to take the tactical choice of the best path through the tile also isn't predetermined either because the enemies and objectives are spawned randomly from hand placed spawn points this is how the space is activated even though almost all this content was pre-authored and the random elements are fairly minor it's still enough to change the ideal strategy and to incentivize exploring the space which makes the level radically non-linear because of these systems it made it difficult to support a standard narrative instead we made the objective to find these audio logs which can play in any order obviously this doesn't result in the most traditional story but i still think it added to the experience that said the story wasn't the strongest aspect of this game that this was part of the trade-off we decided to make another gameplay system at work here is that each new tile is harder than the last one this gives the player an incentive to check all the lanes in a given tile which may require some backtracking as i mentioned when i was talking about miss pacman a systems-based approach to tension is an important pacing tool and i think this could have been boring in some games but because receiver is extremely high tension back tracking serves as a much needed break from clearing new areas this increasing difficulty as you move forward is also provides a layout structure i mentioned earlier called field because these levels don't have a lot of scripted events or built in gameplay they may seem a bit simplistic but part of making a map in this style is to not obstruct the core gameplay in receiver most of the gameplay is built around a real realistic simulation of a gl of a gun including all the working parts and buttons we surfaced all this gameplay for novice players with a context-sensitive help system this makes it clear to the players that every bullet is precious and that every enemy encounter is difficult simply because firing your gun is is difficult this relates directly to the levels because you can scavenge bullets and obviously cover and enemy positions matter more when the situations are so lethal i think these levels do a good job of giving the player interesting choices while staying out of the way enough to let the core gameplay shine by working closely with your gameplay coder and because the core gameplay was so deep my job was made a lot easier and i was able to avoid any really tricky problems part of that was making trade-offs early in development to avoid those kinds of problems in the end focusing on interesting dynamics that shift rewards and hazards keeping layout and pacing in mind and giving the gameplay room to breathe really helped receivers radically non-linear levels come together finally i would like to say i feel like this is a complicated topic and because these maps integrate so closely to the gameplay they're a bit slippery to nail down but i think they really are a distinct and interesting style of map with their own history and design trends hopefully i've provided some concrete ideas on building a map like this and how to approach certain problems whether your future levels are radically non-linear or not thank you for coming to my talk and uh it looks like we have time for plenty of questions uh if you have any questions you can go to the mics hello and thank you for the talk oh thanks so you mentioned that it is a constant challenge being able to nail the level design just right before shipping the game now the industry has changed over the years and now we have tools to measure all sorts of things what's your approach to nailing down the the interesting part uh up to now what i've mostly done is just have play tests do early builds and just get player feedback watch people stream the game i i personally like more than trying to collect metrics through the game i like to um watch live play tests where i can see the player's face so you can get a kind of a more immediate read on what they're reacting to thank you hey what do you think of the dangers and benefits of using procedural generation with these kind of levels uh i think it should be kept to a minimum like it's a very useful tool i know that there's been recently a lot of like kind of negativity towards procedural generation and i don't think that's really makes much sense but at the same time sometimes you just want the level to play a certain way and by introducing just a tiny bit of random elements you can massively expand the playability of a map so yeah i try to use it sparingly um so as an educator i'm brady university in the netherlands um we have a lot of room to do multiple level design projects in our curriculum but one of the things i have to kind of figure out is uh how we build complexity for the students so like where we start simple and then build up um and some parts of that are easy like single player platform game level design is a nice starting point but do you think the uh linearity and non-linearity is that more difficult or i think that especially now making a really compelling linear experience is like extremely difficult so um and it's mostly different disciplines like more about storytelling environmental narrative my last game i made was totally linear totally totally linear totally linear like adventure game so i'm well acquainted with those problems and yeah i i think they're both really difficult just different so uh non-linearity is more like a sideways shape yeah just like a different a different set of skills okay cool hello so thank you for the talk um i wanted to know you mentioned at the start that the making of the levels is like making sure what space is blocked and making when what space is open but in this kind of non-linear level where you opened up a lot of this space what let's say one problem maybe if the player has a lot of freedom of movement no there's not necessarily that they can fly but let's say you know i know they can climb walls or something like that how can you make let's say obstacles or walls that actually are obstacles without impeding the core gameplay in that way so uh yeah i one of the games that i didn't mention because i had a lot of trouble with this exact problem was overgrowth because your character is this rabbit and he can jump like 30 feet in the air and run 40 miles per hour and you're you're just like flying past everything at top speed and going anywhere you want and yeah this is a really difficult problem and i i would you know i think ultimately you have to have those kinds of movement options cost something at least a little bit of something so that you can like have some gradient in the player behavior or else they are just going to be lying to the objective and jump over everything because their character is superman okay thank you thanks caicho escudigasco thank you for the talk i recognize this maybe a little bit outside of your field of expertise but i'm curious about the overlaps with vr projects in terms of linear and nonlinear design yeah uh there is a vr version of um receiver and i've but i've only played it a little bit um and i'm not sure i think it's it it basically works like i don't think it's like massively different uh it's maybe a little bit more confusing um when you can't like whip your head around and get for some people maybe less confusing for others i don't i'm not sure but i haven't done extensive player testing with that kind of thing so yeah thank you anyone else thanks for coming to my talk

Channel: GDC

Views: 146,133

Keywords: gdc, talk, panel, game, games, gaming, development, hd, design, level design, single player

Id: CTBor4rhnQs

Length: 27min 40sec (1660 seconds)

Published: Wed Jul 14 2021