It's that time again: to prepare for the big GDC 2018 Level Design Workshop on Tuesday, March 20th, a handful of the speakers wanted to warm up by chatting (via email) about some of the finer points of level design.
Gathering question from the community, they shared some interesting insights, experiences, and learnings in a shared document that's reprinted below for your reading pleasure.
The participants in this exchange, in no particular order, are:
Robert Yang, NYU video game professor and indie game maker
Blake Rebouche, senior quest designer at Guerrilla Games
Nina Freeman, level designer at Fullbright
Steve Gaynor, game designer and Fullbright cofounder
Heather Robertson, indie game maker
Mike Bithell, game designer and Bithell Games founder
Christopher Totten, game designer and founder of Pie for Breakfast Studios
David Shaver, game designer at Naughty Dog
Nathan Fouts, game designer and founder of Mommy's Best Games
[more detailed bios for everyone are at the end of the piece.]
Each participant was encouraged to respond to every question they felt comfortable answering, so you'll notice that the variety of answers will vary based on the query.
What tricks do you have to lead the player towards a goal?
This is always a big challenge, regardless of the genre of the game, there's a lot of cop-out's such as obnoxious goal HUD markers, but there's surely also a lot of different ways LD's can mitigate this, even solve it.
Robert Yang: I like using enclosed stairwells. As the player climbs upward, you’re getting them to look up -- and the opening at the top also frames the view for them. I also like the old Valve trick of using sudden flocks of birds to direct the eye, but then you need a bird system in your game?... There’s also two attitudes you can take: (1) it’s OK for the player to get lost sometimes, (2) players don’t actually mind obnoxious HUD markers, even if us developers think it’s artless.
Andrew Yoder: For multiplayer maps, I want the goal and the most mechanically engaging gameplay to be in the same space. When wide flank routes offer the best tactical options, gameplay can become diluted across the map, which creates “where’s my team?!” moments for anyone playing the objective. Also, because the goal in multiplayer games often shifts per situation, it helps to let players change paths and regroup with other players without having to backtrack (this is one reason why long hallways can feel bad).
Chris Totten: Designers should reach into their visual art knowledge for ways to draw players through a space. Contrast is one of the best things to use. One example I demonstrated for students the other day was in Bioshock: Infinite: the designers used contrasting colors to indicate the path. The scenes in the beginning are predominantly blue/purple in lighting, but important points of interest are orange/yellow in color, which contrasts and draws players through the space. Other things to contrast include lighting (humans travel towards light instinctively), level geometry (small path leads to big space or tall object), or material/texture (indications of man-made objects in a natural area or vice-versa.)
David Shaver: It turns out, this is exactly what my GDC talk is about this year! I don’t want to spoil the talk too much, but here is a super quick version of some important tips. Combining several of these together gets great results.
Landmarks - Big, iconic objects in the world that orient the player and are often the end goal.
Lighting - Darken everything around the goal and put lights on it. People are drawn to the light.
Color - Pick a guide color that pops out from your environment color. Games like Uncharted and The Last of Us use the color yellow to let you know you’re on the right path. Yellow handholds, yellow flapping caution tape, yellow pipes, etc.
Shapes - Spiky shapes repel while round and square shapes provide safety and stability and attract.
Affordances - Affordances tell the player “hey, you can go here and interact with this thing” just by looking at them. They attract people.
Movement and Sound - Flying birds, flapping ribbon, fire, sparks, a banging door. All of these attract attention and can guide the eye where you want it to go.
Enemies & Buddies - NPC buddies can lead the way or look at an important object. People tend to follow enemies (or avoid, depending on the game type), so you can use them as breadcrumbs.
Breadcrumbs - Enemies to kill, powerups, health packs, collectibles. Whatever is appropriate for your game to breadcrumb the player through the level.
Heather Robertson: There’s a concept in theme park design called “weenies” -- i.e. large landmarks in the environment to which you make sure to give near-constant line of sight. This gives players a direct goal and a constant reference point with which to orient themselves. If that fails, there’s no shame in just putting signs with big arrows on them around your level. Players will get the idea, and it’s much better than leaving them lost in my opinion.
Nathan Fouts: In our current game, Pig Eat Ball, it’s a top-down 2D action-adventure game. The overworld sections are many screens wide and high. In 2D, we don’t get the benefit of large landmarks that are always visible by simply looking up and around as in 3D.
Instead, for Pig Eat Ball at least, we tried “pathway flooring”. The overworlds are open and the player can go many places to explore, but for the “critical path” parts, we use different floors that make a literal path to the next, necessary section.
Yet after years of testing, we found for some players that wasn’t enough! They’d still get lost. So we simply went for it--Every 5 seconds a tasteful HUD arrow fades in, pointing the way for a few seconds, and then fades out.
It helped many players since, in testing, and hopefully isn’t too obnoxious to players who just want to explore for a while.
How would you factor in player choice at a more fundamental level inside the game world? Would you agree that building a world that can hold many ld stories, as opposed to having one story and building the world around it, is a more advantageous way to design levels?
David: In order to factor in more player choice, you can include a variety of game mechanics into the level design and have multiple paths that let the player discover and use them. Dishonored does a fantastic job with this. You could stealth around on the ground, teleport to higher vantage points, sneak through small crevices, go in guns blazing, etc. The levels are like sandboxes that let the player accomplish the goal however they want with whatever tools (mechanics) they think to use.
I don’t really agree that building a world that holds many Id stories is inherently better, though. It all depends on the kind of game you’re making and the kind of tone you’re going for in that level. For example, if you have a bombastic action sequence where you are trying to escape a deadly helicopter, presenting choice can cause people to pause or go the wrong way which kills the action and pacing.
How do you balance gameplay vs aesthetic goals?
Chris: Rather than think of them as distinct from one another, I try to think how my aesthetics help accomplish gameplay goals. Think of environment art as the words or characters of a language and level design as the conversation you’re having with the player in the language. If I want players to go to a specific place or do a specific thing, I make sure the environment art or other visual elements that I use (lighting, geometry, textures) communicate to the player, “go here” or “do this.”
David: As a designer, gameplay often comes first over aesthetics. However, aesthetics are a big part of what makes games so amazing and must be considered too. The key to balancing the two comes down to good collaboration between art and design. It’s the artists’ jobs to make the game look as stunning, and it’s the designers’ jobs to ensure the gameplay is fun. Through close collaboration and a little compromise, we can make sure that aesthetics look amazing but still reinforce the gameplay so everybody is happy.
Heather: The two are absolutely intertwined; better aesthetics nearly always translates to better gameplay through readability and encouraging exploration.
Nathan: Like Heather said, they are deeply connected. For me, they actually feed inspiration. I’ll think “I need a block that damages the player.” I’ll then theme the visuals to the setting. In the case of Pig Eat Ball, it’s sci-fi, so I decided for some striking, red-crystal spikes. Later on, because of the visuals of the hard, but brittle crystals, we decided to make some bombs, and certain other attacks be able to destroy the crystals. Now the player gets even more mechanical options--it’s fun!
My method being 1. Gameplay. 2. Add aesthetics. 3. Be open to new gameplay inspired from aesthetics.
What do you think is the best method for a design test? Is it documentation or a prototype? How much work do you think is needed from candidates?
Andrew: a test greybox level is nice to have, since it shows a candidate’s technical and design abilities. However, because it’s easy to evaluate these aspects, they often get more weight than hard-to-test soft skills like how well a candidate will collaborate with their teammates, or how they will compliment the culture. The technical and design skills are easier to teach than the soft skills, which ought to affect our priorities when interviewing.
David: It really depends on the type of design job. These are just my personal opinions, but I feel for all candidates, start with a small written test asking general design questions to get a feel for your experience and design sensibilities. Then, we get down to specializations for the job in question.
Systems Designer? We need a system to do X in our game. Design and balance a system to do X.
Gameplay Scripter? Here are some gameplay sequences we want in our game. Write some pseudocode (or actual script in language you choose) to implement them. Also, basic 2D & 3D math problems.
Level Designer? Here’s the specs for a level in our game. Build a blockmesh layout in any 3D software (Maya, Unity ProBuilder, Unreal Engine, Google Sketchup, etc) Paper top-down maps are not enough - I need to know you can actually build it.
The key is to have the candidate do similar work to what they would be doing daily at the job because if they can do a bit of sample work in the test, it’s a good indicator they can do the job. As for how much work is needed? Just enough to show you can do the job well. The tests that take weeks can be frustrating. I don’t like wasting people’s time so something small that can be done in a day or so is my favorite.
How do the pros handle lock & key progression in an open world game? Would love more tips and insight on that subject.
Nathan: In our game Pig Eat Ball, we have “critical path” required levels, and then there’s open-world exploration sections that are not required but look enticing. The required action levels teach the player certain mechanics (how to break a certain block, how to get past a certain obstacle).
The open overworlds look big, exciting, and are filled with unusual objects. But as the player makes their way through the action levels, learning how those strange objects work, and then come back to the overworlds, the player sees these gameplay objects in a new light, and how to use them and get past them.
Top 3 levels or games to play and experience -- what is level design at its prime?
Robert: The old Quake 3 GeoComp2 levels are still very dear to my heart. I’m also a big fan of island mansion levels, like The House of the Widow Moira in Thief 3 or the Addermire Institute in Dishonored 2, which have amazing floorplans that feel kind of like complete places.
Andrew: If you’re into multiplayer design, learning how to play Halo CE’s “Damnation” in competitive 2v2s will make you rethink a lot of assumptions about best practices. If you’re interested in narrative-focused or expressionistic level design, Psychonaut’s “Black Velvetopia” is a must-play. Or for mechanics-focused singleplayer level design, I love Quake’s E1M6 for its tight puzzlebox design.
Super Metroid because of how the game nearly-wordlessly teaches you how to use each new weapon or item with level design.
The Orange Box (okay that’s cheating, it’s not just 1 game) for the visual systems of communication apparent in those games and the developer commentary option that breaks them down.
DOOM (old school or Doom 2016) for how to create worlds with linear progression out of non-linear/looping spaces.
The first area of the first level of Super Mario Bros. because it teaches you everything about Super Mario Bros. in one screen with no dialog.
The intro level of Mega Man X for the same reasons while also providing a fast-paced-feeling but very beatable introduction that makes you want to play more.
The maze from Pac-Man as an easy demonstration of how spaces should loop back on one another for ease of navigation and to help players dodge enemies (applied throughout the history of level design in games from Resident Evil to DOOM.)
Thief: The Dark Project has fascinating level design, which balances the mechanics of stealth to create environments which feel lived-in (finding silver candlesticks in dining rooms and food in kitchens) while also using the levels as an opportunity for characterization (the player character is less visible in torchlight than gas lamps, and walks more softly over carpet and wood than marble, making it clear which environments the character (and, in conjunction, the player) finds themself more comfortable in).
The Sexy Brutale does a great job with level design, creating areas of shifting levels of danger that you must understand to solve puzzles while constantly creating new sources of surprise.
For procedural generation, you can’t go wrong with Spelunky -- the systems in that game create areas that always feel tense to navigate, like if anything goes wrong you’ll be dead within seconds but if everything goes right you may find yourself soaring to new heights.
Seen lot of tutorials on how to approach designing levels for FPS, but is there anything specific one should look at when creating open-world/RPG levels?
Robert: In general I think open world games have more diverse demands than the typical FPS -- like, is the open world game about walking (a la Skyrim) or driving / flying (Just Cause) etc -- which make generalizing more difficult. But the one thing that comes to mind is Matt Walker’s Twitter thread on Breath of the Wild composition, about using clusters of “triangles” in the landscape. (I also hate reading Twitter threads for stuff like this, so I did my own write-up: http://www.blog.radiator.debacle.us/2017/10/open-world-level-design-spatial.html)
Chris: All open world designers should read Kevin Lynch’s Image of the City, where he breaks down navigation through a city into nodes, districts, landmarks, edges, and paths that let occupants know where they are and an idea of where to go next. These can exist in open world of any visual style - not just ones that are cities. You want players to be able to understand where they are at any time and provide a visual language that helps them understand what they are looking at (i.e. is this building a random house, a place where I can heal, or the entrance to a dungeon?) or where to go (tall tower viewable from distance likely = important.)
How would you design an open space that draws the players attention to landmarks in the context of environmental storytelling?
Chris: When I want to draw player attention to a specific place on a map, I try to think not only about the environment art used to make that destination distinct, but also how environment art around the destination supports it and guides the player’s eye. If I want a player to go to a tall mountain, for example, I would frame it with shorter mountains or objects that get taller as the player’s eye navigates to the mountain I want them to notice. I’d also maybe use color to emphasize the mountain against the rest of the environment (sunset behind the mountain, fire coming out of it, light emanating from it, snow on top when others have none, etc.) Think of it as a relay race: each time the player’s gaze reaches an interesting object, their eyes will run to the next more interesting object until they hit the visual “finish line.”
Heather: Players are much more likely to notice motion in an environment than things that are still. This can be achieved in a number of ways, from pumping machinery above a door to a quiet stream of fireflies floating across a field. You can use continual motion, specifically that line of motion, to draw attention to more still things in the environment.
What is your process for designing and blocking out level layouts?
Robert: Start with drawing a simple sketch of your layout. You don’t have to follow the plan exactly, but it’s good to have a rough idea of your space. After the sketch, open your level editor and add a floor plane and a human scale reference figure to help you scale your level. Then, start adding walls, etc. Rinse and repeat until you have a substantial chunk, and then test as soon as possible to check distances, clearances, etc.
Chris: I usually start by thinking of what my level is “about” gameplay-wise. Unless it’s later in a game, it’s not a good idea to make your level about every mechanic in your game, but maybe one or two. I try to introduce a mechanic first in a safe way, then give some basic challenges. Afterward I like to find a way to twist the mechanic by putting it in a new context (wall jump, but now with RISING LAVA!) or by combining it with a mechanic that was established earlier in the game (platforms that are turned on and off with switches, but now they block buckets of water that cool the lava and create a safe passage.) In terms of tools, I first try to sketch out my ideas on graph paper, then I gray box them in my game and playtest playtest playtest. I don’t add final artwork until later unless it somehow aids the player’s navigation.
David: First, I gather the requirements for the level - things like: location, time of day, weather conditions, enemies available, buddies available, weapons available, traversal mechanics available, story and character tones, how far this level in the game, etc. Once I have that, I look for reference photos online and put them in a folder. In addition to Google Images, Pinterest can be pretty great if you search for something and add “concept art” to the search phrase. I might play similar games or watch playthrough videos on Youtube to see what others have made so I can avoid repeating it, and to spark ideas. I’ll make a list of properly paced major beats in the level and then I’ll start building a basic blockmesh level in Maya that includes those beats. Then it’s a big loop of iterating on the blockmesh, playing it myself, playtesting it with someone else, and making changes until satisfied or I run out of time. For the blockmesh process itself, quickly block out the entire level’s space very roughly to your game’s scale and then refine it like a sculpter would chisel away at a block of marble. You’ll get a sense of what areas are too big and small and fix them. Building something and playing it as soon as possible is more important than a paper layout because everything will change as soon as you play it and you’ll find the fun as you go.
Heather: I’m a very spatial thinker, so I tend to start by constructing my levels out of cubes in a test scene, iterating on that structure for a while, and then moving over and creating the same level in the proper scene.
What level design considerations and solutions do you have for the camera perspectives that your games are viewed from and the rules that govern them?
Andrew: In third person games, its best practice to design with space for the camera. If the player avatar back into a solid corner, the camera needs to have a solution, or the level designer should avoid those situations by pulling the walls back from the walkable space. There are also kinds of feedback that players perceive differently based on the camera. In first person, it is hard to have enough feedback for stuns and other crowd control abilities that players don't mistake it for a broken input device. Similarly, ground hazards like pools of acid are hard to notice and frustrating to navigate in first person games.
Chris: Once you’ve set up your camera, make a simple test level to figure out what the limits of your ability to see players or gameplay-important things in the world through the camera. If you, for example, are using an isometric camera and can’t see players when they go behind skyscrapers or under bridges, you either need to learn to not use those things or create a script that allows players to see beyond them (fading them out or silhouetting the player when they go behind things.) Tests help you establish a set of design guidelines (that you should write into a document shared with all of your designers) for all of your levels and/or a list of needed scripts or tools to alleviate problems.
Also, think about what artistic powers certain perspectives give: 2D side-scrollers, for example, give you the power to show the player things that in-game characters don’t see. It’s like the old Hitchcock example of showing the viewer the bomb under the table in a scene where two people are having a conversation over lunch. Metroid: Fusion similarly shows the player a monster hunting Samus that she cannot logically see, creating great tension.
David: In first person, many players don’t look up much so you need to do things to get them to look up. Angling the ground so you are facing the sky as you walk, drawing the eye upwards with a flying robot, walls with literal up arrows painted on the wall, etc. Also, I’ve seen several playtesters never want to turn around so avoid 180 degree turns when possible.
Are there any tools for level design out there other than custom ones for each game? Is there a repository for design philosophies?
Chris: There are lots of great middleware solutions for level design (think the ProBuilder plug-in for Unity.) Editors take on whatever specialization their engine has, so if an engine is built for an FPS, the editor probably is too unless you go in and completely rewrite it. Unity, Unreal, and 2D editors like Game Maker or Construct can each make lots of different types of gam