Originally presented @ Level Design in a Day, GDC 2013 Co-Authored and Co-Presented w/Nate Purkeypile Slides available here. This blog is a repost of the original text on my site.
While many developers understand the basic concepts behind a modular system, and some have dabbled in it for a project or two, very few have made a career out of it. That’s exactly what Nate and I have done, however. Our current project will be our fifth together in the near-decade we have known and worked with each other. Every one of these projects has taken this kit-based approach, including those we worked on before joining Bethesda.
We recently realized that there are many unspoken understandings we and our colleagues share; expectations and assumptions taken for granted as part of our process. We've struggled to explain ourselves in detail for new artists and designers who had joined our team. This talk was a way for us to explore and articulate this accumulated knowledge. To understand our approach, it’s useful to know where we’re coming from as developers, and where Bethesda Game Studios is coming from as a whole.
Let's begin with a pretty simple observation: our games are big. If you've played or read about Skyrim, Fallout 3 or any of our other open-world games, scope is a big part of them. It’s an integral part of our games and our DNA as a studio. You couldn't boil Skyrim down to a six-hour game and expect to provide the same experience. Scope isn't a random attribute of our games; it’s a major feature.
When you play a Bethesda game (or when you work on one) you know this going in. The games are big! This influences everything we do at the studio. Production methodologies, tech and tools, workflow, pipeline - it’s all informed by a need for efficiency. We need high bang-for-buck on everything we do. Games like Skyrim take a few years to make, even moving as efficiently as possible. With a game that big, missteps aren’t measured in hours or days, but scale to weeks and months.
These core values, along with the circumstances of team size, and our collective experience making games together all adds up to what you’d commonly refer to as studio culture. Even if your team or your game isn’t driven by the same forces as ours, chances are you could take a lesson from us, just as we could probably take a lesson from you.
With that in mind, consider this a case study. The modular approach to level design we’re analyzing here is just one manifestation of the BGS studio culture.
Before getting too far into things, let's me define what a “kit” is. Kits, first and foremost, are systems. A basic pipe kit, like the one from Fallout 3, may only be four simple pieces of art which can be used together. This kit, as most, snaps together using a grid system. The most important attribute of a kit is that it adds up to far more than the sum of its parts. This artist hasn't just given us four pieces, but a system with which we can create infinite configurations of pipes, all on the editor side.
Many sprite-based, 2D games make heavy use of similar ideas and techniques to those we’ll be discussing today. Sprite sheet or texture atlas technology lies at the heart of many 2D game systems, so it’s easy to see how those games lend themselves to repeatable tiles of a uniform size. Thanks to this, it’s not necessary to paint the entire world by hand. Designers and artists can use art systems of repeated tiles to draw out the map quickly and with less of memory footprint.
Back in November of 2002, Lee Perry (then of Epic, now BitMonster) authored an article in Game Developer magazine which explored a modular approach to level design and world building, specifically as a way to help cope with the increased fidelity of games at that time. Bethesda has been at this for quite some time - 18 year as of this writing. Terminator: Future Shock and the Elder Scrolls 2: Daggerfall, both games which far pre-date our (Nate and Joel's) tenure at the studio, embraced a modular approach to world-building, which would become the foundation upon which much of our current thinking and technology are built.
So remember - the concepts discussed today are far more than the product of our work together over the past decade. Many influences and ideas have contributed to our current workflow, and there are many lessons yet to be learned.
As mentioned earlier, many developers understand the principles behind modular level design, but we hope to offer insights gained from our extensive exploration of this topic. To do this, we’ll go over the various benefits and drawbacks we've found over the years, with an emphasis on how to get the most out of the workflow.
The first, most self-evident benefit of working with modular art is that it’s reusable. Reusing art helps us mitigate the huge scope of our games. There’s a massive amount of world-building to be done in Skyrim, and not having to specifically create and export every fence post and doorway becomes very important to us. We try to be smart about where we spend our time and attention, reusing art where it makes sense, and investing time where we want something more unique.
Consider this list of features, which together make up the rough measure of Skyrim’s world-building needs:
- 16 sq. mile Overworld
- 5 Major Cities
- 2 Hidden Worldspaces
- 300+ Dungeons
- 140+ Points of Interest
- 37 Towns, Farms & Villages
Now consider that alongside this photograph of the Skyrim development team. There are 90 people in the photograph below. With the exception of a small number of outsourced assets, you’re looking at the entirety of the dev team across all disciplines. We've resisted the temptation to grow into a multi-studio team of hundreds, such as you often find behind games of similar scope to those we make.
Today we're focused on what you’d most typically associate with traditional level design at Bethesda; the dungeons component of that world-building work. Only ten people out of the 90 in that photograph are directly responsible for those dungeons. The rest of the team is focused on the myriad other features, content, systems and support needed to build a game like Skyrim. You can begin to appreciate how small our dev team really is when you consider how much responsibility each person needs to shoulder.
Sooner or later while reading this, you may think that the scope of the game is a restriction that limits our ability in some way, that it’s a burden we wish we could shrug off. The thing is - even if our circumstances were different, and we had infinite time, resources, or talent - I don’t think the way we make games would look all that different We've found a way that works for us.
Consider the Dutch artist Piet Mondrian, famous in particular for his work in Paris during the 1920’s. You may not know Mondrian by name, but you’re likely to recognize his work, such as the piece seen below. Mondrian had access to more than four colors of paint, and was certainly capable of more technically complex work. Yet he chose to self-impose restrictions which led him to pioneer a style he may not have otherwise discovered. The influence of that style is still felt today in art and fashion.
Look throughout the history of science, technology, music, anywhere people are creating, and you’ll find examples of restrictions not limiting creativity, but focusing it and leading people toward new discoveries and innovations. I think our workflow is like that. To change it because of a shift in circumstance would be ignoring the belief systems and experience that led us to that conclusion.
Getting back to the topic at hand, we should also acknowledge the most self-evident drawback of using modular art: the fact that resused art can become repetitive. This leads to what we commonly refer to as “art fatigue”. The median play time of Skyrim on Steam has peaked at well over 100 hours, which is a huge compliment . With that kind of time spent, however, players are bound to notice the same rock or farmhouse or tapestry used again and again. And another two dozen times after. Art fatigue sets in where this repetition becomes obvious and erodes the authenticity of the world.
We've found some ways to delay the onset of Art Fatigue. One of the big ones is simply doing away with copy and paste design as much as possible. When I first joined Bethesda, Oblivion was poised for the home stretch of its development cycle. Oblivion was of similar scope to Skyrim, yet built by a team of about half the size. One of the ways the dungeon art team coped with this disparity was to create a number of “warehouse” cells in the editor. These warehouses contained fully lit and cluttered rooms to copy, paste, and then arrange to create "new" dungeons. While efficient, this method left much to be desired, and many players rightfully called Oblivion dungeons out as being “cookie-cutter”.
One thing we noticed was that players were quicker to react negatively to repeated detail elements, as opposed to broad architectural repetition. Consider the following three screenshots, each taken from a separate Oblivion Dungeon.
In each, you’re more likely to pick up on repeated clutter first, then the repeated architecture. This is especially true in actual gameplay from a first-person perspective. To minimize needless repetition, we abolished the use of warehouse cells as they existed in Oblivion. Beginning with Fallout 3, we staffed up a group of level designers and got tool support to make sure we were able to build spaces more quickly, and with the most granular art available, reducing the amount of repetition as much as we possibly could.
You can also fight art fatigue at a more fundamental level. It’s common at the start of a project to strongly associate a particular setting with specific types of inhabitants or gameplay. You may want to only see soldiers in military bases, and zombies in crypts, for example. Resist this. Think of your kits as the architectural identity of the space, and allow other elements to establish the specific identity of any given space in which it’s used. The more you’re able to divorce these things, the more you’ll be able to mix elements up and keep the settings fresh.
This also applies to the kits themselves - try to actively encourage the notion of mix-and-matching your kits. This is sometimes called kit-bashing, kit-jamming, or a kit mashup.
Look the following examples of Dwarven dungeons from Skyrim. The first delivers on the expectation of any Dwemer dungeon in an Elder Scrolls game. There’s a chunkiness of proportion, an emphasis on brass highlights and clockwork elements which are all hallmarks of that style. Skyrim features around ten of these dungeons, however, and they’re all pretty big. If they were all the same, it’d be a dull, one-note experience, and take the thrill out of discovering one.
To try and mix things up, we started playing around with ideas like combining exterior elements. In the second example we've introduced some lightweight unique assets and enemies you’ll only encounter in this particular Dwarven dungeon. This twist on visual and gameplay expectation helps keep the setting fresh and combats art fatigue.
In the third example, the level designer simply tried bashing Dwarven hallways with ice caves. What he achieved was an instant shift in tone and atmosphere. Some artists will be uncomfortable with this idea. The notion of your art being used in unforeseen ways could lead to bad intersections or lighting issues, for example. To work on this kind of game, however, you need to take a couple of steps back and be open to this. Allow for this kind of experimentation and have it in mind when you’re creating the art to begin with. Not every combination will look good or make sense, but you can be part of a conversation about what works, what doesn't - and most importantly, what almost works, but could work great - with a little tweak from you.
Looking again towards the benefits of working modular, one of the big bonuses (especially from a production viewpoint) is that a modular approach helps if your team has a low ratio of artists to designers. Remember the ten people responsible for the dungeon content in Skyrim? Eight of them are the level designers, and only two represent the entirety of our full-time kit art team. They generated seven kits, which the level design team used to create well over 400 cells, or unique loaded interiors, of dungeon content. And those dungeons were built in about two and a half years, from start to finish.
To put this into proportion, the infographic shown here represents the kit art team as yellow, level design as orange, and the blue-gray area as a conservative estimate of the dungeon content created by that collective group.
There are two points to make here. The first and most obvious: this approach allows a small number of artists to support a much larger team of designers, who can in turn generate a lot more content than those two artists otherwise could. The example provided by Skyrim shows just how effective this approach can be, when only two artists were required to provide the core art behind so much content.
The second, less-obvious point is the more important one, however. Think of the other 80 people in that team photograph. Because such a relatively small group was able to handle the dungeon component of the game, it allowed the rest to focus on the myriad other needs of the game, whether it was landscaping the massive world, writing and scripting the many quests, contributing to character art and animation, working in the guts of our game code.
Of course, there’s another reason that Skyrim had only two full-time kit artists; kits are really complicated things to work on. Kits require not only the artistic ability to produce high quality visuals, but also a technical competency in their art tool, a deep understanding of the editor and design workflow, and so on. This unique blend of left and right brain is somewhat at odds with what many art professionals value. I've worked with great artists who make excellent kits but hate working on them - so they don't.
So when you’re trying to identify somebody with the the aptitude and interest to be a great kit artist, you’re basically looking for a unicorn. They're rare.
Another example of the problematic complexity of kits is the process of identifying and fixing bugs with them. Kits aren't like other art assets, for which you might be asked to fix a bad pivot point or texture seam, but are otherwise fire-and-forget. It’s important and necessary that a kit artist can keep the entire system of rules in her head throughout the entire project. Kit pieces are instantiated dozens or even hundreds of times throughout the game. Making an obvious fix to something like a bad pivot point may address a specific bug but create new problems in those hundreds of instances elsewhere that you may not have thought to check.
This does bring us to one of the other benefits or our approach, however, which is the very fact that kit pieces are heavily instantiated. When a change goes into a kit, those changes are instantly viewable across the whole game. As an artist, performing an export is like a fly-by deployment of new art, whether the level designers were waiting with bated breath or don’t really care.
For artists this allows you to see your art in real-world use cases right away. There's no need to construct a test area. Just load up any level making use of your art and check it out. See how it really appears in game instantly.
This deployment process also has minimal impact on the design workflow, which is not free, but something we work very hard towards. We’ll go into more detail later, but consider this. It’s often been the case that early in a project, art is trying to answer a lot of questions about how the game should look, and kits are no exception. This aesthetic process can be unstructured and time-consuming. The trouble is - design is often twiddling their thumbs at this stage, waiting for something to work with. This can result in art rushing through their process too quickly, sometimes making decisions they’ll change later, in ways that can drastically effect level design and force work to be thrown out or heavily revised.
To avoid this, we get our kits to a "functional-but-ugly" state as soon as possible. This is discussed in more detail later, but the high-level idea is that we figure out as much as we can about how kits work first, which allows design to get working while art then can focus on their visual direction without being rushed.
This is one of the contributing factors to another perk we enjoy; Bethesda level designers are quite fast. Because we’re working editor-side with pre-established kits, our level designers are able to iterate on layout extremely quickly. We can sketch with kits right in the editor and push those to the game, playing the results immediately.
In fact, we’d go so far as to suggest that a level designer working with a good kit that she understands is able to iterate faster and truer to final gameplay than any other designer with any other editor or workflow in the industry. This is because she's using the actual, final art and playing it almost as fast as she can work. We don’t have turnaround with art, we don’t have to bake textures or compile geometry, and our markup is fairly minimal.
Ah, but there’s a catch, and it’s one of the major downsides of our approach. When we say that a Bethesda level designer is the fastest, it’s with a big caveat. We specify that she’s working with a good kit. Without that, she’s not just the slowest LD in the industry - she’s a non-starter. We’re totally dependent on the art we work with, which is why the relationship between art and design is so important.
Consider an all-too-common tale from throughout the game industry. There are many variations of this story, but it’s typical that early in a project, designers are figuring out how their levels should play out. They may do this with editor BSP, external art and previs tools, or through paper maps and documentation. Whatever the method, design works on this until they’re happy enough with it to hand it off to art.
Hopefully this process has been well communicated - but in practice it often isn't and the art team ends up in the difficult position of trying to polish something they weren't very involved with. They’ll do an art pass, though, and make the level visually appealing Once they’re at a place they’re happy with, they send it back to design for final markup and scripting. Once design has done that, the level should theoretically be done - right?
Not usually. Level designers often inherit a litany of unforeseen problems when receiving final art for their levels. Cover along a street has been converted into poles too thin to take cover behind. A wall intended as a visual blocker is now a see-though chain-link fence. A bridge now has support beams which occlude sight lines in a major gunfight you had planned.
When work is handed off like this, it's often one of the main fault-lines of communication This story plays out again and again across all manner of games and teams. At best it’s non-optimal. At worst it can be downright toxic and create an atmosphere of hostility between art and design.
We’re no strangers to this. We've had our share of communication problems and finger-pointing, and have sought out ways to avoid the whole mess. This has come largely from seeking more open communication and collaboration.
Some users who download our Creation Kit comment that kits seem impersonal. To these people, our kit-based approach probably seems like anything except collaborative. Kits can seem restrictive when you’re an end-user - and in many ways they are. But this is one area in which the experience of a Bethesda developer differs greatly from that of a Bethesda modder.
Kits don’t simply appear in the editor one day, fully formed from the mind of the artist and complete, with no further pieces coming along. There’s a process we've worked out to establish any kit, and it takes place over a period of time during which a level designer and artist are in full collaboration.
Things To Know Before Beginning
This begins the technical portion, and it's worth establishing a few items. We’ll be referring to abstract units on occasion. If you’re familiar with Unreal units, for example, these are much the same. All you really need to know is that a character is usually about 128 units, or six feet, tall. We are also accustomed to a Z-up environment. If you’re used to Y as the vertical axis, keep this in mind.
There are a number of things you want to establish at a game-wide level before you begin on kits. As mentioned - you want to do this as early as possible. One thing we've found very useful is to determine a uniform dimension for door frames. This has a few benefits. It allows us to transition from kit to kit without unique pieces, as well as allowing easy re-use of doors between kits. More importantly, it gives AI and animation a fixed standard to work from, which can be reinforced throughout the game.
It’s also worth figuring out the most narrow traversable space in the game. We usually stick to a minimum of two character widths. This allows us to avoid pathing problems by ensuring there’s at least enough space for two AI characters to path around each other in any given space. You should also know how steep of an incline your AI can navigate. In our case this has usually been 60 degrees, but you should also know what looks good. The full incline often looks bad for animation, and you may want to embrace 30-45 degree slopes in your kits.
Finally, if you’re making a game with environmentally contextual gameplay, like a cover shooter or platformer, figure out some important gameplay measurements early on. If you’re making a platformer, figure out how far