Building incredible multiplayer games since 2012.

Surgeons and Firefighters: Two Types of Level Design in the Funpocalypse

Surgeons and Firefighters: Two Types of Level Design in the Funpocalypse

image

Within the realm of design-land at a video game company, level design typically comes in two flavors, each with its own set of quirks. There are generally things I love in one type and miss in the other, but neither is inherently better and both offer distinct advantages. What follows is a brief, high-level introduction to both traditional and procedural level design, as well as how we approach it here at Proletariat.

When talking about traditional level design, there’s something I like to call, for lack of a better term, “traditional level design.” Building levels this way gives the designer direct and precise influence over most of the minutiae in the design of a level: obstacles, enemies, objectives, etc. For thematic and mnemonic purposes, this is the “surgeon’s approach” because surgeons use scalpels and those are supposed to be precise. The Surgeon is also a playable character in World Zombination. See what I did there?

image

Traditional level design has a designer manually manipulating and arranging assets to produce a compelling experience that falls somewhere in that just-right-porridge state (technical term). It’s not too tough and it’s not too easy, but juuuust right.

Beyond traditional level design is what Proletariat leverages in the development of World Zombination: procedural level design. In describing this approach you could put on your fancy pants and call it algorithmically-made level design. Or you could stay in those comfy sweats and call it the “firefighter’s approach.” Much like the axe-wielders in our game, I like to think that procedural level designers do their jobs in much broader strokes. Like swinging an axe. Firefighters swing axes in broad strokes. See what I did there??

Now, as to those quirks mentioned above, each approach to level design has associated strengths and weaknesses. I’ve labeled the former approach traditional because that’s largely how it’s been done with content-driven games: one guy manually laying track that is always more-or-less the same at that point in the game for everyone who’s going to experience it. Mario Brothers games are the quintessential example here: it’s a handcrafted experience and pretty much the same for all players. 

image

Traditional level design empowers an expert designer to lead the player along a predetermined experience. A good level designer has done his or her job if the player can put on their smarty pants (these can be dressed up or down) and progress forward under the guise that they accomplished it entirely on their own (when the level designer was telling them how to do it the entire time). Another example that jumps to mind here as it applies to narrative is Portal; you can’t help but feel compelled to follow and enjoy the narrative from little more than cake-related vignettes left for discovery at opportune times.

This is where I’d like to make an important aside: a level designer is first and foremost a teacher. This is one of the strengths of traditional level design: since designers have finite control over the minutiae, they can determine how and when a player will be exposed to new mechanics, new enemies, new narrative, and other shiny new things that are…new. They can introduce new concepts in a safe environment where the player is free to experiment, they can teach the concept through ramping iteration, and challenge the concept precisely when and where the player is primed for the challenge and standing on the precipice of mastery. It’s a great feeling when you get there as a player, like hitting that elusive “flow” state, but chances are if you’ve ever felt empowered by your own mastery of a new skill in a game, or had that “A-HA!” moment in the narrative, you have an expert level designer to thank at least in part.

image

Yes, traditional level design empowers a designer with precision, but this comes at a cost—literally. The common con associated with traditional level design is the expense: time, money, people, you name it. Somebody’s got to sit down and spend the time iterating on an “optimal” experience, and those somebodies gotta get paid.

But with a little ingenuity and a lot of talent, developers can put together a solution for a game like ours, which on one hand requires a ton of compelling content, but on the other hand, is being made on a budget. Where a surgeon uses a scalpel for the smallest details, a firefighter has his axe. Now, I don’t in any way mean to suggest we’re figuratively fireman-axing levels to death in comparison to precisely modifying elements. Rather, in World Zombination we’ve got a pretty rad system that simplifies the process of designing cities and their layout (our version of levels).

Allow me to better explain with an example. Our artists have made a ton of wonderfully post-apocalyptic buildings, many of which are specific to a single city. This is done so each city carries over the identity of its real-life counterpart. Traditional level design in many cases would then call for some combination of level designers and artists to assemble these hundreds of assets into interesting combinations in and around the play space. For a game at the scale of ours—with dozens of cities and hundreds of city blocks—that’s an amount of level design work that could occupy a small army of designers.

Rather than grab my scalpel, I’ll instead reach for my axe. Generally speaking, what are the sweeping strokes a particular city would need to make it recognizable? What’s the min/max for park coverage in San Francisco? How about Detroit’s industrial area? What about New York? How do we represent Central Park? 

image

By determining generalities about these and a slew of other factors, we can successfully approximate the layout, look, and feel of any city. Better still, we can randomly generate near limitless unique variations of each city, blowing the roof off of replayability. We can tell our tools to pull from certain décor or gameplay sets—what we call ‘encounters’—and calculate offensive and defensive ballpark values along a particular route through the generated city. We can determine the complexity of the streets layout, which buildings appear adjacent to roads, and which enemy types are allowed to appear and in what combinations. At the end of the day, we have a ton of variables to tweak and modify.

It’s a remarkably robust system and incredibly efficient, saving countless man-hours of development around each city. The drawbacks come in when we demand a great deal of precision. It defeats the purpose to use the tool this way, but there’s a very good reason why we do it exactly once. It’s called the first-time user experience, or the FTUE (‘fuh-too-ee’—that is actually a thing), which could be its own post so I’ll summarize briefly: the first time a player plays a game is also the first time they can make a judgment and either become a dedicated player or walk away forever, so you want to get it right. Understandably, we bend over backwards to make this first experience as accessible as possible, considering what’s at risk.

Similarly, many games without procedural level design try to shoehorn in the replayability and content load of this approach—but with limited success. Look at mastery systems (lots of mobile and Facebook games), mirrored levels or playing levels in reverse (Mario Karts and shooters), incentives to play through the game again or on a harder difficulty (“new game plus!”), or modified art in an otherwise unchanged experience (ever beat Super Mario World’s Special World?). Instead, many of these approaches just come across as a desperate plea to keep players playing and arguably aren’t as compelling as new, original content.

Weighing both sides equally, there are parts of me that do miss being the surgeon and giving a level that element of craftsmanship you can’t really achieve otherwise. But procedural level design’s got more heft and staying power and can keep things fresh for far longer. It’s still possible to noodle with individual gameplay encounters and drop them in the pool of assets a generated city can draw from. And an axe is way better in a zombie apocalypse anyway.

Josh Hufton, Level Designer

On Twitter