How Battlefield devs learned to build an open world for Mirror's Edge Catalyst

Swedish developer DICE has been putting out games since the twilight of the Amiga, and most of them have had single-player components carefully constructed to be linear, scripted experiences.

So the studio’s latest, Mirror’s Edge Catalyst, is a significant departure for DICE. This revamp of the studio’s 2008 cult classic Mirror’s Edge does away with "levels" entirely in favor of a more open world (the "city of Glass") filled with missions and activities that players choose to take part in.

It’s a sign of the times, to be sure. But as DICE climbs into the sandbox of open-world game design alongside so many other big-budget developers, it’s interesting to study how the studio best known for Battlefield has adapted to meet the demands of open-world game production.

“It is challenging,” says design director Erik Odeldahl during a recent phone conversation. “We always talk about it as a free-roaming title, but it’s still a game where you play through a linear story because we do like telling linear stories.”

Odeldahl and many others on the Catalyst team are Battlefield veterans, and he acknowledges that the main storyline missions in the new Mirror’s Edge are tightly-scripted experiences built using lessons learned from structuring Battlefield campaigns. But in the space between those missions, Odeldahl says the design team struggled to create a world that was exciting to explore -- but technically possible, and not distracting enough to overshadow the game’s narrative.

“We're fairly used to building tight, action-packed, event-driven stories in Battlefield. So I think we've been trying to balance that,” he says. Since Catalyst players can dip into or out of the game's main narrative at any time, the devs were challenged to build a single-player game that was engaging simply to play. “This game is about running, and it's about jumping, and it's about fighting. But it's also about finding optimal routes. It's about shaving off seconds or milliseconds off your time.”

The original Mirror’s Edge was also designed to entice players into searching for the best possible path through each level, and it worked -- to this day the game is a touchstone of the speedrunning community. But that’s a much more manageable proposition when you know you’ll be designing each level by hand, knowing exactly where the player will start and end -- how do you build a non-linear open world that’s optimized in the same way?

Build a rough version of the whole world, then fill in the details

“We couldn't leave much to chance. So what we did was we built the city using huge geometrical chunks,” says Odeldahl. The Catalyst team basically built the game’s entire world first, sculpting a rough city map out of primitives, then the level designers and artists went in and added or adjusted details over time. “So we had the whole city, almost all of it, built up like that...and there’s this constant iteration, as we try to figure out how the movement works.”

That alone is interesting because rarely do we see open-world games built to be quickly traversed by a player who’s highly mobile on foot, right from the get-go. Open-world game protagonists typically get larger and more bombastic as their mobility increases -- think sports cars and jets in Grand Theft Auto games, or the superhuman movement abilities in games like Saint’s Row and Prototype. Catalyst’s protagonist is much more humble, and nimble, in her movements -- and unlike the Assassin’s Creed team, the designers at DICE had to make all the jumping and swinging between buildings look good in first-person.

"What we learned was that we needed to make the controls more responsive than the first game," says Odeldahl, because players were far more likely to change their minds mid-move when navigating the more open, nonlinear spaces of Catalyst on the fly..

That brings with it an interesting challenge: how do you clearly indicate to a player which gaps they can leap, what walls they can climb, without distracting them from the experience?

“We use multiple different layers,” admits Odeldahl. “On one layer we use things like scratch marks, foot marks on walls, to show that it's a suitable place for players to do a vertical run up or a horizontal run across the wall. We also use ventilation systems and stuff, and basically everything of a certain height, whether it's a railing or a ventilation system, everything of a certain height is interactable in a certain way because the movement is contextual.”

The second, meta-layer is a “Runner’s Vision” mode that players can switch on at any time. The game then algorithmically generates a crimson path from the player’s location to wherever they’ve set their destination, highlighting traversable objects along the way.

That means the game has to be capable of charting routes from anywhere in the world to anywhere else, and the world itself (which is totally hand-built) had to be designed and balanced so that multiple such routes exist. More importantly, DICE had to build it so that there weren’t any frustrating gaps or “dead zones” players couldn’t quickly move through, while also trying to make sure every area of the city felt distinct, believable and real.

“Some parts of the city we've actually rebuilt several times over, from scratch," says Odelhahl. "Because we realized that okay, it's not readable, or it's visually not as interesting as we want it to be, or stuff like that. It's been a lot of work. And we don't have an immense amount of level designers or anything."

While the Catalyst design team did study telemetry data to track player movement patterns through the game during development, they didn’t have access to production tools capable of doing things like automatically generating a minimum viable number of player-traversable paths in a given area, or flagging gaps as being leapable or not.

"Other EA studios will learn, and look at what we've done"

Instead, Catalyst level designers had to rough up their own solutions until they got the hang of what players could and couldn’t do in-game.

“We set up all those [movement] rules fairly early, and then we built prefabs, brightly-colored prefabs, which we clearly marked like -- they were basically like rulers that we could place in the world,” says Odeldahl. “We actually didn't have a tool that generated these things. We did have a lot of automatic testing though, to make sure that the movement system didn't break down after animation changes and stuff like that. We had, basically every check-in, everybody working on the project always runs a number of tests before they're allowed to check in whatever it is they're working on.”

And since the world of Catalyst is basically one giant “level”, the design team used prefab chunks they could pull out and work on as required.

“That way they didn't have to work directly into the level,” notes Odeldahl. “ And the whole city is more or less split into streaming zones, and as the player traverses the city, depending on where they are the zone they're in will stream in, and also neighbor sets. Each zone has its own neighbor sets that stream in based on what you can and can't see.”

That’s a new way of working for DICE, and Odeldahl believes the lessons learned from development of Catalyst will influence the way other Frostbite devs build their own worlds.

“I'm 100 percent sure that other teams within DICE, but also outside of DICE, other EA studios will learn and look at what we've done,” he said. “And probably use it in some way, especially when it comes to these big, big worlds. Loading and streaming and stuff like that.”

For devs embarking on their own open-world games sans Frostbite, Odeldahl has one piece of advice: plan thoroughly, build early.

"Try to create the world -- in our case, a city -- as big as you want it to be, early. Because you don't really know, until you see that, if it's even possible for you to build it," he says, adding that some of those revisions DICE made to the city of Glass through development were done because some areas wound up being too small -- or too big. 

"As a designer, I think you should plan your world, plan your city or whatever it is you want to build, early -- so you get a good firm grip on whether you can actually build it or not."