Chapter 26
Chapter Select

World of Mystery

A god on a rampage and a smoother approach to level design helped Obsidian's level, gameplay, and narrative designers navigate uncharted waters.



The history of Obsidian’s Pillars of Eternity franchise can be written in numbers.

Five Infinity Engine games. Three wagers worth a penny apiece. One epic-sized console-based RPG that resulted in a devastating round of layoffs. Two funding goals: each set at $1.1 million, each more than doubled thanks to the contributions of hundreds of thousands of players. One project director willing to share rather than cede control with his dedicated team.

Many developers at Obsidian referred to the development of Pillars of Eternity and Deadfire as a studio-wide effort to bring Josh Sawyer’s vision to life. Sawyer doesn’t see it quite the same way. Yes, he’s the game director, and yes, most decisions are filtered through him. But game development is a collaborative effort, and Sawyer has always been willing to defer to peers if he finds merit in their input.

Clothing concepts for the Huana in Deadfire.
Clothing concepts for the Huana in Deadfire.

“Whether it's Adam, Bobby, Carrie, or whomever, there are a lot of people I used as sounding boards,” Sawyer explained. “I'll be writing something, and I'll go over to Bobby Null and say, 'Bobby, what do you think about this?' He'll give me some ideas, and I'll go, 'Okay,' and I'll go back and write some more.”

Sawyer followed Carrie Patel’s work on both Pillars of Eternity titles, saw that the tasks she performed and the quality of her work were of leadership quality, and extended his leadership on the narrative team to her. He trusted Adam Brennecke to make technical decisions that would facilitate faster development and more creativity. And, because he comprehended how much of his time writing would take, he asked lead area designer Bobby Null to step into the role of lead designer after Dave Williams, another systems designer, was needed on another project.

“Josh called me in one day and said he wanted to make me the lead designer,” Null remembered. The meeting took place almost one year into work on Deadfire. “At one point, Josh was doing lead narrative work because Eric Fenstermaker had left the company, and I was still the lead area designer. Between those jobs and the project director, there are five full-time jobs that Josh and I just figured out ways to do ourselves.”

Five lead jobs. Only twenty-four hours in a day. “We have a pretty good relationship, so it was all about figuring out how to be the most efficient. Sometimes I helped with systems. Area design was still my bread and butter,” Null continued.

“On Deadfire, we actually had systems designers, so I gave high-level direction on that stuff, but I didn't do tuning, and didn't build out the character-advancement trees,” Sawyer said. “It was a sequel to a game that already had fairly strong systems, so I knew I did not need to micromanage all of that stuff.”

Deadfire’s systems designers aimed to strengthen systems from Pillars of Eternity. The original game’s factions and related quests had been a sticking point. Players as well as developers wanted more diversity in quests and characterizations that might make one faction more appealing than another depending on the player’s gameplay style and character choices. For those reasons, faction design was a team-wide effort on Deadfire.

“We felt that, just using our past games as a reference, Fallout: New Vegas was one thing we really enjoyed working on, making different factions and seeing how they played off each other,” recalled Adam Brennecke. “We also felt that was a good way to reinforce the open-world feel that we wanted to go with, so we wanted to expand on that and make sure the different factions felt more integrated into the experience.”

Clothing concepts for the Principi in Deadfire.
Clothing concepts for the Principi in Deadfire.

The nomadic Huana faction sails from island to island in the Deadfire archipelago as the seasons change, making them perfect for players who like to stay on the move—even though putting down roots would favor the Huana. The Vailian Trading Company, a holdover from the first game, is bent on commerce. Their goals put them at odds with the Huana, who believe they have claims on the land that the VTC desire for their ventures.

The VTC isn’t the only burgeoning trade operation eyeing the archipelago. Ancestors of the Royal Deadfire Company inhabited the region as well. They’re interested in turning a profit, but they’re also trying to survive: Their stormy homeland made seeking out land where crops could be safely grown and harvested essential. “They were tricky at first because they were the pseudo-military faction, which as an aesthetic, was not what I would have chosen first,” said Carrie Patel. “So I thought, Okay, how do I make these guys interesting even to a player like me? How do I make them feel flavorful and unique without making them the grunt military faction, and there's nothing else to them? Giving them a survival-level interest in Deadfire was actually one of the changes I suggested when taking them on to make them a bit more sympathetic and to give them stakes that were distinct and on par with those of other factions.”

Players must ask themselves such questions if and when they get an opportunity to throw in with a faction. Joining the Royal Deadfire Company eventually closes off any chance at friendship with the Huana, including Tekehu, a marine godlike who will leave rebuke players if they align with the militaristic faction.

Along with factions, tactical combat sat near the top of the designers’ priority list. Fans of the first game had complained that combat was either too easy, or too excessive. Certain areas, especially regions near the game’s climax, boiled over with enemies. “It's feeling-based,” Sawyer said, explaining his macro-level approach to monitoring systems design on Deadfire. “I can look at a map, and if there are two rooms next to each other that have combat, I'll say, ‘Yo, does this really need to be this way? Let the player breathe for a second.’”

In design parlance, let the player breathe means give them something to do besides fight everything that moves. An elaborate puzzle might be the centerpiece of one floor of a dungeon. Once it’s solved, there might be treasure chests for players to loot. One of those treasures may be an examinable, jargon meaning an item with descriptive text that players are meant to pore over, perhaps gleaning a hint for how to progress or gaining insight into their current area or one of the members of their party.

Most of Deadfire’s dungeon environments are spread out between the titular setting’s islands. Sawyer and Null studied areas and asked pointed questions about encounter design: How many fights would be enjoyable in a space relative to its size? Is encounter X too similar to encounter Y in terms of which monsters are involved? “There are a lot of ways to approach that, whether it's diversifying combatants, changing the layout of a room, introducing traps or hazards—all those things can make one fight feel very different from any other,” Sawyer explained.

Combat was overhauled at a deeper, blow-by-blow level as well. In the first game, the graze effect let players deal a slight amount of damage if their virtual dice roll generated a number on the lower range of a scale. Some players opined that graze made combat feel too difficult to track amid other elements such as spell effects and the UI. Others—including Sawyer—felt that dealing some damage was better than no damage at all.

The Royal Deadfire Company faction.
The Royal Deadfire Company faction.

Deadfire’s designers revised graze’s implementation to satisfy both groups. An attacker’s accuracy rating is weighed against the target’s defense, then the difference is added to the value of a virtual dice roll. A number between twenty-five and fifty results in a graze at half the damage of a solid hit. “Misses were rare, and usually the result of a big defensive mismatch, where a target's defense was way over accuracy. Grazes, hits, and crits were the norm, so damage output felt much better. It didn't feel like a bunch of combatants flailing at each other. That made combat feel better overall,” Sawyer said.

Pillars of Eternity II’s change in setting dictated fluctuations to dungeons and encounters. Although the sequel packs in more content, it’s a shorter game than the original. Sawyer and the designers intentionally made sure players interested in the story could get to the end much quicker than it took to get to the final battle against Thaos in the first game, assuming they wanted to skip all the fun characters and lore revelations found in the game’s many side quests.

The main path was shortened because players and developers had found the original game’s last act bloated with content that slowed the pace. Therefore, encounters found along Deadfire’s critical path were given more attention than those designed as part of quests found off the beaten path. “In retrospect, we probably should have tried to make that a little more momentous. I think a lot of people were like, ‘Ah, finally! The lost city of Ukaizo! Oh—I guess I'm done.’ Even after doing this for almost twenty years, I still make mistakes and learn from them along the way,” Sawyer said. “Yes, if you just do the crit path, it can be very short. But also, Ukaizo, as a final area, is quite brief. You can actually skip the Ukaizo Guardian. You can just not fight the boss, so to speak, because we wanted to make the final boss pretty challenging.”

Systems design reaped the benefits of Deadfire’s revamped engine and toolset. Letting Null take the wheel on encounter design, Sawyer tested them by playing through levels to get a feel for pacing, difficulty, and context. Thanks to OEI Tools’ advancements, he had a method to quickly give feedback based on the types of characters and abilities in his party. “I would do a run-through with a standardized party and try not to vary them too much, but because there were eleven classes, sometimes that would require me to go in with a different setup. We do have debug tools that allow us to swap out party members and rapidly level them in Deadfire. We didn't have that in Pillars I, so having that in Deadfire helped a lot.”

Writing a comparatively shorter story—though one that can still take dozens of hours for new players to complete—combined with a larger design scope affected encounter design and tactical combat systems in other ways. Areas on the critical path, such as the Drowned Barrows dungeon set in a mountain the façade of which resembles a skull, received special attention from designers, but the size of the game relative to the small design team meant that some encounters underwent more balancing than others.

“When Deadfire came out, we erred a little—or a lot, depending on what your play style is—it being too easy,” said Sawyer. By the time Deadfire launched on May 8 of 2018, wheels were already in motion on rolling out patches that would increase difficulty for players. “I said, ‘Well, between too easy and too hard, I'd rather err on the side of too easy, and tune up from there,’” Sawyer continued.

Black and White

Whisking players away from medieval, Forgotten Realms-inspired locales to set sail across stormy waters and tropical islands was the perfect opportunity for Obsidian’s environment artists and level designers to rethink their processes.

“I think the change of scenery opened it up to being more colorful than the first game, which some people felt was a little drab,” said Bobby Null, who wore his lead level designer hat throughout Deadfire’s development. “I think that's another thing that really helped us. We wanted to increase the color palette, and the setting helped with that.”

Obsidian’s developers had dipped their toes in more color variety when they had painted snowy mountains for the first game’s White March expansion sets. That work had also given lead artist Kaz Aruga time to finesse details such as crossing from one type of terrain to another, such as from grass or stone to snow. After completing a render of an area in The White March, he would open it in Photoshop and apply blending techniques to give snowy regions a glittering, frosty appearance.

One of the tools Kaz had used on The White March was ambient occlusion, render data that holds the intensity of ambient light that surfaces are exposed to and contains them in an image. “The way we used it was, when you render a scene, you can tell the renderer to output that information in black and white. It brings out all the details of the map, and it's a cool-looking render,” said Aruga.

Examining black-and-white maps was a little like inspecting an X-RAY. Dark spots indicated sand or other topography tucked into spaces where light couldn’t reach. Artists harnessed ambient occlusion by using it as a mask when adding details into environments, such as dust and grime to give dungeons an old, rotting look, or an extra layer of sparkly ice in the White March’s maps. “On Deadfire, our environment artists put a lot of effort into developing materials, so they had that extra level of detail built into renders from the get-go,” Aruga said. “That also freed me up so I didn’t have to touch up every single render. The renders come out almost as if I'd done a touch-up, which is a huge visual upgrade.”

Ambient occlusion came with other advantages that were rolled into pipeline advances. “Using that pre-rendered shadow map, we can properly shadow dynamic objects in a scene,” explained Michael Edwards, lead graphics programmer on Deadfire. “The characters will receive shadows from the pre-rendered background, and the ocean will receive shadows that make sense. If they go into a recess that's shaded, they'll look darker, but if they're near a bright surface, they'll receive that bounced lighting.”

Edwards rendered the oceans that players sail in Pillars of Eternity II by doing what many programmers do: He bought a plug-in for Unity rather than waste time and money building water textures from scratch. Oceans in Deadfire are made up of giant meshes that, when rendered in-game, break up to create waves and other organic animations. “We can mask certain areas so they don't pass through ships, and we can add foam layers as well to make the ocean look like it's being disturbed by the ship. We also have a separate water shader that's used for pools, lakes, that sort of thing. We did some work on how the overlays are handled so we can add lots of foam and change the way water reflects light. We can do whirlpools, different waves, and cool things like that.”

Traveling on foot and making camp in the player’s stronghold had suited Pillars of Eternity’s Forgotten Realms-like wilderness. It wouldn’t do for Deadfire. “The stronghold wasn't well-integrated into the game because it was a location you had to travel back to,” Brennecke said. “The ship would be a location that traveled with you, going wherever you go. It would be easier to integrate into the flow of the game.”

Sawyer and Brennecke talked about bringing back the stronghold early in preproduction. The more Sawyer chewed on that, the more he became set against it. It simply didn’t work in a setting where players would be encouraged to sail around and go on adventures. “Certainly I think that for whatever shortcomings our ship systems have, I think they're way better integrated than the stronghold was in Pillars I.”

“It's a means of transportation, but it's also the player's stronghold,” added Carrie Patel. Like the stronghold, ships can be upgraded and customized, fitted with new sails, hulls, and ammunition for cannons.

Ships also tied in with the team’s collective effort to revamp factions. “You can also get different ships throughout the game, and will encounter different ships that factions have,” Patel continued. “In most cases, you can get one of those ships of your own if you so choose. You get to do some interesting things, like if you fly the colors of a particular faction and encounter those members at sea, they'll leave you alone, but their enemies may be more inclined to attack you. You're discovering and building a story around your ship.”

Changing the way players explored meant changing the structure of Deadfire’s world. “We'd have to change our narrative structure to be more open-world, so that change entailed going from the Baldur’s Gate II style, which was based on locations opening up, to a more free-roaming map like Fallout I and II, and Wasteland 2, where the player would be a token on this huge map, and you could freely explore and discover new areas. That would fit into the Deadfire region, which was an island chain.”

Open-world games are less linear by definition than games set on a straight line. With few exceptions, players can go anywhere and do anything by traveling along paths that open into new regions and often looped back on themselves to provide multiple points of ingress and egress. That style of exploration resulted in Deadfire’s map being overhauled as well as its areas. Instead of a chart showing locations players had visited, the map would be interactive, and would be a faster method of travel. “It was more about using the overland map as an exploration space to get to caves, castles, and places of that nature,” Bobby Null explained.

Null had experience working with overland maps. He was aware of their advantages, and their pitfalls. His first project at Obsidian had been Storm of Zehir, an expansion set for Neverwinter Nights 2 that had introduced an overland map. Players had liked using it to get around, but complained about enemies that appeared out of thin air to slow their progress.

Overland map.
Overland map.

Overland maps were also more difficult to implement. “Once you move into overland mechanics, you're creating a separate game, almost,” Null cautioned. “Any time you're doing that, you do have to tread cautiously. I think we did a great job in certain places, and in other places we probably could have done better.”

Painted by Kaz Aruga and fellow artist Matt Hansen, and animated by John Lewis, Deadfire’s overland map was much more versatile. Players navigate their ships in real-time, pick up items, battle other ships, and dock at locations where they can then disembark to explore and fight by way of Pillars of Eternity’s traditional controls and isometric camera.

The overland map helped immerse players in Deadfire’s setting and themes. They had to explore to reveal terrain, and special events would pop up as they performed actions such as sailing into uncharted terrain. Just as importantly, the overland map added continuity. In Pillars of Eternity, reaching the end of a map sent players to a loading screen so a new region could be rendered to the screen. Sailing around an interactive map and disembarking at various locations smoothed out those transitions by blurring the line between map screen and levels.

“You can sail to islands where there's a lot of content,” Patel said. “You can sail to smaller islands where you've got just a few dungeons. You can explore an overland map, using an icon to indicate your party, search for resources, get into fights. There's a lot more exploration to be done. In that sense, a lot more screen-area for the player to cover. It's different and quite a lot more open from what we had in Pillars I.”

Dynamic weather played a larger role on land than at sea. At any minute, sunny skies can give way to light winds that swell to thunderstorms. “I had envisioned the weather system being more integral than it ended up being, honestly,” Sawyer admitted. “We did not wind up leveraging weather as much for quests, mechanics, things like that.”

Despite the success of Pillars of Eternity, Obsidian still had to work within parameters of time and money. Sawyer’s weather system had too many moving parts for programmers to expand so that it could affect systems such as quests and encounters. Ultimately, it did more for atmosphere and verisimilitude than gameplay, which many developers appreciated. “The wind, foliage, rain, and rolling clouds. When you look back at Pillars I, the levels do feel static compared to Pillars II,” Sawyer said.

“It's cool because it adds more visual variety,” Patel agreed, “and we have more ambient NPC interactions, like when it starts raining, they'll run for cover, and [make comments] about the change in weather. There are some encounters at sea that respond to certain weather such as fog.”

Pyramid Scheme

Not every map in Deadfire is interactive, and they don’t need to be. Every town or city players visit comes with a map that divulges statistics about the area and displays a beautifully drawn illustration of the setting.

“Fantasy, for me, has always been about scope,” Bobby Null explained. “I read D&D books and liked to learn about how big a city is. Our town maps at least helped, in my mind, sell the illusion that Deadfire is a real archipelago, and the size of these places are not scaled down. Even with the walkable spaces still approximately the same, I think those maps make the world feel more realistic.”

Deadfire downsized the first game’s two-city design down to a single metropolis, a decision that helped streamline exploration and development. It also impressed on Null a need to make that singular big city, Neketaka, all the more memorable. “All of the Infinity Engine games—and Pillars I—do similar city designs where they're flat. I thought that if we were going to have a city with five or six districts, instead of them being in a relatively flat city, I wanted to build a pyramid: three districts on the bottom, two in the middle, and one at the top.”


Null built Neketaka alongside John Lewis, with whom he had designed levels for South Park: The Stick of Truth. Those memories were bittersweet. Much of South Park’s development had occurred following the mass layoffs stemming from Stormlands’ cancellation. Desks and offices had sat dark and empty. Null and Lewis had commandeered one or another and filled whiteboards with sketches of South Park maps.

On Deadfire, Null and Lewis pushed to integrate features and effects that had not been possible on the first game. One was parallaxing, a visual technique that makes background elements scroll by slower than objects in the foreground. Parallaxing fed into Null’s pyramidal layout for Neketaka, as did Deadfire’s expanded color palette and the growing skill of the environment artists. “The artists just got better at their craft. They were always good, but they just got better. They'd been doing this for a while, and they hit their stride on the second game,” Null said.

Deadfire’s level designers capitalized on the artists’ growing talent by increasing their ambitions. They still gray-boxed areas to get prototypes up and running, but the gray boxes evolved thanks to templates. Buildings looked more like buildings so that artists and designers had more to work with if a gray-box prototype passed muster. “They're more complicated, but it gave the area artists a much closer template for what we had going on in our imaginations,” Null added.

The extended capabilities of gray boxes came with drawbacks that could only be exploited by growing overexcited. It was tempting to spend too much time gray boxing instead of constructing levels one element at a time and testing them through OEI Tools as work progressed. Designers could import starter parties—groups of pre-made characters of varying experience levels so that late-game areas could be tested without having to start at the game’s beginning—and explore their gray boxes, keeping eyes peeled for spots that needed adjustment: gaps between geometry, swathes of terrain that were too crowded or empty.

Placing nav art.
Placing nav art.

“A lot of times what can end up happening is you can have a blockout that looks great when you're looking at it from a bird's-eye view, and then you run through it zoomed-in and there are big swathes on the screen where there's nothing: no props, nothing to help guide players to where they should be going next,” Null explained. “It's up to designers to catch those things, because what we've dictated what we're hoping for, a lot of times the area team will take what you give them and make it look great, but they may not have as much time for them to play through it. Level designers have more time to play through levels as we build them.”

The larger an area, the more playtesting it warranted. Their focus was on tightening the core gameplay loop that drove Pillars of Eternity and the Infinity Engine games: enter an area, kill things, find loot, rinse and repeat. That loop changed depending on the area being designed. “Once you get into a civilized location, particularly in the games that we make, it's a lot more talking to people, figuring out quests, and, in cases, playing them a different way and inject your core gameplay loop into them,” said Null.

Deadfire’s core gameplay loop is one of two. The first, core loop operates on a minute-by-minute bases: explore, kill, loot. The other is larger, and combines dungeon delving with civilized areas such as towns: explore, kill, loot, then leave the hostile region and return to town to claim rewards for quests and buy supplies before getting back on the road to adventure.

“It's something we're always trying to balance,” Null admitted. “As soon as you start getting in the mindset of trying to make a living space feel more realistic, you get further away from the game's core loop. It's always a challenge, and we do have to hand-wave some things.”

A level blockout in Deadfire.
A level blockout in Deadfire.

Constructing spaces that feel realistic and livable is perhaps the greatest challenge Null and his designers face. It’s a moving target due to the demographics that enjoy RPGs. “Some fans are completely happy going around talking to people all day long, not getting into the combat loop, whereas that drives other people nuts. You're always trying to juggle those two things to pace out your game as best you can.”

Neketaka presented designers with opportunities to fuse combat and more serene exploration. The pyramid-shaped town is divided into districts with assorted populations. Early on, Null wrote documents outlining each of the city’s districts: what type of content should be there, how much of the game budget designers had to work with in each section. “Early on, every one of them had a dungeon,” he said of Neketaka’s seven areas. “That sounded great, but things change based on what we need for the story. The ones that ended up retaining the dungeon spaces, I like them better, because I think they're paced better.”

The Gullet district features a smaller civilized space juxtaposed against a larger dungeon zone. “I blocked it out early in the project, and I wanted the space where you're walking around talking to people to be relatively small, and it was, compared to something like going into the Queen's Berth,” Null said.

Neketaka’s Queen’s Berth district is the single largest map in Pillars of Eternity II. Decking out such a huge space with characters and attractions excited Null, so he blocked it out in gray boxes himself. “I wanted it to be the biggest scene in the game for this reason: When you get to a big city, you want to open the map and say, ‘Oh my god, this place is insane.’”

Neketaka's Queen's Berth district.
Neketaka's Queen's Berth district.

Null mapped out Queen’s Berth over two weekends. Its final layout, from canals and the placement of its tavern to the path players take through it, to what players see and when they’re meant to see it, was honed over rounds of iteration. Once a designer finished blocking out an area, it went to the environment artists. Null added an extra step on Deadfire, a method of calling attention to fine details he or other designers wanted to make sure the art team included in their final render. “We didn't have time to do concept work on scenes in the first game, but we did on Deadfire. That's another reason the maps look so much better. Matt Hansen and Lindsay Laney would go through screenshots of what I'd blocked out, and they would go through and draw awesome line illustrations to further define what the city should look like.”

Kaz Aruga iterated on levels, too. He went into Deadfire determined to clean up the game’s user interface so that UI elements co-existed with environments and on-screen action such as character movement and spell effects. “Visibility and clarity were my concerns,” Aruga clarified. “One thing we took away from Pillars 1 was that the environments were noisy: You'd get high-fidelity, crunchy, nice textures on the ground, and they look great when you're just looking at a render, but in the context of combat, it often contributed to visual chaos. That's something we wanted to weed out from Pillars II.”

Minimizing chaos meant rethinking color palettes, such as making sure UI elements did not conflict with or blend into backgrounds. Aruga’s concentration on clarity paid off. Special circumstances such as activating stealth mode to sneak around characters more clearly highlighted each character’s line of sight so that players knew exactly where they were looking, and when.

Null was happy with how Deadfire’s tropical, stormy world turned out, though he reflected on the abundance of smaller dungeons compared to larger, multi-level spaces. “There's a lot of smaller living spaces, and a lot of really cool quests,” he said of Queen’s Berth, “but I would say that area could have used—and originally had this—a strict adventure space that gets you into that gameplay loop more. Those are the kinds of lessons we go through. Lots of people have lots of great ideas, but sometimes we make tradeoffs and try to get different types of gameplay going to mix things up.”

Chasing a God

When Eric Fenstermaker and Chris Avellone left Obsidian, Carrie Patel became the only narrative designer who had worked on every installment of Pillars of Eternity: the base game, the two-part White March expansion set, and Deadfire.

“I had a lot of institutional knowledge and a greater awareness of how our pipelines worked, of the Pillars tone and flavor, and what our vast lore included,” she said. “Also, because all of the other leads were incredibly busy, I was often a much easier person to approach about a lot of those questions. Writers and people from other disciplines who wanted to know something about Pillars lore or how we did things, like using our tools, would oftentimes come to me with those questions.”

Already spinning several plates as Deadfire’s game director, Sawyer took on the role of lead narrative designer. Someone had needed to fill the void left by Fenstermaker. He also wanted an opportunity to take a hand in the game’s story and characters.

“I think it's important for me to do writing. I always do something in the guts of the game on every project I work on,” Sawyer said. “When I started in the industry, my first thought was, I'm going to make all this crazy shit!” Sawyer recalled. And the programmers said, 'Well, that's fascinating, dude, but none of that stuff is supported.' The more I understand about the reality of how an engine works and doesn't work, the better informed I can be when I'm talking with the designers who are working on that stuff every day. I fear that if I never write a dialogue, if I never create a spell, if I never set up data for a character class, I'll be too far removed from the reality of technical limitations of the engine.”

To juggle his many other obligations, Sawyer relied on Patel. Other narrative designers had gravitated to her since work on Deadfire had begun, asking her feedback and for help navigating OEI Tools. Other designers likewise looked to her for guidance. “I feel like to an extent, I'm the same way with managerial stuff that I am with a map, for instance,” Patel said. “If I'm traveling with people who know exactly where we're going, I will just check out and follow them. If it turns out that someone needs to step in and figure out where we're going, figure out the best route to get there, well, okay, I'll do that, and I don't mind it. That's not a complaint about a lack of direction, but just the fact that there's always a need for someone with both the inclination and the availability to do all of the same-scale follow-up, even: with people, with issues, with creative or technical questions. If there's someone else already doing that, I'm happy to let them do that.”

Patel quickly fell into step with Sawyer, collaborating with him as she had with Fenstermaker. “I find that Josh is very reasonable and willing to hear things out, and then take those things into consideration,” she said. “For Pillars II, whenever there was a question about a big creative direction or decision that needed to be made, I'd pass it to him and say, ‘Where do you want to head on this?’ When it was something that was kind of established or understood, or if it needed more definition or something we already had an internal answer on, I'd usually follow up on it on my own.”

Sawyer gave Patel and the other narrative designers the credit for carrying the heaviest loads. He had to be more concerned with the ten-thousand-feet view out of necessity: story structure, making sure the story gave players freedom to confront moral dilemmas and respond to them in their own way, and providing opportunities for players to flex their tactical-combat muscles. “I wrote Pallegina, which is the companion I wrote in Pillars I,” said Sawyer. “I wrote Eothas, who is the main, driving force behind the plot in Deadfire. And I wrote a handful of side characters.”

Coming from the groundwork she had set with Fenstermaker, Patel understood many of Deadfire’s beats and themes. What she liked about the story was that the big picture framed two epic conflicts: powerful, god-like beings such as Eothas competing and cooperating with one another to reshape the world of Eora; and smaller, mortal races, personalities, and factions who were just as determined to run the world. “As the player, you're encountering both of these forces, looking at the solutions they offer, and asking yourself questions, hopefully,” Patel said. “‘What's best? What's the right shape for things to take?’ ‘How are stability and salvation best found? Do we look to our past and honor it? Do we innovate and plan for the future? Do we guide and direct one another? Do we offer one another total freedom?’”

Players find answers depending on the gods and factions they align with. The narrative team’s goal was for players to have time and space to ponder questions between soaking up the Deadfire region’s attractions and partaking in deep combat.

Above all, the game’s colorful palette mingles with the narrative team’s change in tone to encourage players to enjoy themselves, a sharp contrast from what Patel had described as the “pessimistic realism” of Pillars of Eternity and its string of tragic events. “I don't think pessimistic realism was appropriate for Deadfire, and I don't know if it was a question so much of what would be appropriate. I think it's still pretty pessimistic because we've got colonialism, slavery, unchecked capitalism. But I do think we wanted to balance all of that with a little more humor, more levity, and more of a sense of fun.”

The narrative designers were as interested in buffing the dents out of Obsidian’s pipeline as other disciplines. By taking guesswork and frustration out of OEI Tools, writers could be more creative and less concerned with figuring out how things worked. Most quests in Deadfire started as a stub, a simple and often vague framework that the narrative designer assigned to a task filled out. In no time at all, the designer was able to connect events and dialogue to the stub and playtest them just to see how the beats of a quest line or character interaction play out.

For their first pass, narrative designers played with characters and flavor. For side quests, they felt their way through finding out what story they wanted to tell. When writing for story quests, they worked to understand what segment in the plot they were filling in: how it connected to what story beats had come before, and where they needed to leave things for the designer in charge of writing the next leg. “Once you're comfortable with that structure and with what you've set up, the narrative designer will go in and do a more polished writing pass: character details, nuance, voice, and some of the different types of dialogue interactions and reactivity that we like to see in finished conversations,” explained Patel.

Among the interfaces and programs within OEI Tools is a conversation editor. That, not word processors such as Microsoft Word, was where narrative designers wrote and revised their work. “It also includes item text, examinables, UI text, character names. All of those strings live in the string editor, and many of those strings will not necessarily be present in the conversation editor because they're not part of conversations,” said Patel.

Many of OEI Tools’ narrative-oriented tools and processes were in place long before starting Deadfire. A designer may decide, for example, that he or she wants to draw the player’s attention to a particular spot in an environment. Designers dragged-and-dropped an examinable component into the area, opened the component, and entered the appropriate string ID—a reference to a pre-written piece of text—and OEI Tools and Unity automatically pulled it from a database of text strings. “You can technically do some writing from Unity, although the Obsidian Tools is a much cleaner interface,” said Patel.

An important phase in refining the narrative pipeline was coming up with templates for common tasks such as area design and quest design. A template included information designers would need to know such as where a quest or character was located, what needed to be accomplished, what media such as animations and audio was needed to make it happen. Templates were shared with every lead, artist, programmer, and designer responsible for working on that content so that everyone grasped where that content fit—especially in terms of budget and time—of the project’s development.

“You could look at a quest, for instance, and say, ‘If these are the NPCs and what they're doing, this is about how large their conversations will need to be in order to feel satisfying and complete. That means about this much time for this content,’” Patel said. “Then you can look at that and say, ‘Is this in our budget, or do we need to trim some part of this out or tone it down? Are we being realistic in what we're saying we can get done in this period of time?’”

All of those processes and tools proved instrumental to writing a story for an open-world-style game. Originally, the designers wanted to give players their ship at the beginning of the game but prevent them from sailing into deep waters until they had paid a visit to Neketaka. The designers did away with that restriction. “Then there was this idea of, well, what if we just let the player travel wherever they want?” Patel remembered. “We warn the player: ‘Here be monsters. Content gets pretty hard if you go this way, so what you should really do is go to this city,’ which we've put on your map, but why not let the player go anywhere they want?”

Except in a few instances, players can sail where they want, when they want to go there. Deadfire does block off end-game content so players cannot drift into the final encounters ahead of time. Virtually everywhere else can be accessed. “Players who have been through the game and know exactly where to go—we're talking about speed runners to a point—and want to ask, ‘Can I do this?’ the answer is, ‘Yes, you can.’ You can get through the game relatively quickly if you're skilled enough,” Patel continued.

Not all players like to be dropped into a world and left to find their way. The scope of a game as vast as Deadfire can be overwhelming. For those players, the narrative team constructed quests—as well as collaborated with other departments to put together quick tutorials—to teach players the basics: how to move, how to attack, how to sneak. Additionally, the narrative team sets down clear paths to the next area players should explore if they want to continue along the critical path, but does not push. Players can explore while knowing where they need to go and what they need to do when they’re ready to move on.

The overland map was designed in a similar fashion, with areas of interest such as other islands and obstacles like enemy ships peppered over the waters to catch players’ eye. “Even if you try to sail in a straight line from the first island to the big city, you'll have to go around a few little islands here and there,” Patel said. “You'll see ships passing, and can engage in combat if you want to try it. You'll find islands that have landing points that you can click on to go to, and you can walk around and explore them.”

Referencing the overland map reveals icons for things like combat, dungeons, and resources to pick up. Players can breeze right past them if they have other plans, enabling the narrative team to use them as breadcrumbs rather than forcing players to acknowledge them. “We made sure to include that type of content between any two points we would send you so that you'll always find something else to do. If you decide to deviate from the route you've been given, you'll be rewarded with something interesting and new,” explained Patel.

A staple of classic and classic-style RPGs, companions are divided between old friends and new blood. Edér is one of the first familiar faces players see after they complete Deadfire’s prologue, which reintroduces them to being a Watcher—one who can read souls—and then deposits them in the cabin of their ship, where they find Edér waiting.

Then there’s Aloth, another fan-favorite who had his beliefs shaken when he discovered the true cause behind the Hollowborn crisis. It’s been half a decade since the player’s Watcher avatar and Aloth have crossed paths, and during that time he’s made peace with the events of the original game. “I think with him, it was kind of cool to have someone who was a familiar character, but also to show how he'd changed as well,” Patel said. “It was a little bit of a challenge to figure out how to do something that's not the same old Aloth, but that still feels appropriate for the character we've built, and that fans will recognize from the first game. It was fun to write a stronger, more sure-of-himself Aloth, but also someone who had some conflicts and doubts that we knew from before.”

Dereo the Lean was a new character, written by Patel, and one of the most creative challenges she faced while writing Pillars of Eternity II. He belongs to the Principi, a faction of pirates who, through a ruling body of seven captains, attempt to control the activities of pirate fleets in and around the Deadfire. “I remember doing a first pass, and it was kind of what you'd expect: ‘Oh, yeah, he's a nasty, sneaky guy, living down there with all the criminals.’ That was the feedback I got,” Patel recalled.

Sawyer read her first take on Dereo and recommended that she inject more of the nobility that flows through the blood of the Principi’s ancestors. That feedback empowered her to bring out more of the character’s innate flair. Dark-skinned, somber, and methodical, he bears a striking resemblance to drug kingpin Gustavo Fring from AMC’s Breaking Bad TV drama, a comparison that perhaps influenced Patel as she worked and reworked Dereo’s interactions. “He was still someone who fit the role, but I found ways to make him an individual. I think he's fun because you can tell he's someone who's having fun doing what he's doing. I think that's fun to write and fun to encounter as well.”

Narrative designers maintained a balance between text and audiovisual elements. “The rule I use is, if the line a character tells you could just as easily come from another character, it's not character building, it's just information and a waste of time, for the most part,” said Sawyer.

Writing meant more than typing a bunch of words and plugging them into Obsidian’s conversation editor. Narrative design was a multi-discipline effort. If a narrative designer wanted to, say, show a character performing a meticulous task such as setting a table, the designer would need to work with an artist to figure out how or if to show that action. “Maybe the art needs of having this table laid out with lots of food as opposed to very bare makes it into the artist's notes, and maybe it doesn't. Communicating that and catching it early is important,” explained Patel.

Leading from the vantage of game director, Sawyer worked with narrative designers to diversity as well as tighten dialogue. In RPGs, writers have a habit of giving every character information about the area to tell players. Sawyer thought that was fine, but a seedy character who skulked outside the town walls and holds a grudge for land that was swindled from him would have a very different view of local issues than, say, a magistrate.

“If you say, ‘Yeah, so what's up with the town?’ they might say, ‘You wanna know what's up with the fuckin' town? I'll tell you about the goddamn town,’” Sawyer suggested. “What you're doing is, you're giving the player this information that is character building and conflict building. It helps develop the world more than just reading a Wikipedia entry to them through a voice.”

Sawyer viewed good prose as the cherry on top of a story sundae. When critiquing narrative design, Sawyer pointed out instances where prose was better off left to dialogue, or vice versa. “For example, describing a character doing something and then making a gesture that is not very distinctive of the character, and feels like it just reinforces what the character just said? Not a good use of prose. You're already getting the vibe of what the character's mental state is from the line itself.”

One of Sawyer’s favorite uses of prose over dialogue was Orron, a dwarf written by Carrie Patel. When players meet Orron, he’s sitting at a table pouring equal amounts of wine into glasses. Quiet and intent, he tops off each glass until the bottle is empty and each cup holds the same amount of wine. “It's something we can't see something happening on the screen. It evokes your imagination. It lets you understand, Holy shit, this guy is super-obsessive compulsive, which informs some choices you can make in the conversation later to really throw him off. When I saw people using prose like that, I tried to reinforce, ‘Yeah, that's great. That really builds the character. That gives you insight you couldn't get from the scene.’”

Patel credited the importance of rewriting to hitting high notes such as Orron and the Royal Deadfire Company, another of her charges. “I think sometimes it's easier to let yourself experiment with something early without worrying too much about making it really good and polished, and then come back to it when you've had time to let that idea sit in the back of your mind, and feel a little fresher toward it,” she said. “I feel like it's when you try to do all the passes at once and get a really polished draft in your first go, that unless you're just on and know what you're doing, that's what a lot of times leads to the whole ‘staring at the blinking cursor, write something, delete something, write something, delete it again.’”

Patel integrated iteration into narrative’s pipeline. An exercise toward that end was the entire team taking a break from tasks and playing through all of Deadfire’s content and submitting feedback when possible. “That's great for catching bugs, because you're most likely, as the designer, to know what you're missing. But you're obviously not getting fresh eyes on your work, and outside perspectives from people who may not know your intent [ahead of time],” she said.

As Deadfire’s release date closed in, narrative designers—and the team at large—had less time to playtest and iterate. “Certainly there comes a point when, as text is getting locked down and as other areas are being locked down, as things are getting translated, you have to look and say, ‘Here are the things I have to fix. Everything else after that is about triage.’ You get into bug-fixing rather than content creation,” Patel explained. Triage could mean trimming text or cutting it completely, or accepting the fact that the best she or another designer can do is polish what’s there to make it the best it can be, even if it falls short of expectations.

“I've been a lot more involved in trying to keep us coordinated with other teams, make sure everybody's on the same page, make sure people are getting the feedback and direction we need to create good content, and make sure the content we're building syncs up well with everything,” she said. “We've got a much larger writing team now, which is good, because this game has turned out to be quite a bit bigger than the first one. I still do a lot of writing, but also a lot of, on a day-to-day level, trying to keep everybody in sync and where we are on all fronts.”

In January 2018, a few months before Deadfire was due to launch, Sawyer asked Patel to meet with him in his office. He had been impressed with the creative work and responsibilities she had undertaken since joining Obsidian, especially on Deadfire. He asked if he could formally recognize her as co-lead narrative designer, a promotion that would lead to other opportunities following the completion of Pillars of Eternity II. Patel accepted.

“Carrie is fantastic,” Sawyer added. “She did great work on Pillars 1, and on Pillars II. I can't wait for her to be a lead on a project where I'm not hovering over her, because I think she's going to do great stuff.”

Hello, Meet Lola