Blood, Sweat, and Pixels

by Jason Schreier

  Early on, the developers knew that they wanted The Witcher 3 to be far larger than anything else out there. Most video games aimed for a campaign that was between ten and twenty hours long. The bigger ones, the RPGs and open-world games, usually set out for a range of forty to sixty. With The Witcher 3, CD Projekt Red wanted a game that would take at least one hundred hours to finish. To hit a number that absurd, The Witcher 3’s designers needed to start as early as possible, sketching and writing as much as they could during preproduction, before there was even a game to play.

  Everything would begin in the writer’s room. “We start with a very general idea,” said Jakub Szamałek, one of the writers. “Then we expand it, then we cut it into quests, then we work closely with quest designers to make sure it all makes sense from their perspective. And then we iterate and iterate and iterate.” The main quest would center on Geralt’s hunt for Ciri, they decided, with interludes throughout the game in which you’d play as Geralt’s adopted daughter. There would also be a series of important quest lines that were optional, but would have a major impact on the game’s ending, like a potential regicide plot and a love triangle between Geralt and the sorceresses Triss and Yennefer, both of whom had appeared in previous Witcher games. Then there were the minor quests, which included an assortment of mysteries, monster hunts, and errands.

  As head of the quest department (a role that he’d inherited from his brother), Mateusz Tomaszkiewicz would work with the writers to lay out a basic theme for each quest (“this one’s about famine”), then assign it to a quest designer, who would plot out exactly how that quest would proceed. How many combat encounters would it have? How many cut scenes? How much investigation? “This whole logical chain of events would be done to come up with how this situation actually occurs, what are your objectives as a player, what are the challenges you have to face,” said Mateusz Tomaszkiewicz. “The pacing is super important, because you can have a great story, but if it has too many dialogues or cut scenes it can drag. So you need good pacing, and that was a big part of our job in this process.”

  The workload ahead of them was overwhelming, and with plans to ship the game in 2014, their time was limited. By some accounts, The Witcher 3’s world would be thirty times the size of The Witcher 2’s, and as the quest team started looking at early art assets and plans for the map, they began to panic. “The first time they showed us the scope of the world, we were terrified because it was such a huge landmass,” said Mateusz Tomaszkiewicz. “And since we didn’t want filler content, we had to provide some worthwhile content in those regions so they were not empty, because that would be terrible.”

  During these early design sessions, Mateusz Tomaszkiewicz and other designers enacted a simple rule: don’t make boring quests. “I called them ‘FedEx quests’—the quests that are just fetch quests,” said Tomaszkiewicz. “Someone says bring me the cup, or ten bear skins or whatever. You bring that stuff to them and that’s it. There’s no twist, no nothing. . . . Every quest, no matter how small it should be, should have something memorable in it, some little twist, something you might remember it by. Something unexpected happening.” At one point in preproduction, worried that they weren’t hitting this quality bar, Tomaszkiewicz cut around 50 percent of the quests they’d sketched out, “first of all because I thought we had no time to actually do all of them in time, and second of all because I just used this as an opportunity to filter out the weakest ones,” he said.

  They knew the way to make The Witcher 3 stand out was to subvert people’s expectations. One early quest, called “Family Matters,” would introduce Geralt to the Bloody Baron, a noble with information about Ciri’s whereabouts. To find out what the Bloody Baron knew, Geralt would have to help track down his missing wife and daughter. But as the quest continued, you’d find out that the Baron had driven his family to flee by drinking, abusing his wife, and acting like a violent piece of garbage to everyone. Now he appeared to be remorseful and apologetic. Would you forgive him? Try to help him reconcile with his family? Assist him in digging up and exorcising the demonic fetus of a stillborn baby he’d buried in the backyard? (The Witcher 3 always took things to the next level.)

  Other RPGs tended to draw strict lines in the sand when it came to morality—BioWare’s Mass Effect trilogy, for example, separated your dialogue decisions based on whether they were good or evil—but in The Witcher, there were very few happy endings, which CD Projekt Red saw as a reflection of Polish culture. “That’s how we Eastern Europeans see things,” Marcin Iwiński said in one interview.* “My grandmother survived the Second World War. She escaped a Nazi transport and they hid in villages for a couple of months. This left an imprint on every single family of a big part of the team. Although the team is very international, the majority are Polish. This leaves something within you.”

  With quests like The Bloody Baron, The Witcher 3’s design team wanted to give players tough choices. They wanted to make you question your morality, and they wanted to leave you thinking about those ethical questions long after you finished the game. In the early days, when CD Projekt Red’s writers and designers were trying to figure out how to pull off that sort of narrative complexity, they ran into the challenge that oh so many game developers have encountered: How do you determine whether a quest has an impact when you don’t even have a game yet?

  The writer Jakub Szamałek realized he was in trouble one day when he had to take a scene to get reviewed by some of his teammates. He’d written what he thought was a solid script, containing a funny dialogue exchange between hero Geralt and the sorceress Yennefer, then he’d implemented it into the game’s engine to see how the scene looked. The art team hadn’t finished Geralt and Yennefer’s models, so Szamałek had to use a couple of generic fishermen as placeholders. There was no animation or lip movement in the game yet. In the background, what would eventually become detailed homes were currently big gray boxes. Occasionally, the camera would glitch out and fly into someone’s head. There was no voice acting—they weren’t going to record dialogue until it was final—so everyone had to read and imagine what the delivery would sound like. “And you sit there and you’re trying to explain to them,” said Szamałek. “‘So listen, imagine that this happens here and Geralt makes this face and there’s a pause here, and then they say this, and then we show Geralt’s face and he winces.’ It’s supposed to be funny. And we have ten people in the room, and they’re looking on the screen, and they say, ‘I don’t get it.’”

  Szamałek, a novelist with a wry streak, had never worked on a game before The Witcher 3, so he hadn’t realized how difficult writing a video game could be. During one quest, when Geralt and Yennefer would walk together through an abandoned garden on their hunt for Ciri, Szamałek had to write dialogue that captured the characters’ complicated history. Geralt and Yennefer would snark and tease one another, but there was meant to be an undercurrent of warmth running under their banter. During early testing, it was impossible to convey that sort of subtle human emotion. “It all works when you have voice acting in, because a good actor can be both mean and warm at the same time,” said Szamałek. “But when you just see letters at the bottom of the screen and you have two grey Skellige fishermen talking to each other, it’s very difficult to get that and convince whoever’s reviewing your work that it’s going to work in the end.”

  For Szamałek and the other writers, one solution to this problem was to file draft after draft after draft, iterating on every scene as the rest of the team implemented more and more of the game. Once the game had basic character models and rudimentary animations, it was easier to tell how each scene was doing. People often wondered how CD Projekt Red sharpened the writing in Witcher games so well, especially when there was so much of it. The answer was simple. “I don’t think there is a single quest in The Witcher 3 which was written once, accepted, and then recorded,” Szamałek said. “Everything was rewritten dozens of times.”

  In February 2013, CD Projekt Red announce
d The Witcher 3, revealing the game through a cover on the popular magazine Game Informer that showed Geralt and Ciri mounted on two horses. Konrad Tomaszkiewicz and his team made lofty promises: The Witcher 3 would be larger than Skyrim. It’d have no loading times. It’d be out in 2014 and feature at least one hundred hours of gameplay. “We spoke about this cover for the game for a long time,” Tomaszkiewicz said. “We knew what we needed to expose there [was] the open world.” This was their coming-out party. By emphasizing that The Witcher 3 was a game in which you would be able to go anywhere and do anything, CD Projekt Red would show the world that Poland could make pretty good RPGs, too.

  The developers were purposefully ambiguous about the game’s platforms, because Sony and Microsoft hadn’t yet announced their next consoles, but everyone on The Witcher 3 knew they were putting it out on PC, Xbox One, and PS4, skipping last-generation hardware. This was a big gamble. Some analysts believed that the PS4 and Xbox One wouldn’t sell nearly as well as their predecessors, and most publishers were insisting on making their games “cross-gen” to reach the largest possible audiences, as EA and Activision did with Dragon Age: Inquisition and Destiny.

  CD Projekt Red knew that the older hardware was too limiting for their goals. If they had to restrict The Witcher 3’s memory for last-gen consoles, they wouldn’t be able to achieve the level of photorealism that they thought they could hit with this game. CD Projekt Red wanted to build a world with a functioning ecosystem and day-night cycle, with elaborate cities and grass that swayed in the wind. They wanted players to be able to explore the entirety of each region without waiting for parts of the game to load. None of this would be possible on the PlayStation 3 or Xbox 360.

  Not long after announcing the game and showing their demo to Game Informer, CD Projekt Red’s engineers made a drastic overhaul to the rendering pipeline that changed the way graphics would appear on-screen. The good news was that it made everything look significantly sharper, from the wrinkles on a leather bag to the characters’ reflections in water. The bad news was that to make it work, the artists would have to change almost all the models they’d already developed. “This happens fairly often,” said Jose Teixeira, a visual effects artist. “A big feature gets worked on, and if it’s deemed important to the game—if the game will really benefit from it—then even if it is a big change and assets have to be redone and whatnot, it’s worth doing it.”

  Teixeira, a Portugal native who was primarily responsible for visual effects like weather and blood splatter (“I had a very questionable browser history”), spent a great deal of time not just creating graphic enhancements but researching and experimenting with new technology that would make everything look better. He also had to find ways to optimize that technology. A single small village, for example, might have dozens of different light sources, like candles, torches, and campfires, all sucking away at The Witcher 3’s memory budget every time they flickered. “We had to work a lot with what we call levels of details,” Teixeira said. “The further away things are, the less detailed they are. So we had to work very carefully to make sure that we’re not, for example, using a really ridiculously complex particle system that’s being spawned a mile away.”

  As development proceeded, The Witcher 3’s designers again began to panic that they didn’t have enough content. They’d promised the world that this game would take at least one hundred hours to play. It might have been a ridiculous number, but it was the one they’d promised, and they felt compelled to hit it. “We had this feeling that the game might be too short, that we might not deliver on one hundred hours,” said Mateusz Tomaszkiewicz. “And we were like no, we told everyone, we have to do it, we have to deliver one hundred hours.” Tomaszkiewicz and his department kept churning out quests, trying their hardest to follow the edict that none should feel too FedEx-like, but the fear of emptiness lingered.

  One fundamental conceit of The Witcher 3 was that Geralt would be able to own and ride a horse, which he called Roach. (A running joke in the Witcher book series was that Geralt gave all his horses the same name.) Horseback riding was not only significantly faster than walking, it was the main method of moving through The Witcher 3’s various regions. As a result, every area of the game needed to be big. Really, really big. Every day, the game’s world would get bigger, requiring more and more content from the quest department. “We knew we wanted an open world, and we knew we wanted to have horse travel; we wanted to have a very big, realistic scope,” said Mateusz Tomaszkiewicz, “and this basically caused the locations to grow and grow.”

  The most significant issue, and the cause of the most anxiety for the quest department, was how many of The Witcher 3’s core mechanics remained unfinished. Combat wasn’t entirely complete yet, so the designers had to put Geralt into “god mode” during their testing, which meant he’d defeat every monster in one hit. This made it tough to judge whether quests were paced properly. “[It’s hard] to estimate precisely how many hours of gameplay you will have when you don’t actually have the gameplay balancing and mechanisms in place yet,” said Tomaszkiewicz.

  All these unknowns added up to a game that nobody knew how to properly judge. As Mateusz Tomaszkiewicz put it, “It’s super hard to do quality reviews when you can’t see the quality.” And throughout 2013 and even into 2014, as CD Projekt Red sat through reviews and tested out various parts of the game, they remained nervous that The Witcher 3’s world felt too empty. They had pulled off some incredible technical accomplishments thanks to the new graphics renderer, and it was common for members of the team to gather around and gawk at how pretty everything looked. The foliage looked like it had come right out of a Polish forest. Geralt’s ringmail was so elaborately detailed, you could identify each ring of metal. The leather looked sufficiently leathery. But the world was enormous, and the whole team grew anxious that they wouldn’t be able to live up to that nagging promise of “one hundred hours.”

  Each of CD Projekt Red’s major regions was bigger than most open-world games, and The Witcher 3 had three of them, plus smaller areas like the keep of Kaer Morhen and the prologue area, White Orchard. To flesh out this world, a team of level designers combed the map and placed landmarks that they called points of interest (POIs) throughout each area. Each of these POIs would hold an activity ranging from minor (a group of marauding bandits) to major (a merchant wants you to find out who killed his assistant). Some POIs led to dumpy villages; others took you to ancient ruins filled with monsters and treasure.

  The level design team was responsible for ensuring there were enough POIs to fill up the entire world. “We wanted to get an idea of the scope first: How many POIs feels good, how much is too much?” said Miles Tost, a level designer. “We’d just place them on the level. And then we would take the barely functioning horse and ride in between them and actually measure the time it would take, and then we would go, all right, so every minute we have a POI or whatever. We would even reference other games, like Red Dead Redemption or Skyrim, check how they did it, and then we wanted to figure out, all right, does this feel good for us, do we need it denser, or vary from that.”

  Tost’s team didn’t just want The Witcher 3’s world to feel like a collection of disparate quests—they wanted it to support an entire ecosystem. There would be a village that built bricks, connected to the city of Novigrad via elaborate trade routes. There would be manufacturing, agriculture, and all the other accouterments of a realistic medieval world. “If you look at the farming areas around Novigrad, all of these exist to support this huge city which believably could exist in this world,” said Tost. “All of these people living there, they get support from different kinds of infrastructure around the world. This was something we put a really big focus on creating. There’s even a village where they make carts.”

  As The Witcher 3’s staff grew larger, this insistence on realism led to some complications. At one point, Tost’s team noticed a serious problem in Velen: there was too much to eat. “Velen was always sup
posed to be this famine-ridden land,” said Tost, “where people don’t really have a lot of food.” For some reason, though, an environment artist had stocked up many of Velen’s homes, filling the cabinets with sausages and vegetables. It bothered the level designers too much to leave as it was, so they spent hours digging through every village in Velen, taking food away from the people like twisted reverse Robin Hoods. “We had to go through all the houses in this area and make sure there was barely any food,” Tost said.

  It was this kind of attention to detail that CD Projekt Red thought could elevate The Witcher 3 above the competition, not unlike Naughty Dog’s approach on Uncharted 4. Most players might not have noticed the amount of food in a Velen cupboard, but those who paid attention would be rewarded. There was something special about realizing that the developers of a game actually took the time to make it so tree branches rustled and crackled when it was windy, and so the sun would rise earlier and earlier the farther north you went.

  Of course, one of the problems with being a studio full of perfectionists is that you spend too much time on the little things. “From a player’s point of view, you could almost never have enough detail,” said Tost. “People like to explore a lot, and you can totally understand that. But of course at some point you have to consider the scope of the project, and whether this rock angled in a certain way actually adds something to the world. Or maybe you should instead go and fix another two bugs somewhere.” Every second spent adding an extra detail to the game, every minute spent fine-tuning quests, and every hour spent wrangling with the unfinished world led to the schedule progressing faster than CD Projekt Red would’ve liked it to.