Try to imagine an alternate universe where, instead of playing through a rollicking action-adventure with affable treasure hunter Nathan Drake, Uncharted took the form of a post-apocalyptic survival horror game. Or that instead of playing "Uncharted" -- which was about finding El Dorado, the mythical city of gold -- you instead played a game called "Zero Point," focused on the concept of dark energy. Those were some of the ideas that the developers at Naughty Dog threw around at the conclusion of Jak 2's production. Yes, Naughty Dog had already started prototyping ideas for their first "next-generation" console title way back in 2003; though, Naughty Dog co-president Evan Wells elaborates, "When Jak 2 closed, that's when we started thinking about it. But then the PlayStation 3 got delayed, and we back-burnered all those ideas for about a year, and picked it up again right after Jak 3."
Two weeks before Uncharted 3: Drake's Deception hits shelves, I find myself in a conference room in Naughty Dog's office in Santa Monica. For two and a half hours, I get to hear developers discuss topics such as the creation of the original game, and some of the major hurdles in creating this next installment. Sitting down along with Wells is fellow co-president Christophe Balestra, creative director Amy Hennig, lead designer Richard Lemarchand, cinematics animation lead Josh Scherr, and actor Nolan North.
Hennig briefly shows me the title page of the May 2005 concept document for "Zero Point," while Scherr describes that pitch as focusing on "zero point energy, eco-terrorism, dark energy, and all sorts of crazy nonsense." He continues, "At one point we wanted to set the entire thing in a vast underwater facility or city type of thing. We didn't do that, and when we saw BioShock come out, we breathed a sigh of relief and said, 'whew, dodged that bullet.' So obviously it was a good idea: just not for us." Even when Zero Point transformed into Project Big (the internal codename for Uncharted: Drake's Fortune), the underwater base was planned to constitute about 75-percent of the game, with the tropical island environment serving as the first 25 -- a far cry from how Uncharted ultimately went 90:10 in terms of island-to-base ratio.
FROM ZERO POINT TO PROJECT BIG
Hennig points out, "What's interesting is that the fundamental idea has always been the same: third-person action adventure. [We had to think] about what's the genre, and what are the pillars: gunplay; brawling hand to hand combat; action stunts; chase sequences; big action setpieces; and partners and allies. This has been our roadmap ever since. What we do now is say, 'We got 10 percent of the way down our goal, and now we're 50 percent of the way down that goal.' It's [all] to the service of saying, 'how do we best replicate that action adventure movie experience?'"
Once the team solidified the Project Big design document, in late 2005, they created a proof of concept video to pitch to Phil Harrison (then the president of Sony Computer Entertainment Worldwide Studios), a portion of which is reproduced below. Wells comments, "A lot of the stuff we were seeing pitched, from Sony studios as well what we were seeing being developed on competing platforms like Xbox, was all sort of post-apocalyptic with lots of browns, grays, and muscled space marines. So how do we stand out? When you're walking through E3 with everything blasting at you, what's going to draw your eye? Well, let's use color. And we wanted our character to be relatable and something that people can still aspire to be, but not in a superhero unrealistic macho jerk style."
Wells states some aspects of the pitch video above have finally been implemented with Uncharted 3. Lemarchand points out the specifics: a melee combat system that allows for heavy use of contextual animations, multiple opponents, and improvised weaponry were all ideas that the team wanted to do for the first game, but ultimately didn't get implemented to Naughty Dog's satisfaction until the third game six years later. He continues, "We didn't see [some of these gameplay] systems for months and months, and it's all right here. We had to explain it to ourselves through this animatic."
Yet once the game got greenlit, that was only the beginning of some of the growing pains the studio had to go through during Uncharted's development. Balestra elaborates, "At the time, we just came off from working on PS2. We had to learn what we were doing on the PC, with all the shaders and other things. We had no idea, and so we had to catch up on many years of technology from working on just the PlayStation 2. We had to rewrite everything from scratch and change the way we were working. It was really bad for the team, because people were kind of lost. There was no transition from what we used to do on PS2 to what we would do on PS3. I'm sure of all the stuff you see here [referring to the pitch video], not one line of code has made it to Uncharted 3. It was tough; it was a really difficult process and we lost a lot of people. But it made the team stronger. The people who could not deal with the situation were probably just not good enough to deal with what we wanted to do anyway."
He continues, "We started working on the PC, and we had no expectation about the limit of the hardware -- we felt like it could do anything. So a lot of people were very disappointed when we got the hardware -- a lot of 'What, that's it?' -- I mean, we just made stuff up. We thought we were going to have motion blur everywhere and infinite memory. And on top of that, [when] you're working on a giant game where you know what you're doing -- you can be productive for every minute. And suddenly, you're now trying to do the simplest thing and it doesn't work. Tools break, and you're also trying to figure out what kind of game you're making. It was really rough."
Hennig adds, "Anytime you change hardware, there is that big transition. But this was a real sea change, like Christophe was saying, because the old way of doing things -- thinking of polygons and textures -- you have to think about things more like film or high-end TV. Things like lighting and normal maps and how things interact with each other. You had to relearn everything."
Wells goes on to describe how just working on the technology affected the way Naughty Dog approaches design during the production process: "It was very very difficult, because we didn't have the tech to support the experimentation that design required. We had lots of great ideas that we made documents about, but they're just documents. That doesn't get you anywhere; you got to try things out." As a specific example, he notes, "So we shipped in 2007 -- up until January or February 2007, we had a lock-on aiming system. And around that time, we tried manual aiming, and that changed the flavor of the game in a good way. We might have done experiments with that before, but it wasn't until then that it became a core part of the game. There was a number of things: we had an underwater swimming mechanic around the same time that we cut. We wanted to do a lot of things with fire also. So we had this long list of things we wanted to do, and it wasn't until about probably nine or ten months before shipping the game that we had enough of an engine or animation system or character control system that you can really start the experimentation process to refine what's going to stick and what's going to go."
To which Hennig adds, "And then you have nine months to make the game." She continues, "We did a lot more paper-mapping, design-wise, back then; since you couldn't quickly visualize or test ideas in an engine. But a lot of times, it was just drawing maps and thinking about architecture."
FROM GAME-GAME TO ACTION MOVIE GAME
While Hennig, Lemarchand, and the rest of the content and design folk were paper-mapping and waiting for the tech to be in a comfortable enough place for proper iteration and production, Hennig then turned her attention to the casting and cinematics. In August 2006, Naughty Dog conducted the initial auditions and callbacks, before locking in cast members such as North, Emily Rose (Elena), and Richard McGonagle (Sully). Hennig's aim for casting Nathan was as follows: "We knew it was going to be hard to cast the role of Drake, because we wanted that intangible quality that beloved actors of the genre had, like a Harrison Ford. And the reason he has the appeal, staying power, and relatability is because he's vulnerable. Most people we brought in played up the tough guy or the snarky funny guy. But you can immediately see some flawed vulnerability, doubt, just in the simple audition. In the audition you did, you got that this guy was the protagonist hero of the game, but he could easily be surprised by things, taken aback by things, and hurt by things. Nolan understood that."
North recalls how, traditionally when an actor records dialogue, they just do so isolated within a recording booth with a complete script. For Uncharted, he found himself in the same room with the other actors, and they had to actually act with each other. Another difference that Scherr points out: It's important to use the same actors for both voices and physical performances.
Wells explains, "The way we did [things] for the first game was very different than how we did it for the second and third game. At the time, [motion capture] was really new, and no one had really done a performance capture process yet. We were shooting in a warehouse with a mo-cap stage setup, and the audio capture there was not good enough to go into the final game. But we liked all the actors, and we wanted the audio as reference so they would come back and ADR [Additional Dialogue Recording] it later."
Another lesson that the studio learned from working on the cinematics and motion capture of the first game: Animators only need the actors to work with a close approximation of whatever object the actor was involved in. That is, if Nathan Drake is driving, then the animators only need motion capture footage of North sitting in a large box with some PVC piping serving as his steering wheel. Wells and Scherr relate how Naughty Dog built this elaborate set with a replica airplane and a rented Land Rover for an early cut-scene in the first game.
Wells details, "The evolution from Uncharted 1 to 2 is that we worked with [motion capture studio] House of Moves to build a sound stage around our mo-cap stages, so that we can mic the actors up and capture audio on that same stage to get that one-to -one match. For the third game, we were fortunate enough to have Sony build us our own dedicated motion capture studio over in Culver City. That was great because not only did we have the ability to capture performance and audio at the same time, but that facility was dedicated to us. We probably shot close to twice as much mo-cap as we did for Uncharted 1 and 2."
That's how it went. Naughty Dog learned how to establish a new production pipeline, had gotten the technology under control, learned how to capture the cinematics, and were then able to make the design, technology, and cut-scenes work together. Project Big in 2005 transformed into Uncharted: Drake's Fortune in the fall of 2007.
CHARTING DRAKE'S DECEPTION
So you'd expect that after the growing pains of Uncharted: Drake's Fortune, Naughty Dog would then refine the development process on Uncharted 2: Among Thieves and breeze through the creation of Uncharted 3, right? Surprisingly, that's not quite how it went.
Early on, things went as expected. The team had several meetings about what to do for the next game soon after Uncharted 2 hit shelves. Lemarchand remembers that melee combat was the primary design topic for the meetings that the team had in November 2009. The expanded melee combat the team had wanted to implement in the first game had become a priority. Other aspects of the game were established as well, ranging from specific setpieces that the designers wanted to create -- the cruise ship came from an idea of having a level that rotated around the player -- to actionable and reasonable requests from the fan community. The idea of delving into Nathan and Sully's bromantic friendship was in response to a frequent fan complaint that Sully was barely in Uncharted 2, for example.
Sometimes, Naughty Dog personnel mingled with other developers. Wells reveals that one of these meetings served as the catalyst for Uncharted 2's multiplayer: "After every game, we go around and visit different studios. We visited Jason [West] and Vince [Zampella] over at Infinity Ward, and they were very influential in us doing multiplayer in Uncharted 2, because we went and visited them a couple months after finishing Uncharted 1. In fact, we visited them after having a meeting with Phil Harrison at the time where we said, 'Oh yeah, we're not doing multiplayer.' But we went to visit them, talk to them, got inspired, and now it's one of the most important features in the game."
UNCHARTED 3: I'M THIRSTY
After getting a handle on the core principles of Uncharted 3, the first major development hurdle arose: The announcement trailer for the Spike Video Game Awards in December 2010.
Wells says, "So this concept came to us around the end of summer. The only reason we knew that we were even going in that direction wasn't because we had the entire story laid out -- in fact, the story was still very skeletal at that point. But right after Uncharted 2, a couple of designers and animators started playing around with ideas about epic setpiece moments for Uncharted 3. And that's where the cruise ship came from, and some others had this cargo plane sequence. So we knew that we had a gigantic cargo plane, and that it would have to crash at some point, and we knew at some point the story was going to take us into the desert. So we knew that there would be a crashed plane in the desert. We thought, 'that'd be a pretty cool announce for the game: Show this crashed plane in the desert.' That's where we started building this, but we still didn't know where he was going to walk to. How does he get out of the desert? Why does the plane crash? Why was he on the plane in the first place? These were things that some people had ideas for, but nothing concrete, and we just took a leap of faith. "
Scherr comments, "Part of the thing with the announce trailer is that, we were starting to work on our sand technology and our sand dunes, and as it turns out, nailing the look of that stuff is effing difficult. Maybe five or four days before we made the final images for that thing, we completely and utterly scrapped the sand dunes in those wide shots you see, and remade them from scratch because they weren't up to snuff."
Above is a comparison between what Scherr calls, "the 50th iteration" of the announcement trailer, and for reference, he considers the final trailer "version 149." They point out that this early version was serving as a sort of proof of concept, to nail down the pacing and general beats before going into the polishing phase. Lemarchand notes, "I remember seeing this video from this guy John Stevenson, the co-director of Kung Fu Panda, giving a talk at the DICE [Design, Innovate, Communicate, Entertain] Summit, and he has this brilliant philosophy that we already kind of used here: 'Show it when it's s***ty.' You have an idea, make something quickly, and as you do, it starts to become concrete. Then you show it to someone, and they tell you how to make it better. And that lets you take the next step." Hence, they mocked up that version of the trailer after 50 tries, and then took another 99 passes before the studio considered it good enough to show to the world.
The announcement was probably also the main "active development: go!" signal. Wells notes, "As soon as you show something publicly, then s*** gets real. You think you have everything prototyped; you go, 'We got this environment and that environment going, and it's making progress.' But as soon as you go, 'OK, people are going to start judging us on this,' then you realize that people's expectations are that it looks as good as Uncharted 2 when it shipped, and you have to do that 12 months before you're ready to ship. But it's got to look like it's finished and look as awesome as the last game. So that's when it gets frightening. It takes so much effort to prop it up and hide the rough edges."
Balestra adds, "And we announced the date. That was scary." He remembers, "I think, 'Yea, sure, we'll make it.' Then we get closer and I wonder, 'What did we get ourselves into?'" Wells notes, "Especially when we see Mass Effect 3 go from 2011 to 2012, and we don't have that luxury."
FROM PLANE TO DESERT TO BURNING HOUSE BACK TO PLANE
The press demo immediately after the VGA awards (and the source of our first preview) also indicates how flexible Naughty Dog has been in U3's development. Wells describes, "The very next day, we were going to demo the gameplay to a bunch of journalists. Up until the beginning of October , we wanted to show the environment we were going to: the desert level. So the most recent level we showed [in October 2011], Desert Village, was what we were going to debut with. We wanted to show off our awesome sand tech and establish it; show that we were going from jungle to snow to sand."
He continues, "But in October, we looked at Desert Village, and thought, 'This isn't going to impress anybody.' The sand's not looking good, and we haven't even modeled the dunes correctly. So we had to shift gears and look at what else was cool. Fortunately, the Chateau level was coming along well, and right there, six or seven weeks before showing it, we decide to shift gears."
Here Scherr mockingly interjects, "Time to make lots and lots of fire particles!"
Hennig points out moments like that called for frequent re-allocation: "A lot of people and resources were told, 'You were on this, now you're on that.'"
The announcement trailer and its creation-in-a-vacuum, along with the switching around of what the first gameplay demonstration would be, were habits picked up from Uncharted 2. Hennig elaborates, "That's what we do. We don't say, 'We're going to go off and figure everything out and just make it.' You can't. This business is based on software development; it's not like the film business. It's an employee-based business, not a contract-based one. So we just start sticking pins in things. We commit to some things and say, 'Okay, that's in the game.' Just like the train in Uncharted 2. Early on, we didn't know where it was going or coming from, just that it was there. And that it crashes."
Scherr recalls, "We had brainstorming sessions where someone says, 'I want to do a train level' and someone else goes, 'How about having it hang off the edge of a cliff?' When we were exploring ideas for a teaser at the VGA Awards that year, the thing we came back to was, 'That would be a great way to tease the game.' At the time, we didn't know that it would be the beginning of the game -- we just knew it happened."
It sounds a bit haphazard and scatterbrained, but Hennig believes that it ultimately led to the strong story and script that Uncharted 2 gets lauded for. She says, "With us, we know that it matters why he's on the plane, the train, or the boat; what happens there, how he gets off, and all that kind of stuff. So we take the time then to say, 'Yes, that is an epic set piece that we absolutely have to have; now let's make it relevant.' And spend the time talking about how it fits into the story, how should we weave it through, and where does it belong. It's the story that accretes or coalesces; as opposed to this authorial experience where you go off to write something from A to Z. It just doesn't work that way for us."
North recalls that rather than shoot scenes with full scripts, the actors would frequently just have chunks, and they would go with it. "It was like performing Hamlet, but when you ask in Act 1, 'So what happens to this guy?', the director tells you, 'Not sure; we'll find out.'" This still happened for Uncharted 3; Wells admits, "For the first nine months, we didn't know if either of the two girls [Elena and Chloe from Uncharted 2] were even going to be in the story." For reference: Hennig notes that the first cinematic was filmed in July 2010, and the last shoot was in July 2011 -- most of Hennig's script just came as she and the actors went through their year of shooting.
Wells provides another example of Uncharted 2's seemingly scattered and ever-changing development: "The Sanctuary level was the monastery in the mountains where we were going to show off a snow level and all of our new snow tech. We came to about the same conclusion in the same timeframe: If we wanted to show off snow and how kick-¤¤¤ our snow tech was, well, it wasn't ready. So we shifted gears and showed off the Nepal level." He jokes, "I don't know why we didn't learn our lesson," as Hennig adds, "We somehow always pick the wrong level."
If anything, going through this process twice has given Wells a new takeaway: "Basically, if we want to impress with our new tech, it's not going to be ready twelve months in advance." But he stands by the shifting decisions that the team has made to date: "The sand did turn out awesome, and the Desert Village level is awesome now, but they were not ready to show [within the original timeframe]."
THE PRESS CONFERENCE AND THE LAST MINUTE TRAILER
But it wasn't smooth sailing, as E3 2011 then presented another significant hurdle in Uncharted 3's development. Wells reminisces, "So we knew that we were going to be presenting on stage, and we had big shoes to fill from Uncharted 2's rooftop collapse sequence in 2009. So we said, 'Okay, the cruise ship demo is pretty huge and awesome.' About a month before -- we were also developing the cargo plane sequence for a behind closed doors demo, so we were prepping these two simultaneously -- we thought, 'Oh shoot, the cruise ship level is getting too long.'"
Balestra adds an additional wrinkle: time on-stage. He elaborates, "Early on, Sony was like, 'Oh yeah, you have 10 minutes on stage.' OK, cool. Then we walk to them, and they go, 'Oh no, it's going to be five minutes.' Then later again, they tell us, 'It's going to be two and a half minutes now.'"
During the "you have five minutes" window, Wells noted that the team didn't think they could do a five-minute version of the cruise ship demo, so they switched over to finishing up the cargo plane demo. Yet again, after working on the cargo plane demo, the team switched emphasis on having the cruise ship serve as the E3 stage demo again. Hennig remembers, "One of the other things we defined was that to show the cargo plane sequence, we also would have to show the cinematic between Drake and Elena." Wells continues that train of thought, "We want to show epic sequences, but also that warmth and personality. So we wanted to make sure we could also include a cut-scene. So now we're barreling towards E3 and switching our demos [again]." The team decided that, by cutting out certain sequences, they could get the cruise ship demo to fit within the allotted stage time.
That's just for the on-stage demo. Wells recalls another development wrinkle: "So about three weeks out, Sony also said, '3D is a big push for us, so the demo is in 3D, right?'"
He remembers responding, "You're out of your mind. We operate this way: we polish for 2D and then it takes another three-to-four weeks to optimize for 3D -- to make sure it's not crashing, and performing at a decent framerate. We're barely crossing the finish line in 2D, much less in 3D." He admits that the 3D issue, "became a point of contention, and we're talking with executives, and we said, 'You know what, what if we make you a 3D trailer?'"
As stressful as adding a new trailer on top of two E3-specific demos was, at least it was a manageable request. Wells admits, "We had a trailer that had lots of little snippets, and we literally just pulled those snippets just for the trailer. They were not complete in any way." Scherr adds, "Yea, we would have things like, 'Well this scene takes place in this part of the castle, but it's not built yet, so let's put it in this other room.' We just smoothed over that stuff. And there were a few characters that were just finished in time to go into the trailer. We pulled people off of scenes they were currently working on just to get those little snippets in the trailer done."
The 72-hour period before Sony's E3 press conference was one of the most stressful in the company's history. Wells reflects, "We're working on a trailer and the two demos of cargo plane and cruise ship, and things are just not done. I was having panic attacks and thinking of calling up Sony and saying, 'We can't do it.' We had never been in this situation before. The press conference was on Monday at 5pm and we had put the finishing touches on the code [for the demo] on Monday at 7am."
THE FINAL PUSH
Balestra reflects on how the project schedule somewhat followed the close timing of the E3 press conference: "[For] this project, everything ended up being done at the last minute. You're talking about the difference between Uncharted 2 and 3; this one was more tight in terms of getting things done and shippable." Even when it was time to stop working and submit a disc to Sony, things would crop up.
Wells relates, "We were three days from gold master -- a [few] weeks ago -- and our lead programmer comes in with Christophe and [game director] Justin Richmond, and they shut the door. I'm like, 'Why are we having a closed door conversation so close to gold master?' He sits down with the most depressed look on his face, and he says, 'Guys, I took the game home, and it's a mess.'" The problem: the game performs just fine -- for the first half. After reaching the halfway point, numerous bugs would crop up. Objects would disappear. Walls would flicker in and out of existence. Nathan could find himself in a hall devoid of anything -- geometry, texture, lighting, etc.
Wells continues, "All of these bugs point to the exact same problem in our streaming system. We are streaming stuff constantly; we're abusing the PlayStation 3 like a bad child. We're streaming audio, music, animation, video, levels, textures, everything. We're filling the memory, and about halfway through, it gets jammed up so that when we ask for a texture, it's not there; we ask for an animation, and it's not there." Balestra interjects to note that the reason this bug came up was because the programmer happened to play the game on an older test unit -- most of Naughty Dog's Quality Assurance team were using newer debug hardware and hence weren't running into this issue. The developers realized that a lot of fans still probably play games on launch-era PS3 systems, and this bug could end up ruining a lot of players' experiences.
Balestra remembers having an intense two-hour-plus discussion with lead programmers Travis McIntosh and Christian Gyrling to nail down exactly what triggers the game's meltdown at the halfway point, and they concluded that somehow, the streaming system was causing the PS3's hard-drive to fragment, which therefore led to lots of seeks when requesting data, and said seeks would cause the "traffic jam" that Wells described earlier. Wells notes, "Even though it's literally past the day that we told Sony, 'We're not changing the code anymore, trust us,' we went in and changed the most fundamental and frequently called function in the game."
The lead programmers made a tweak to the code and then burned a new disc with the revisions. Then they had two testers play the old and new code side-by-side; one tester played the old code on a relatively old system while the other tester played the freshly fixed version on the oldest PS3 in the office. Wells finishes, "Right away, we notice that one system is doing better while the other is pretty bad. There was this [specific] threshold we were looking for, 3000, and we knew that was when things would get bad. The number kept creeping up and up, and right when both systems hit 3000, we saw that one guy would go around a corner, and nothing in the world existed -- while the other guy went around the corner, and everything was fine."
Two days after that, Naughty Dog submitted the Uncharted 3 gold master for active production. And based on their track record (recall that the studio has always shipped a game in two years), Naughty Dog personnel are probably already having meetings about their 2013 title.