Designing the Museum Flashback: 'The Last of Us Part II'

Super excited to be here everybody, but yeah. So first off, we're going to talk about The Last of Us Part II and specifically about designing the museum level. We're going to cover its process from concept to composition through to production. First off who the heck am I? My name's Evan Hill. I've been a level designer for about 10 plus years. Obviously, I recently worked at Naughty Dog on The Last of Us Part II. Currently I'm at Obsidian working on Outer Worlds II. Also worked on a couple of other things, but main focus for today obviously is going to stay firmly on The Last of Us Part II. So it has a reputation for being a pretty brutal game, really focused on its combat, on its stealth, all these other

really hardcore elements. But I'm also really happy with how well we were able to display narrative and very human moments throughout the game. This is going to be focused on one of the levels that I think let us push that the furthest. You know, for example, shout out to Gene Park. This was his reaction after playing this level. I don't apologize for possibly ruining his shirt, but yeah. Yeah, this level connected with people particularly well. We got a lot of fan art, tattoos, and the like. And hopefully with this talk we can kind of dig into some of the elements that really drove this connection with people. So yeah, that's what we're going to be covering today. This level that we designed. What's it about? Uh So let's just start with the concept. This level takes place 3 years before the main events of the game at the height of Ellie and Joel's relationship. Uh it's Ellie's birthday. Joel is teasing her with this surprise.

Just constantly just dangling it over her. She tries to guess and it's focused on the interplay between the two of them. On him delivering this amazing birthday surprise. Uh, and I'm going to lead you through how we developed these specific sequences taking you from this super positive element of the relationship all the way to where it starts to unravel. They start to fall apart and then we see where the rest of the game goes from there. Um, so yeah, that's just the concept layer though. The real meat of this talk is going to be on two parts, side A and side B. We're going to cover composition, which is how we conceptualized all of these different elements. How we actually thought about making this game and how we kept evaluating it as it was continuing to be made. Uh, you know, the real design heavy stuff. But that's only going to be part of the talk. The other part is production. How did we take this from idea to a shippable product? And specifically from the lens of me as a level designer

kind of being in the thick of it all. We're not going to necessarily focus on hard deck technical details but just the actual motions of production. So composition is going to be about 2/3 of this talk and production is going to be the last third. We've got an hour. Hopefully I'm going to be able to fit all of this in. Uh, we're going to try but let's jump right into it. So first off, side A, composition. We're going to start with the first track, which is this really hot take of mine that is everything is story. Everything. The reason for this is because every game is experienced one second after another. I don't care what you're talking about. You still live through the moments that you play a game and as a result you see a plot. You build one with yourself even if there's not an explicit one. Playing a game is a narrative experience. Even without a set plot, something like Guilty Gear can give you narrative

fulfillment, be it in the structure of a match where you're just making it through by the skin of your teeth to longer-term goals and longer-term experiences like improving with a character. I see these things as deeply narrative and they're emergent from their systems. But, you know, the same thing can be applied to something like Valorant. That same story that you're experiencing through the matches, the story of progression that you're telling yourself. This is what applies to systemic games like RimWorld, Crusader Kings, Dwarf Fortress. The main thing to understand with this is it's not about what games do. It's how we as humans, as a species, deal with time. So, stories themselves are these structured alliances of experiences. And we normally think of things like movies and books, but just look around us. We have SportsCenter. We have, you know, the same kind of thing I was just talking about, but for athletic events. We have history.

So, it's a cool idea, but how does that help us? So, the main thing that this lets us do is three things. Interesting moments share the same anatomy. So, we're able to analyze any type of thing that's not traditionally story using the same frameworks and kind of push them around. For games, this is really essential because we're so multimedia, we're so all over the place in terms of the ways that we engage players and the ways that we push things towards them. It also centers the player's experience. Again, we're not talking about what my goal was, what the team's goal was, what the meta-commentary is. It's entirely focused on the individual or the group that is experiencing your story, your medium, your content. They're at the center of it and it lets us focus on that. And the third thing is this benefits immensely from iteration. Like stories themselves are these kind of whittling down affairs when you're making them.

And you know, it's how we got to make things like Last of Us as good as they are. Just trying and trying again, layering and layering again. So, really quickly, let's kind of break down what I mean by all of that. All interesting moments share an anatomy. So, I actually did a talk on this a little more thoroughly yesterday. You guys can check it out on the vault. But um the basic way I like to look at it is using this methodology called Kishotenketsu. Going to go really fast through this. You basically have a setup, you have a development, you have a twist, and you have a payoff. This is how jokes work. This is how most story beats work. Again, how a football game can even work. It's just a analysis of like what makes things interesting. You set expectations, you twist them, and then you pay off the consequences. The Mario games uses quite a bit right now. Uh Mark Brown has done an amazing video just breaking down Kishotenketsu for about 15 minutes. Uh we're not going to cover that here, but really recommend it. Um and yeah, centers the player's

experience. They're the center of this universe. Everything we do is to affect their subjective top moment-to-moment experience in this. Benefits from direct iteration. Story is writing. Good writing is rewriting. It lets us kind of break ourselves out of this idea that there's a right way to do things or a perfect version of stuff. We're just layering and layering and layering. Uh so, for some good resources on this by the way, I'm going to cover this in each section. I'd recommend Manga in Theory and Practice by Hirohiko Araki, Robert McKee's Story, and The Visual Story by Bruce Block. Don't worry, these will be back at the end. Cool. Uh that was a that was a good idea, at least I think. That's why I talked about it. But what about the level itself? So, goals and acts. These are the roots of the composition. Um So, we're trying to do this thing. We're trying to make a story. How do we actually make

sure it accomplishes everything else that we need this to do? What are we even trying to experientially accomplish? Um So, we kind of started with this at a high level before anything else, before a pixel was pushed, a vert was placed, or sentence written. Uh we knew we wanted to show the high point of Ellie and Joel's relationship. We wanted the outcome of that to reinforce the conflict between them. We wanted to create uh compelling gameplay without combat. We want to have emotional impact, and we also want to teach the player how to swim. No big deal. Uh But the next step is then we have to find moments to meet our goals. What's this rough uh evaluation that we're then finding to accomplish all of this? Uh So, what's the story? Is really what we're asking. What sequence of events can do all of this and still be interesting? You know, we're not just doing a checklist. So, we broke this into three different acts at a pretty early stage using the same kind of framework I just covered. Setup,

development, twist, payoff. Three really simple acts. Break down the story as quickly as we can. First was hike to the surprise. We come in 3 years earlier. Joel's teasing Ellie about whatever the surprise is. She gets annoyed. He shoves her into a river. And we suddenly teach the player how to swim. So, again, we set it up, we developed it, big twist, you're in river. Oh, Ellie can swim. Great. I've taught you. We're now able to move on. They move further. She shoves him into the river, gets him back. And as they're continuing on, you know, this tension keeps building and building until the big reveal. He's taken her to a natural history museum. Blows her tiny mind. You've You even as the player you've mostly just been focused on that back and forth between the two of them. You forgot pretty much about the surprise itself. But then we get to the meat of it. Exploring the museum. Uh we

enter into this new space, this seemingly impossible thing just out in the middle of the woods as far as we knew. And they bond. They get to have this just frankly magical moment where, you know, everything just melts away. They just get to be. They go to the space area. Joel doubles down, shows that his surprise wasn't even done yet. He's recorded a uh recording of the Apollo missions. Which again, internet's gone. Nobody's ever heard about this uh for a while or at least been able to hear it. Ellie's ecstatic. This moment itself, like you can look at it as that high point, as that moment where they're at their best. And then they move on. Ellie's curiosity to, you know, keep the good times rolling, her drive to see everything uh gets them to check out this is other building that Joel hasn't explored. Of course, they get separated. She gets lost.

Winds up in a dark spooky place. And as you and her are you know, starting to get eeked out. Uh gets attacked by a boar. Not actually any real threat, just really shaking. Uh But in the midst of this she also finds a piece of graffiti that really catches her attention. Uh Joel busts in, see if everything's okay. You know, no threat. No infected. But something that reminds them of the other problems that they have. That uh you know, the events of the first game still happened, and that Joel is still lying to her. She doesn't know about what, but she knows he's hiding something. And this then leads to their unraveling. So, yeah. Pretty straightforward. Just a recap of the story, but hopefully you

can see how we kind of stack these on top of each other to create that rhythm, to create that pulse, to accomplish everything we tried to do and deliver it. It's delicious. It's great. Good storytelling. Nothing crazy. That's cool. But, how did we actually find and develop these moments? Like, I just gave you a list. I didn't tell you how I got there. Uh so, let's go into that. Track three, characters. The main thing to understand about our style of storytelling at Naughty Dog and mine personally is we try to work from this premise that characters drive the story's beats. Everything we just described was due to an action of these characters. Joel is the one leaving the surprise. Ellie is the one reacting. Ellie pushes forward into the annex. All of these things it's pretty basic, but they're driven by what these characters want and who they are. So, we understand the characters drive the story's beats.

But, what drives them? We need to understand the people coming along with us into the story. And well, how do we do that? We break it down. First, we do things like look at their character history. Who are they and what have they been through? This is Ellie and Joel. They have been through an entire The Last of Us together. All of that. Their trauma, their grief got resolved in the end of that. They are they are thinking that they are living their happily ever after. James done. Ellie was saved. No cure. Bit of a loss, but they're trying to move forward. They're they're both looking to the future. Since we understand these contexts and you know, this covers everything else with them. Their personality, their history, you know, Joel's from Austin. Ellie's grew up mostly in Boston. Different vibes, but the main thing that this character history gives us is it helps us understand what these characters want.

And then we are able to take those wants and drive them into actions. So, quick breakdown. Joel's main want in this scene, in this flashback, is he wants to live the life he had before he lost Sarah with Ellie. And he's he's getting to do that. He he's a dad again. He's got the daughter shaped hole in his heart filled. Ellie wants to see the world progress again. Like she grew up really really harshly and and now has experienced what she sees as like the only bit of stability she's ever witnessed. She's she's very much on the pro like, oh hey, the world can actually be good though. Imagine that. Um and you know, her love of space and science always looking ahead kind of reflect this, but then we take the next layer, which is this is what these people want on the surface. This is their most primary drive, but they also have secret wants. Joel's pretty simple. He's wants to keep hiding the truth about the Fireflies.

That man does not want anybody to know what he did. Does not want Ellie to know what he did. Ellie still secretly wants to know what happened. You know, they have these two aligning superficial goals that let them bond, let them really just become a family unit, but that core tension underneath that we get to tug out at the end shows how they unravel. How they go from there to here. Uh so yeah, the secret once come forward, unravel both of them. Joel starts to lose Ellie. Ellie fears she misses her chance to matter. She's starting to kind of question and worry. It's like, wait, what part was he lying about? Um and yeah. The best way to think about this too is that story isn't picking words, it's finding the actions that bring characters to life. So how do we go about doing that? We're not trying to sit down necessarily and type out a script or just bunker up. Uh one of the things that I really liked with our process at Naughty Dog is it

was very kind of focused on acting and improv. So we would just brainstorm, feel it out, play as the characters, use all of this knowledge and history to just come up with dumb moments. Like turns out a lot of Naughty Dog's ex-theater kids. And if you were to just like pop into the office, somebody would probably be rolling on the ground, other people would be using voices, they'd be having fake shouting arguments, but you know, can't really show that here. Um but yeah, it honestly though, like that approach is what I think got us some of these like very lively feeling moments. We didn't pre-plan a script, we just worked from the characters, tried to actually imagine what they would do in these situations. Um so yeah, if you guys want some more resources on characters and how we kind of built all of this out, uh we've got The Art of Dramatic Writing by Lajos Egri, Creating Character Arcs by K.M. Weiland, and of course Robert McKee's Story. So that's cool, but

what about the player? Let's talk about gameplay. So I've mostly just talked about what happens, who these characters are. This is a game. There's a player, there's somebody that like is in the room with us that isn't there until the game ships. So, how do we bring them into this experience? How do we involve them in this thing that we're doing? This, you know, this acting, this improv. Simplest way that I've been able to articulate this is we approached it as if the player is an actor that we cast as Ellie. That means that this person, this performer, collaborating with Ashley Johnson, basically, uh has the agency to drive her actions. They might not be able to branch and decide what those actions are. They're still the impulse. They are still the one doing this. They are acting from a script, basically. That's how acting works. You don't go on stage and do a choose-your-own-adventure play. You do your part.

Uh they are the impulse. They get to decide those still some decisions, the timing and the tone of those decisions. They get to decide the what, the when, the why. And that takes us to the next part. We also want to ensure that these decisions, these choices we're giving the player, provide emotional expression. They are either letting them facilitate or express an emotion that they think Ellie should have. So, they also get to let her explore uh her different reactions. So, I can press a button going like, "Oh, I wonder what Ellie's going to do with this?" And then get surprised, get this feedback that it's like, "Oh, she's going to make a stupid joke. She's going to get on Joel's nerves. I want to see that. I align myself with that. I'm exploring this character interactively in a way that, you know, it's not quite acting, but it's I think something enriching nonetheless. Um and then the final part is feedback. So, we have agency, expression, but the next

part is yes-anding. Is like taking what the player is trying to do and saying, "Oh, that's cool. Also, developing it, throwing it back." Again, kind of going back to this improv uh framework that we're working from. It also then drives the next expression. It drives the next decision for the player. You're actually having a back and forth instead of I don't know, filling out a spreadsheet or going on a hike. And you know, some examples of this that we can kind of cover are you are the real person to press the button to shove Joel into the water. It isn't a cutscene. It's not perfectly timed or canned. It's whatever you feel like it. Whenever you I honestly as the player uh think he's expecting it the least. And where you get that bit of extra satisfaction. Uh same thing with the swimming. It's really simple, but just this exploration of a new mechanic. Up until this point, you've not been able to swim in this game. We're about like 8 hours in. You get a new expressive way to navigate the environment, learning things like diving

under water. You can also find some secret stuff like climbing on the dinosaur, just being this reckless kid. Uh clambering up the nearest cool thing that she finds when she's excited. And getting to again, like emotionally express that. Getting to just align with the performance that she's just jazzed to be there. She's actually also kind of happy that she's screwing with Joel. Um or you know, her finding a hat, messing around. Uh going into the backrooms. Cracking jokes. These are all optional. These aren't essential to her performance, but they're things that you as a player get to own having added to the performance. You're the one who has decided that like, yeah. This is Ellie. This is who she is. You're collaborating with us. You're adding to the scene rather than just doing it by rote. And yeah, that that's just how we approached a lot of these small sections, you know, putting hats on the dinosaurs, putting the hats on Joel. Even going back to emotional expression,

literally letting you just go into a mirror and make stupid faces. Can't get any more direct than that. Um choosing your hats, uh deciding how long you want to examine this gift that Joel's given you, you're the one to jump off, move forward, and then when things start taking a turn, there's nothing at all in the game asking you to turn on your flashlight or crouch, pretend that it's you know, you're in danger. You You're the one doing that as the player. You're embodying these emotions. By this point we've gotten you so deep in that you're you're not really consciously doing this, but all the same, you're providing the performance. You're providing the uh emotional alignment that Ellie is then acting out in the game. So, yeah, that's how we approach this gameplay. Uh the player is an actor cast as Ellie. Uh if you want more resources on this, I would recommend uh directing a directing actors by Judith Weston, uh Manga in Theory and Practice by

Hirohiko Araki, and of course, sorry about my key. I told you they'd be back. Uh but again, that's cool. But, how does the player know all of this? We have all this elaborate plan for them. We We've tried to convince them to be an actor in our weird digital stage play, but uh we didn't send them a script. We didn't let them know that this was going on. Uh how does this work? And the simple answer is level design. So, the way we try to use level design in this scene is that space informs the player of direction and mood. It lets them perform without that script. It lets them actually just kind of take it by vibes. Just do what feels natural. And all of a sudden, hey, you're playing by the script. Of course, any game you can just stand perfectly still, spin around in a circle for 3 hours, or I don't know, walk in the opposite direction and turn it off, but

that simple bit of collaboration really reads that rhythm, that momentum where all of a sudden you as a player are buying into the script. Taking the stage directions. You're performing. You're You're the person we need you to be. So, how do we do this? Uh main things we need are clear information. This is much more than just paint the path forward yellow. That's all really, really useful. You know, as you guys have seen those examples where we'll just have a piece of caution tape fluttering in the wind let you know where to progress, but it's also about what types of things to expect. Whether you're driving navigation or not, whether you're in danger or not, all of these other subtle cues really come into this idea of clear information. And that you need to look at this as like this informing the player is much more than just where to go. Uh and also it's a it's a difficult balancing act because too much information can be bad. I mean, we look at uh over cluttered UI, we look at levels where the noise in the balance is

just not really there. Even if I might tactically be able to understand where I'm going, it's not giving me a mood. Second thing we need is variety. Aesthetics, verticality, traversal and interaction. All of these different things that just provide contrast. Uh it's cool, but the real thing that it's useful for is it also increases the memorability of information. Lets you navigate better. Lets you segment. It lets you kind of get everything from pallet cleansers to new directions, emotional shifts. Again, you're directing this player here. And the third thing is mood and impact. Impression, immersion, and anything else that's emotionally focused. You know, we're giving you the mechanical uh process of what to do, where to go. But we need to focus too on what these emotions are. What should you be feeling? And I think it's so important that it's like I've separated this out. We really need to get people to embody into the space, to buy into the emotions, and we signal a lot of that with

all the techniques in level design. That's why we're here. Uh So, everything from bright sunny days, Joel leading you around as you're being teased about this thing. Uh As you go in, an example here is Joel's leading you this whole way, and then he stops. He just He just chills out and says, "I'm not going to tell you anything else." And this is when we actually provide the reveal. If we go back here, there's actually s- like multiple different ways you can approach this thing. We make sure it listed perfectly each and every path. But, the key thing, going back to everything we've been talking about, suddenly the player's in a situation where they don't know what's going to happen next. But, they know what to do. They understand the emotional context of what's happening. And then the moment that happens afterwards, despite it being exactly to the script that we wanted to make, gets to feel like their contribution to the story. You know, just running through some other things. Like, we we do this with uh letting the dinosaurs uh

be these interaction points for the player. We show one off, and then we show that Oh, the player has dozens of options, all these different ways that they can keep adding to this performance. We shift gears. We change aesthetics. We make things claustrophobic or big. We uh provide really clear framing on where to go here. Like, if you pay attention, you can probably figure out like, "Oh, hey, let's head through that window. Jump down these bars." And then, even more importantly, when we get to the to the annex, uh I just [ __ ] with people. I put down uh collectibles. I put down crafting objects. All of the cover here is valid combat cover, even without enemies. Using all of these subconscious cues to just signal to the player like, "Hey, watch out. And unsurprisingly, players do. They follow along. They really kind of shift into the mood that I'm asking them to. I'm not manipulating them. I'm

directing. I'm asking. I'm like, hey, shift into this mode. Shift into this uh perspective. And that way when we end up surprising the player, shoving them down by the boar, they get to experience that emotion in a much more valid way than if it was just rote. And as you're still reeling from all that adrenaline, we take you through that last beat, that last point of separation. Yeah. Again, space. It lets the player perform without a script. Plenty of resources on this one. I'd really recommend Framed Ink, an approach uh architectural approach to level design, Architecture, Form, Space, and Order, and The Visual Story. So, neat. But, what about, you know, things we haven't covered like timing? Let's talk about pacing and process. So, we have all of this stuff. We have the space, the gameplay, the characters, all

of the elements that we need. How do we actually ensure that it's digestible, that someone would actually sit through from beginning to end without falling asleep or turning the game off? How do we find the right rhythm? Honestly, we kind of cheated in this case. I just let the player do it. Uh I made it more like running and just pacing yourself than like trying to accomplish some kind of graph match. Um in fact, I could do an entire talk on pacing design. And I did. Uh it was yesterday, though. Sorry. So, this is just going to be the short version. Um The main things we need to know are when we're building this level, we want to reward people who want to linger uh linger, enable people who want to rush, give the player clear options, and eventually surprise them with a follow-through. You know, these weird little moments like being able to put a hat on Joel, having Joel buy into these stupid ideas

that Ellie's been throwing out. Um but yeah, really recommend it. It's going to be on the vault. I wish I could cover it more in depth, but we need to move on. Specifically to the actual act of getting all of this done. How did we keep looking at this? How did we go from concept to completion? We really tried to make it a living process. Remember, talked a lot about how iteration drives story. So, we talked about the player's experience. We've talked about how we've involved and put them at the center of this aesthetic universe that we have, but what about us? Simple truth is, you move forward, you expect to get it wrong, you reflect, and keep trying anyway. Most of your ideas will end up in here. Just just get used to it. Bask in the warmth. It's never not going to be painful, trust me. The amount of iterations I had to do on this and most of the other levels was

intensive, but also I wasn't surprised by it. We just kept using all of these lenses, all of these perspectives to try and make it better a little bit every single time. That's cool, but like the other honest question we need to ask ourselves is how do you deal with that struggle? How do you keep getting back up? You know, I'm not just going to sit here and say like, "Oh, get good. Get harder. Get less sensitive to throwing away work." Which is just one of the hardest things a creative can do. Uh in the process of making this level though, I feel like I found a pretty reasonable answer, which is keep looking for inspiration. And specifically, I want to talk about Naoko Yamada. This person is my favorite film director. And I'm going to spend about 2 minutes to completely derailing this entire talk because it's mine and you need to know about her. Uh so, about halfway through making this level, we were just

kind of lost. Like, we knew what we wanted to do. I didn't know how we were actually going to do it. And then I found this movie called Liz and the Blue Bird. She is like made by this lady right here, by this studio. She is of the same caliber as Miyazaki to me, quite frankly. And I just want to do some quick points on why I think that is. So, let's jump into that. First, she is a master of character and body language. This is the biggest takeaway if you watch her movies for anything, this is the reason. She is able to imbue her characters with a sense of life and cognition and rhythm that is spellbinding. And like, I don't even have to make a comment here. Just look at it. Like, every single character has their internal lives just on full display. She finds strange little quirks and moments to just pull out of characters. In fact, she's so good at this, she's actually notorious for just shooting people from the knees down in a scene and fully being able to articulate their

emotions from that. I mean, look. She's doing it right now. Uh she makes the mundane spectacular. She is just able to take any normal world. She doesn't use fantastical elements like Miyazaki. Every single one of her films are real world, real problems, simple people struggling with emotions and love. And then she'll take all of that and emotionally destroy you. Like, I have trouble staying hydrated in most of her films. Like, I I just sobbed. Like, it's it's great. I mean, like, look look at this [ __ ] Like, you're going to feel this way. It's bleak and it's beautiful. and I think all of this happens because like as a leader I also find her as an inspiration. She brings out the best in other great artists. This guy right here is called Kensuke Ushio. He is a phenomenal composer. But most strikingly his music sounds totally different when he's working on her projects. He's normally doing like dark, grungy, insane, almost metal-like music for things like Devilman Crybaby. For Liz and the Blue Bird he literally scattered

paint onto the bars and then composed off of that after like seeing an animation technique she was using. Uh she has a masterful studio that like really backs her up. And then in the most selfish part of this talk, the other reason I like her is her favorite director is my other favorite director. This guy who looks like a Dark Souls boss is Jodorowsky. He's this weird Argentinian. We don't have time to get into him. The main thing is his stuff's wild, it's surreal, it's poetic, and I think you'd regret missing it. So two minutes are up. Just wanted to do that aside. I could do an entire hour on this director and her work and I might. Uh but if I could just sum it up in one word, it would be grace. Her work is graceful. And if you like anything I've done this is the template for a lot of what I pursue as a as a creative. This is the template that

I'm like striving for now after seeing this. Like it changed the trajectory of my creative ambitions. It was just that important to me. But enough about that. Let's get into the next stage, which is production. The how-do. How do we actually take all of these compositional things and make it happen? So let's take a quick look. So we had teams and timelines. My role on this project officially was level design. Basically, I was the first designer on the ground. I created the layouts, some of the prototypes, and then coordinated the teams at each stage. I was basically the person who {quote} owned the level. Doesn't mean I did everything or was the main person in charge. I was just the person responsible for making sure everybody was talking to each other. So, at the start, it was just me. Uh the way we just generally do it is like we start with a level designer and a lead kind of working together to incubate it. We move it up to a four-person strike team until it's kind of cooking a little bit better.

And then, as we move into alpha and play testing, we slowly ramp up to more and more people and more and more people until at full production, it's all hands on deck. The maximum amount of people are working on it. All departments are involved. Everything is just firing on all cylinders. Uh also, I didn't work on this level beginning to end in one big chunk. We each of us had about three to five levels uh that we accomplished and wove together for over about two years. So, I want to talk about this process a little bit. Uh the first step to me, or how I'd like to describe Naughty Dog's uh goals is to create a playable storyboard. Step one of this is actually breaking the story. We sit down with the writers, we read the pitch, we grab the lead into a meeting, and we just talk about it. We we figure out in a brainstorm what we're trying to accomplish. What are interesting things that we can do using all these things that we've talked about before. We literally went to a seminar hosted by Robert McKee uh early when I started there just to have a shared language that we could start

chewing on this, start breaking this down into manageable parts, and make it happen. Um after that meeting, uh level designer would break off with the lead and just plan a first pass. We would try to get all of this done in like less than two days. It's a very very concise but focused start to all of this. The next step, though, is uh uh what I like to call 3D first design. We'll just jump straight into Maya. No top-down maps. Maybe just a rough outline, maybe a few scribbles. Nothing substantial beyond actually getting a mesh done that I can walk around in. We get the block mesh in engine and we try to actually move the character and the camera around to see if it starts working. The reason I think uh this is one of the best ways to do it is it's the fastest way to test an idea. You can see if this even makes sense, if this feels too big or too small, if the camera doesn't make any sense, if you can even accomplish something like revealing a dinosaur to somebody uh by wandering through a forest. Like

you can say all of those things, but it doesn't matter unless you can prove that it actually works in engine. And it's also important to do it so fast because it lets you assume that things will be thrown away. Going back to that living process and that iteration. Like you are intentionally making this to die. Like don't care about it. Don't feel as little pain to shove it out and try again. But uh but yeah. What also do we mean when we're talking about storyboards? Like what do we mean playable storyboard? So I think for extra context on what I was just saying, a lot of storyboards are like this. Like this is from Knives Out. The image on the right is their initial concept for this shot. Just soak it in. You don't need to be a master illustrator to convey your ideas. And now I'm going to show you what we dealt with in the early stages of development. This was the first version of the dinosaur.

This is how we tested it. This is how we saw off like, "Hey, is this even Is this Can this even work?" And once we realized it did work, we took it to the final version. Uh unfortunately, due to a hiccup with the Wi-Fi, I was going to show some uh videos of early playtest levels, but uh they're all on my website ehilldesign.com. Basically just showing a bit of the block mesh process, the kinds of things that we do to uh walk the player through and have that playable prototype. But just as another example, looking at this, uh look at this mess. Look at this scribble. This is an early storyboard for JoJo's Bizarre Adventure. The person behind this is like one of the best draftsmen alive. This guy's been in the Louvre. This is what he normally draws like in a finished product, but again, he starts with this. Starts with something that can be thrown away just to see if the idea works. So yeah, once we have this thing, once we have this playable storyboard, we walk through the blockmesh with the leads. We actually sit down face-to-face

and just talk through it. We fill in the gaps of animation and art and dialogue with our performance. Like I mentioned, we're all ex-theater kids. Well, not all, but um we all acted like we were ex-theater kids. We we just lived in the scene with each other. We tried to figure it out. We we just wildly gesture and say like, "Oh, this will look like this. It'll be It'll be great." You know, and and it led to a lot of great moments. The first time I showed this off to the leads, one of them, right out of their mouth, "Can I climb the dinosaur?" Like the grin on their face was phenomenal uh when we said, "Yeah, sure, we can totally do that." And then we did. Uh the same thing with the hats. It was just a dumb joke I made. Just totally off the cuff, and it wound up being a feature that a lot of people really like. Um so yeah. We would talk about what sucked and what worked. We would use all of that compositional stuff we covered in the first part of this talk to analyze what was good to move forward with and what we needed to just cut or redo.

And we just literally throw it all out again and redo. It's the power of iteration. You rework it until it clicks. Um normally we would spend about several weeks iterating, like 4 to 12, depending on the complexity. That's total, mind you, not per iteration. 4 to 12 weeks of figuring out this blockmesh, this playable storyboard. Again, like uh including starting from scratch. Like you we we tried to keep this as limited and light as possible. This is the part where the level designer was the only person working on it. Um but yeah, it lets it grow and breathe as we get a feel for it. We don't really even know at this point how long the final level's going to be. You could add a third, double it, cut it, shift it around, try radical different flows and beats with the elements. Like the dinosaur kind of came in early, but for example, the uh space capsule, that whole entire scene, that wasn't until like the last quarter of this like ideation process. Before the graffiti was just going to be right after the main part of the museum and on the space capsule itself. But in going

through this process, we me and the leads like wanted more. We figured out what to add, and then I just went back into Maya and added more. Um but yeah, once we have this rough sketch done, this playable board, that's great, but that's not enough. The next stage is getting into alphas and playtesting. Um we try to build a playable prototype. That's where the first stages we're taking to other designers. We're trying to limit our, you know, gesturing, our manic energy to convey the story and just let it speak more and more for itself. Um that's when we also start adding in more developers. This like I said, the four to six-person strike team that supports the level designer to actually finish and develop features. Um and then it work towards that first playable. First playable means the first time you can just shut up and let someone play it on their own. No guides, no oral presentation. I hand

it to you, I leave the room. Does it work or not? Uh again, I would have some video for you, but unfortunately not. All on my website. Um and then once we hit that stage, once I can actually hand it to you and leave, we grab some outside people, hand it to them and leave. So, every few weeks we would have outside people come in, they'd be have really detailed feedback, and we do this at a cadence of every 2 to 3 weeks as we're shifting through production. So, this would be a studio-wide event. And trust me, when you start playtesting thoroughly like this or not, you will see how bad your layout really is. Everything that was duct-taped over with good intentions and slick presentations just falls away. You will watch people get lost, not care about your story, complain, and worst of all, be right about all of it. Just just rip it apart. But, uh also because of this cycle, this like

tempo where every 2 to 3 weeks we knew we were going to be bringing in new people. We knew that Oh, I might I [ __ ] up. That level sucked. What can I do in the next 3 weeks to make it better? What can I do to ensure that I don't break this and it's still playable in 3 weeks, too? Um Yeah, so all the changes need to be playable. We had so much data to work with, though. It was massively helpful. And another level with it, too, was it was studio-wide tempo. It wasn't just playtesting your level, it was playtesting all the levels at once. Then we would go back to iteration, and then playtest all the levels at once. It really helped facilitate, uh, you know, just a cohesion of everybody knew what was on the schedule. Everybody knew what needed to get done. We were able to focus on tighter milestones. Um But, from there, we would also take all of that data and try to dial things in. That's when the more formal reviews with studio leadership would tend to happen. We'd refine features and scripting and prove out any ideas that we have with playtest feedback rather than just wildly gesturing, "Hey guys, I think

this is a really good idea." Instead, we can go like, "Hey, this part objectively people like it. They keep talking about it. Let's improve this or fix this. Again, it's just all about trying to facilitate effective iteration. But even at this stage, you know, we try to cut things early, throw things out fast, but there's still going to be a lot of rework. Cuts, revisions, additions. We use the entire team for input at this point. Starting from scratch is almost limited to specific sections, but it's still fairly common. We call it back to layout where a chunk of the level might just be kind of sawed out and redone. An honest example of this was that space section. We would we were pretty confident in the rest of the layout and the rest of the area, but there were a few times where we just looked at that upper area, that second floor, and we had art done for it, we had some features in, and it just wasn't right. And we kind of pulled the trigger on redoing it again. And yeah, just going back to that

iteration, that living process. You just keep trying to make it work. You keep redoing and redoing and redoing. But what happens when it does work? Like you you've just burned all of this effort, like desperately trying to make this thing work, and now it does. And now what? That's when you lock the layout. The level is moved from prototype to alpha. Now the rest of the team can be brought in to actually make this thing shippable. Which brings us to full production. This is the part where everything gets ramped up. We bring off hand off to the other departments. This isn't a comprehensive list, but you know, an example is like audio, VFX, mocap, dialogue, character art, AI, foreground, technical art, environment art, animation, cinematics, QA, and on and on and on and on. This is when we take it from just oh, the designers are working on this cool thing with a small group of people to everyone needs to know about every detail all of the time. It also really helps collaboration.

We then start really focusing on feedback and direction more as level designers more than anything else. This is when it goes hands-off. Like, I should not be in Maya, I should not be in editor messing with things other than gameplay scripting. Um, so yeah, it's like I shift time to other levels at this point. Like, maybe I start the whole process over on another section or start scripting gameplay for the scene. And a big part of this, at least in the Naughty Dog process and in making this level for me was coordinating the team as well. I was one scheduling meetings, relaying information, managing task loads and deadlines, and uh also facilitating the review process. I really liked what we did on this project where we would have uh large recorded meetings where we would just play through the level and note down every single thing that was wrong. And this would be like maybe eight people at sometimes just pausing, commenting, pausing, commenting. So, 20-minute level might be 3 hours of commentary video. But, we also had this annotation software where you could take the recording and just make a note. Turn that note into a Jira

ticket. Hand it off to the level designer. Hand off all the Jira tickets to the people that need to get it done. Um But yeah, this was mostly like focused specific feedback. Very rarely would this turn into a full rework or back to layout kick. Um, but yeah, how did I do this? Like what kind of fancy tools were there? Nothing really important. Uh face-to-face communication, legal pads, Skype for business, email, Jira, no silly proprietary scrum board or anything else. Just actual paper, a few uh tags, and it kind of looks something like this. For clarity, this is not at all for anything Naughty Dog. These are just some personal projects that I kind of work with, but no, I would just kind of run around the office like it was an Aaron Sorkin movie just talking to each department about every single thing on my giant list I need done for the day. Uh I would break it down into sections with tabs. And yeah, it was facilitating iteration. And then the hard part comes, which is lock and polish.

So, before we said we handed things off, it was kind of a lie. Uh there were still reworks, but this was the hard point where we would shift into we have deadlines. We're getting this done. No major changes. Push it from this janky prototype mess to this thing that is Naughty Dog quality. So, in this polish stage everything matters. The review and the playtesting cycle I mentioned before is continuing through this entire process, this entire span. Uh and the notes are picking up. They're getting more and more. It turns into a flood. Every single department has tiny tweaks and everything, but through all of this it starts to finally take shape into something that's going to ship. And yeah, a lot of this time uh even though there's no layouts to be done, us in the level design department are running around doing additional scripting, fixing things, and we're just responsible to get things done. I was basically a messenger boy on some days, and it was hectic as heck. Because that's the real takeaway on this part of production.

The team does it. All of these other developers that I worked with are some of just the most talented people I probably will ever have the privilege to work with. And they're just they're just great. They were on top of everything to the point that yeah, I was relegated to just running around conveying information and notes. Um because yeah, this type of work isn't focused on somebody like me or the leads. It does take an entire massive organization to deliver. And the result is honestly still just one of the things I'm most proud to have helped on. Like I yeah, this was a joy to work on. And thanks for uh letting me do this talk. Uh so yeah, you know, that's it. We've covered composition and production. Hopefully giving you a bit of an insight into what it takes to make a Naughty Dog level, specifically a weird one like this. But, yeah. Final housekeeping, let's cover some of the takeaways. Uh you want to find a useful framework. For us, it was story. For you, it can be

something else. Competitive integrity, counterplay, whatever it is, you use them to guide iteration. Start it rough as fast as you can. You fail. You pick yourself up, often by finding some inspiration. You iterate more, fail more, try not to stop until it's great. And you rely on your team. And that's it. Uh hope you guys can follow me. This is my Twitter. I'm going to be working on some secret uh podcast stuff that'll hopefully come out soon. And then, the main thing is I'll be contributing to Outer Worlds 2. And here's a list of resources. Also, go watch Liz and the Blue Bird and A Silent Voice and The Heike Story. I'm not kidding, they're really, really good. And we have time for questions.

Hi. Great talk. Question about the play testing every 2-3 weeks that you mentioned, um especially early on in the process, what does that look like in terms of the questions that you're asking the play testers? Is it very directed to a specific thing that you're trying to accomplish? Or do you start out general? How does that work? get too, too specific because uh Sony really has been refining their process constantly. Um but in general, it's a lot of data, um big questionnaires, group feedbacks, focus tests. Um so, yeah, just literally anything that we thought would help, we would throw at it. There wasn't like a framework that we really worked from. We're just like, what would we like to know and what would they like to tell us? Hi. Thanks for the talk. Um I was kind of on the same page as him. Asking about the feedback from outside players, how do you kind of handle that? And how do you know what's good feedback versus bad feedback? I mean, you're the level designer, you know best. How do you know what to listen to? If that

makes No, that that's a really great question. So, the case wasn't picked up. How do you know what's good feedback? How do you know what's bad feedback? How do you turn them into notes? A good rule of thumb I've always liked is that the player is usually right that something is wrong, but usually wrong about what specifically is causing the problem. The classic example was there was a first-person shooter where the two sides had identically statistically damaging guns, same fire rate, everything else like that. But the sound effects on one sounded more aggressive and more intense than the other. And as a result, the players kept giving feedback like, "Oh, this gun's more powerful than this one. This gun's more powerful than this one. You need to nerf this." They were able to analyze that like it's statistically identical. The only difference was actually the sound effect. They boosted gain on it. Nope, went away. Again, players are telling you something's wrong because there's usually something wrong. But they will always The other thing to understand is they'll always reach to the most obvious answer, the most proximate answer, the thing that's the most in their perception because they

feel it, they see it, they give you the feedback. But yeah. Thank you. One more playtesting question. How much do you have the same group of humans come in to see if your new iteration works differently versus try to vary them to make sure that you're not just designing for that group of people? Oh, we would try to get fresh feedback as much as possible. Very rarely I I forgot if we ever brought entire groups back or entire individuals back, but in general with outside playtesting, the idea is fresh first player experience. And then that kind of longitudinal iteration would just be mostly design team, the people who are also the other developers. The people who are making the game would kind of provide that like, "Oh, where were we 6 months ago and where are we now type analysis. Thank you. Uh quick question about the I guess the proportion of time that you spent on the block out versus the polish stage. What was that like? Uh it was pretty variable. I can't give you like a hard time on it because it was all just very fluid, very focused, and uh the directors were the ones kind of setting the final deadlines.

Um but again, like the I would want to say like an eighth of the time or less was that initial iteration process. And then it was probably somewhat close to like a quarter a quarter and then two quarters on polish. Like that final stage of taking it from beta to ship at Naughty Dog is uh a bit longer than most other studios. So, I would say yeah, that's that's the main balance. I can't be like, "Oh, we spent 3 months on this, 3 months on that." cuz it was spaced over five levels over 2 years. We had a lot of shifts in production and the like, but in general, just providing more time for that like last step, that last team focused step, I think is another really important thing uh that Naughty Dog does that's not necessarily associated with our composition or our level design. So, as a designer, there's always a moment like you have to iterate till you get it perfect, but there's also this expression where an art project is never done, only abandoned. I'm wondering when do you know to abandon it?

Uh Like especially for that one thing that you cut out entirely because in my mind you can just try to polish a turd as much as you can. Yeah, that's where uh leadership I think is really effective. So, kind of by separating that decision-making uh from the person making the ideas and the person deciding like this is what gets made and that's what doesn't. Um so, like to me a good lead is always like a good editor. They're the people who get to be a step removed from the process of making the thing and can kind of say like this is done, this is good enough, this isn't, and um Um, yeah, so it's like a structural thing. It it's not uh Don't do it alone. I guess that's the best way I can do it since I can say it. Thank you. I was surprised to hear you all used uh Maya to block out stuff. Uh so, you know, my career I've mostly used, you know, uh brush work or primitives or stuff to block out those rough shapes pretty quickly. What What did you find the advantages or disadvantages were

using Maya and to block out stuff instead of, say, in editor tools or other things like that? Yeah, so the main reason is just that uh Naughty Dog has its own uh personalized tech stack. There's no like we're not using Unity or Unreal or anything else like that. So, there was a lot of integration with Maya, but overall, the main benefit uh to it in my uh mind is just that 3D first design, being able to focus on bespoke sections. So, I wouldn't recommend this for all games. Like uh working on Outer Worlds 2, we're obviously working from more of a kit structure. Um but in a narrative-focused game like The Last of Us, being able to just custom make every piece of geo in this storyboarding process, uh think we found that pretty effective. Uh we all worked fast enough, we're all capable enough at this particular mode that we're able to get things standing and done fast enough. We also didn't have a lot of uh complicated physics objects or the like in our gameplay loop. So, we were then able to, you know, mostly rely on just mood,

structure, sightlines, and pacing. Uh and then just go from Maya to engine. Hey there. So, under the production schedule you detailed, other disciplines and teams are rolling on pretty far into that design process. And under those conditions, how do you maintain alignment and get all the other disciplines on board and engage with the original vision you had as the LD? Also, what sorts of steps do you take to find potential compromises, mitigate wake and more as the design owner once it reaches their hands? I'm sorry, I'm having trouble hearing you. Can you start the question over again? I'll just reiterate the first part of the question. Um So, other disciplines and teams are rolling on pretty far into the design process. Under those conditions, how do you maintain alignment and get all the other disciplines on board and engaged with the original vision that you had as the LD? Uh yes. So, the main question is how do you keep everybody on the same page? Correct? Mhm. Yes. So, uh

communication. It's real real dead simple. We would meet regularly. We would talk about it. Um and in the same kind of mindset where I'm trying to involve the player in the act of playing the game. Um I'm always open to good ideas. And the more good ideas that my team members bring to me, the less work I have to do. So, I'm just really pro getting suggestions from everybody on how the game should be designed. And it has the secondary benefit of uh that extra buy-in helps not only just people commit to it, but then all of a sudden they're evangelizing their ideas. They're talking about the cool things they add. And then you get this group um perspective on how things work. So, yeah, like communication, direction. And now at a dog, it was a bit easier because uh we didn't have dedicated producers. So, that extra communication had to be picked up by everybody. Um so, that's part of the benefit. I don't know if I would necessarily recommend that as a universal fix though. Uh the second thing I wanted to ask is if you're reaching concept and geo lock that early in the process, what sorts of

steps are you taking to find potential compromises with teams that can't deliver on the original intention, mitigating weight, uh etc.? So, the the main way we kind of approach that was just uh trying to avoid that as much as possible with that communication early on. Trying to set expectations and the like. There were very you know, there were a lot of reworks kind of taken from the top, but uh rarely was it ever like, "Oh, this team was so particularly behind schedule we had to cut things. Um Again, that isn't anything with my certain process. I think it's just the um the people involved have been truly excellent at it for a long time. Got you. Thanks so much for the awesome talk. Thank you. Hi, thank you. Um you're talking about in the blockout your base goals are to communicate the points and beats you're trying to express in the level. Are there any other basic interactions and uh systems that you use as part of that blockout phase? So, the earliest earliest version, um

honestly we would just get by if we could walk around the level. And like I said, the rest is just oral presentation, uh riffing, just kind of pitching it in person. It's not really a game, it's just a 3D storyboard that you're puppeting around at that point. Um when we shift into getting things towards first playable, that's when we kind of shift out. But uh yeah, we just ran out of time. Uh hope that answered your question sufficiently. Uh we'll be in the wrap-out room uh just next door if you guys want to ask me any other questions. I'll be here for a while. And thanks again for this talk. I'm really glad to be able to talk about this.