Solving Titan-Sized Problems: Evolving Titan Combat in Titanfall 2

All right. Thanks for coming. Um before we begin, let's uh do a few reminders. Number one, please silence your phones and or noise makers. Um just be polite for everyone. And number two, please uh don't forget to fill out the surveys in your email. Um don't forget to check your spam folder cuz sometimes it goes there. Uh and uh yeah, let's uh I guess we'll begin. All right. Good morning, everyone. And thanks for coming. Welcome to solving Titan-sized problems, evolving Titan combat in Titanfall 2. My name is Carlos and I am a game designer at Respawn Entertainment. Um before that, I was at Infinity Ward and before that, I was company called Shell Games. Click. Click. Um here are some of the past projects I worked on. Most notably, of course,

Titanfall 2, which is the game we'll be talking about today. Um a couple more things before we begin. Um first, as you all know, video game development is a messy process. And while this talk may make it seem like we just figured everything out, you know, just like that, this is really uh the reality is a lot of these lessons are a result of trial and error. And we're only really able to organize them in this manner after, you know, all the dust has settled and we go take a look at what we've done. Um second, making games is team effort. And while I'm the guy standing in front of you giving this talk, a lot of what is in this talk represents the hard work of the different teams that, you know, put their work into the the game. So, with that, let's get started. Button. So, what's this talk going to be about? Um in this talk, I'll be discussing how we evolved Titan combat in Titanfall 2

and how we tackle the problem of combat in a high time to kill shooter. We're going to be breaking down the talk into three major sections. The first section is the introduction to the problem, where we'll talk about um kind of try trying to insert a high time to kill game or trying to insert low time to kill shooter sensibilities into a high time to kill game. Um and what problems can crop up with that. Um the second part, we'll go into the solutions that worked for us in Titanfall 2. And, you know, we'll start how with how we attacked this problem in multiplayer and then move on to how this affected single player. Finally, we'll discuss some pitfalls that we came across or problems that we, you know, might have might not have completely figured out. Real quick, let's talk about Titanfall 2. What is Titanfall 2? For those who don't know, Titanfall 2 is a first-person shooter set in the far future, in which players

play as pilots. Oh god. Agile soldiers that can wall run and double jump around the battlefield. And during the course of a match, pilots can call down Titans, huge war machines designed for destruction, and manually control them on the battlefield. So, Titanfall is really a mix of both low time to kill gameplay in the form of pilots and high time to kill gameplay in the form of Titans. Um and funny story, just for the purposes of this talk, I realized, you know, when I talk about high time to kill, most people will think about games like Halo and Overwatch. So, just for the purposes of this talk, when I actually say high time to kill, I'm talking more of like 20 second or greater time to kill, as opposed to something a game like Overwatch or Halo or Team Fortress, which lives in a more like 5 to 10 second time to kill. Um so, I guess in the real world, we would call something like Titan combat extreme high time to kill if we

were to stick with high time to kill for Team Fortress and Overwatch. But, you know, like I said, I realized that a little late, so just for this talk, we'll that's what high time to kill means. All right. So, now that that's out of the way, let's begin by talking about what exactly the problem is we're dealing with. At the beginning of Titanfall 2, Titans were identified as an area in which the great the game could greatly expand. With mobility no longer being unique to the franchise, Titans were seen as a greater key differentiator that could really set us apart. The problem was that internally, Titan gameplay was largely seen as um boring or lacking depth. Um people would usually throw out the word mushy as a way to describe Titan combat. And references to Rock'em Sock'em Robots were often I hate you, you're not playing my GIF. But um Rock'em Sock'em Robots was often brought up as a uh as a point of reference for Titan

fights. So, if we were really to push Titan gameplay, these would be the biggest problems to solve. So, we started asking ourselves, why? Why does Titan gameplay feel so mushy? Why does it lack depth? You know, it seemed to have all the pieces, um guns that felt good and rockets and dashing. And the Titans even moved quickly in in an agile manner, so it felt better to drive around in a Titan. Um So, we got back to basics and really started asking ourselves some hard questions. Questions like, where does shooter depth come from? And we know that gameplay depth can come from having interesting decisions. So, where would the interesting decisions come from in a shooter? So, the model we came up with, and most of you, you know, one way or another might be already familiar with this, is that shooters tend to give players different layers of decisions that they're making. And these are broken down by the amount of time players have

to make those decisions. So, we've identified three key layers that most of the decisions can fall into. There's the atomic layer, which lives in the sub-second space. The Oh, I hate you so much. The tactical layer, which lives in the 10 second space. And the strategic layer, which lives in the 5 to 10 minute space. So, we'll look at each one of these layers real quick. Atomic decisions um live Again, they these live in the sub-second decision-making space. And they involve of the player identifying a target on screen and then deciding how to move their cursor over the target, which means to deciding how they're going to move their thumb on the thumbstick. And then deciding when to pull that trigger, hopefully killing the target. And these types of decisions are happening at a split-second interval. So, you're not really always conscious that you're making these decisions, but they are there. Um and these types of decisions have to do with what we call twitch skill or

hand-eye coordination. So, you get better by practicing and building muscle memory. Tactical decisions happen at around 10 second-ish intervals. And usually these happen when players have just spawned in or maybe when you just finished clearing area in a team death match setting. Um and these decisions will often involve the player assessing the current state of the map, where am I, where are the enemies or what do I think the enemies are, and then deciding on the on his engagement approach. Okay, so if I'm over here and the enemies are up in that tower, and there's two ways to get to that tower, they're probably looking at the main road because there's a window over there, so I'm going to take this other road um and hopefully catch them by surprise. And these types of decisions often have to do with what people often refer to as map knowledge. So, the more you know about the map, the better you can be at making these decisions. Finally, strategic decisions happen over

a much longer timeline, sometimes over once per match, but so, like I said, normally you you're familiar with these. You would select them at the beginning of the match, but it for example, in a team death match scenario, it can happen in the middle of the match, say when I go on a like a bad losing streak or bad death streak, I might have to reassess my loadout and then go change it and change my strategy. Um So, re- and really, another good way to think about all these layers is like the atomic layer will revolve around thinking about this next bullet that I'm going to fire. And the tactical decisions are thinking about what's the next fight I'm going to have. And the strategic decisions are thinking about like, what's the next three, four, five, six, seven fights I'm going to have. So, what we really have is this kind of food pyramid of interesting decisions. So, you know that the atomic scale

decisions are all the way at the bottom for shooters specifically because these are the decisions you're making you're making a lot of these decisions at a time or per game, right? At So, for every five atomic decisions you're making, you're maybe making only one tactical decision. And so, I mean, made-up numbers, but you kind of get the idea, right? There's way more at the bottom than there is at the top. So, when we apply these layers to to thinking about Titan combat, the most apparent problem is that high time to kill or HTTK, it works it really works against the atomic layer decisions. Um and to illustrate, imagine we have our low time to kill shooter. I'm playing this game. There's my cursor. I find a guy that I want to kill. So, now I'm I'm deciding in my brain, my brain is fully engaged in moving my my deep my thumbstick to try to get this cursor over the guy and I'm moving it and now I'm trying to

figure out when I'm going to pull this trigger and I pull the trigger and then he's dead which means I can move on to my next decision or if I fail then it's more likely that I'm dead and I can move on to my next decision anyway. Um contrasting that with something like a high time to kill game the same thing that happens I move my cursor over the guy there he is I make a decision to pull the trigger and then nothing happens so I guess I'll just keep this trigger held and now I'm not really making any interesting decisions I'm just retaining the last decision that I make and nothing's prompting me to make a subsequent decision and that can really lead to boredom and frustration and this is what people commonly refer to as the bullet sponge problem aka why bullet sponges are boring to fight. Um and in Titanfall this is exacerbated even more by the fact that Titans take

up a huge amount of screen space so even putting your cursor over them is a lot less interesting than putting your cursor over a pilot which takes up less screen space. Um and so this is really the problem that a lot of high time to kill shooters will rub up against more people will oh sorry I already said that. So we really need to solve this problem if we were to make Titans more important for Titanfall 2. Real quick uh these are some other not as crucial problems that came up during assessment. Uh so looking at the tactical layer it seemed that Titan combat overall it it's got some good tactical depth because our map designers build our standard three lane maps for Titans um and that makes it so that it's fairly easy to predict where a Titan's going to be and you can have kind of your standard traditional shooter tactical depth can can be drawn from that. The only difference is that Titans can't jump up or climb anything so they're all really playing on a single plane so you lose a little bit of depth with that but

overall it's not the end of the world. On a on the strategic front we noticed that players would often default to what we called dual purpose weapons like the XO16 chaingun which is a rapid fire hit scan weapon uh and the reason for this is that in Titanfall you're not you're never really sure whether you're going to come up against a Titan or a pilot so it turned out to be more advantageous for you to bring a weapon that could just handle both um both at both types of enemies um so it turned out to be safer so most people ended up picking that what weapon and so that was kind of our weapon meta for Titanfall 1. So we'll be touching on these along the way but the real problem to solve with regards to high TTK has to do with the atomic layer. Okay so now that we have a greater understanding of the problem we can start figuring out how we're going to solve it. So first we'll talk about how we went

about solving this in multiplayer. We decided would it would be better to start solving the problem in multiplayer because it's much easier to apply multiplayer depth to single player rather than the other way around. Single player tends to use AI logic as somewhat of a crutch um you know forcing AI to do certain things so that it helps the player or challenges the player in certain ways. So it but it's almost impossible to force players to do what an AI is doing whereas it's much simpler to make an AI pretend to be a bad player. So first I want to talk about specific things that we brought back from Titanfall 1 you know it wasn't all bad so here's some things that we just we you know we looked back and we decided we wanted to keep. Um the first is the concept of weak spots which is basically our equivalent of headshots for Titans. So similar to the headshot concept this creates a high risk high reward zone that the player can aim for which helps him with the atomic layer it it helps him give like a higher skill higher way to express a higher skill over other players that is

gives him higher risk. Um the problem with weak spots in Titanfall 1 was that they were only present when a Titan's shield was lowered so most of the time you encounter a Titan the weak spots wouldn't even be showing and you have to take down their shield so you have this portion of gameplay where the weak spots don't exist and then they exist so that was a problem for Titanfall 1 which we uh by removing shields in Titanfall 2 we were able to expose the weak spots more often. The second thing we carried over from Titanfall 1 was the concept of dashing and blocking as a form of damage mitigation. So this video isn't playing. Wait are you serious? Oh there it is. Oh my god. So these skills we were able to add on to the atomic layer but they were not viable due to due to the presence of hit scan weapons like the XO16 which made it you know impossible to dash but we'll we'll talk more on that later.

Okay so moving on to Titanfall 2 the first thing we did was we tried to figure out if any other multiplayer shooters have solved the high time to kill problem and we came up with fairly empty handed so if anyone has any ideas you know I'd love to talk about it after the talk. Um so instead we looked at other genres like Street Fighter and MOBAs and we said hey these games you know they have high time to kill and they have good depth so maybe we can learn something from these games. Um so trying to find the commonality between the two you know it being pretty obvious that these both of these types of genres had the idea of preset characters or classes. And once we kind of unlocked that discovery it became really clear that having Titans be preset characters was going to be a really good fit because for high time to kill you know it it takes advantage of the increased face time between players and not only that

but the problem we were having of Titans taking up a lot of screen space is now all also somewhat of a strength since you have that much more screen space to visualize them. And so this is how that turned out.

So obviously nice thing with having Titan classes too is now we can kind of give players these common mech fantasies that uh are pretty popular so like we have the fire Titan we have the Titan with a sword we have a laser Titan we have a Titan that can fly so um in a way it was we it also helped us kind of achieve fantasy fulfillment for the players. But the real win with Titan classes was that high time to kill means you have more time to assess what an enemy who an enemy is and think about what tactics he might employ. Um and a lot of games like Team Fortress 2 they'll talk about the necessity of the snap read which is that players need

to be able to identify their enemies in a split second thereby increasing the importance of distinct silhouettes for them. And for us specifically our fights lasted so long that the snap read wasn't even like 100% essential. I mean we still strive for good silhouettes at the end of the day but during dev testing all we had to do was put the names over the guys' head and it was you know good enough it was it was readable enough for for in kind of internal testing and it helped us kind of develop really quickly. Um in addition high time to kill also means now that players now actually have plenty of time to think about how they can combo their abilities. And as a good example um one of the Titans Scorch has some has an ability called the incendiary trap which you can throw out and ignite and this on its own doesn't do anything um it just sits there and it but it requires Scorch to light it up with his main weapon and this is the type of ability that you would never ship in a low time to kill shooter because it's

really inefficient. Like you're you're just better off pointing your gun at the enemy and shooting him. Um having fixed loadouts also meant that abilities didn't have to directly compete with each other for players' loadout selection. So in an a la carte system what would normally be on your right bumper you know if you had four selections four weapons there normally you would want to balance all of them in such a way that their power level was pretty similar cuz otherwise players would just pick the best one all the time. Um but in a in a fixed loadout system my right bumper ability could be weaker than someone else's right bumper ability cuz you could technically transfer that power somewhere else. So someone could have a weaker right bumper but then a stronger primary or a stronger um defensive ability and so on and so forth. Uh It also oops it also allowed us to create more niche abilities that weren't necessarily useful in every single situation. And since we're going down the road of

fixed load outs, we wanted each Titan's abilities to synergize with each other. This creates depth by allowing players to discover the different ways that they can use a Titan's abilities. So, we wanted to make sure that all Titans were thematically matched. Um but also that all their abilities kind of you you would be able to to be to think about how you could use one or two or three abilities together. But, we still need to address the atomic layer problem directly. And to do this, we found that staying away from rapid fire weapons actually benefited that layer. Low frequency, high damage attacks create spikes in the flow of the fight where players can have small moments of success and failure. So, this also mimics kind of the low time to kill dynamic since the acquire aim fire loop after each shot is made since the lack of rapid fire capabilities means that the player needs to reacquire the target after each shot.

So, this also allows for a better better rhythm in the fight since it creates pauses between the shots. So, players can have time to assess the situation between each of his each of his enemy's shots. So, let's take a look real quick. First so, we're going to take a look at two videos real quick. The first video is from uh is a is the XO-16 chain gun, the infamous XO-16 chain gun. Kind of gives you the feel that comes out of that. So, in that video you'll see the guy with the XO-16, he's just running around pointing the gun at his enemy and there's really no new decision points that that he's triggering other than the fact that when

maybe A, when he reloads, you know, he might think about taking cover then cuz he can't DPS at that point. Or B, when his enemy pulls up a shield cuz then he might think about well, that enemy can damage me, but I can't damage him, so maybe I should break away. But as far as his aiming and shooting, it's fairly uninteresting. You can see he's just pointing and then just casually turning and nothing's really changing. Now, let's contrast that with the next video, which is our more low frequency attacks. So, in this one you can see after each shot, he's having to reassess what he's

doing, reacquire his target, which makes for a more interesting kind of play pattern for his primary weapon. But that doesn't mean that we don't have rapid fire weapons because the good thing about rapid fire weapons is that they're really good for new players. Um it's really they're really easy to use, so it provides a way for new players to get into your game and immediately start having fun. So, we didn't want to completely give up that accessibility. But, what we did was we made sure that the Titans that had rapid fire weapons would have something else in their load out that could compensate and create depth. So, a good example is the Titan Ion. He has his primary weapon is a rapid fire rifle. Um but his right shoulder but Oh, sorry. Let me do that again. His primary weapon is a rapid fire rifle, but it's really weak. And his secondary, which is a shoulder laser, is a one time kind of high impact

skill shot that deals a lot of damage. So, that's So, as an Ion player, you're going to you're really you're really using your splitter rifle to just kind of throw jabs at your opponent, but you what you really want to do is hit him with your strong laser. And so, here's what that looks like. So, in this video in particular, actually, you'll see the first laser shot was missed and that was pretty crucial. Like if he would have hit that laser shot, this Titan would have been doomed and then he could have gone in for the execution. Um as mentioned previously, blocking and dashing was something we carried over from Titanfall 1, but we realized that this only really works when everyone's weapons are projectile

based. Uh projectile weapons have travel time and that time is essential for players to be able to respond to it. So, combining projectile weapons with slow fire weapon designs, um damage mitigation becomes a much more viable concept in Titanfall 2. Next. But, projectile weapons tend to be really difficult to use, especially combined with the fact that many of our weapons were not rapid fire. So, to help with this, we have tech called fat projectiles that we can apply. Uh and what this does is it gives our projectiles width, making it easier to hit targets with them. Um the nice thing with this tech was that it was also used for pilot weapons and it allowed us to create non-hit scan weapons for pilots that could go toe-to-toe with the assault rifle. But, hit scan can still exist and a good example of this is one of the Titans, Legion, who has a minigun.

Um but his minigun has a built-in spin-up time, so that compensates for the lack of travel time in his projectiles. And we do a lot of work to change his silhouette depending on his weapon state. So, players can easily read whether or not he's ready to fire. In addition, this also makes positioning way more important for a Legion player. So, we kind of accept that he'll have less engaging atomic decisions, but that's okay since we've traded it for an increased dependence on his tactical decisions. Video.

Um another thing that we did was design abilities that created permanence on the battlefield. Scorch's load out in particular features abilities that linger for some amount of time, so it forces his opponent to constantly kind of rethink his position or it pushes his opponent out of position. And something we realized was that abilities or Sorry. Sorry, rather. Something we realized was that Titan fights would often last a long time and Titans would often play in the same space. So, it's having something that can change the meaning of that space would help keep those fights fresh. Um some things other examples of this of permanence that we made were the cluster missile, particle wall, uh the uh Scorch's incendiary trap or the laser trip mines for Ion. So, but static permanence are good, but we found that permanence that move in a slow, predictable path are even better.

So, the slow moving part is important because you want these permanence to last a good amount of time on the battlefield or enough time that players can react to them in a meaningful way. So, if they move too quickly, they'll end up hitting a wall or flying by your opponent and then they're basically gone at that point. And we came upon this discovery when we made this ability called the arc ball, which oddly enough didn't ship as a Titan weapon. Um the arc ball or the thunderbolt which as a pilot though that's the pilot version of this weapon, but this specifically is the arc ball. Um this is an ability that launches a slow moving projectile. Can you see my pen? No, you can't see my pen. Can you see my cursor? So, this is our slow moving projectile as you can see. And you can see it's this projectile that kind of zaps the opponent that's within range of it. Um so, when we play tested this, we found that it really added an extra kind of uh an extra layer that would really change the the dynamic of the fight, but

only for a certain amount of time. So, here's what that looks like in practice. If this video would play. Maybe if I click. So, you can see in those in that video, there's multiple ways to kind of use the arc ball, not just to damage your opponent, but to force him into certain bad situations that you can take advantage of. And this kind of mimics Hopefully this GIF will play. Oh, I hate you so much. Yes. So, this mimics what in Street

Fighter people um the slow fireball in Street Fighter which players can use to um I think they call it fireball zoning. So, what it does is it forces their opponents to react to the fireball and possibly put themselves in a bad position. Um finally, the last element that we worked on was improving our Titan's core abilities. In Titanfall 1 core or sorry, in Titanfall core abilities are essentially supers or alts. They are ultra low frequency moves that can have a high impact on the battle. Titanfall 1 had core abilities, but they were mostly things like double damage or increased health which were okay, but they didn't really create the spikes in gameplay that you would want from a from an ultra low frequency attack. For Titanfall 2, we had to rethink our core abilities to be more attack-like with the goal of having them create opportunities to tip stalemates or create reversals. One thing that we wanted to make sure

when designing these abilities though was that they weren't instant win buttons. We still wanted the notion of allowing players to express skill when using these core abilities. So, there's the idea of being able to use a core ability poorly and being able to use it well. The best example of this is Ion's laser core which has wind-up time to warn the opponent that it's going to happen when it and when it activates, the attacker has slowed movement and he has disabled dash and a slowed turn speed. So, a player has to make a judgment on whether or not it's a good time to use that ability. Uh this also opens up a new type of gameplay which is commit and punish which is something that you find in fighting games a lot. So, when in fighting games when someone does a specific move, a character will commit to a certain movement and could possibly expose themself um that can and that the enemy can exploit. So, here's what that looks like for us. If the video would play.

Laser core ready. Firing laser core. Ion laser core activated. Ion laser core activated. We're not ahead by much. Let's pick up Just a sec. So, what we're really doing here with all these changes is we're trying to create new decisions to for the player to make outside the atomic layer. So, we kind of just made this new layer which for now I'm calling the technique layer. And the technique layer is this loop of

decisions that occur in roughly 2-second windows and they have to do with assessing the current board state. So, not the whole map, but just this fight in front of you um and thinking about what abilities you have, what abilities the enemies have, and then executing your next move. Uh somewhere in there you're also thinking about, you know, your cool downs and trying to figure out, you know, whether or not your next available ability is something that you can possibly combo your current abilities with. Um and this layer is sort of present in low time to kill shooters, although it's very underplayed and it's present in the form of grenades as a pre-engagement attack. So, in a low time to kill shooter, if I know that there's an enemy in the next room, I could throw a grenade in there and increase my chances of winning that fight by pushing the enemy out of position prematurely and then going in knowing that he would be out of position when I walk in. Okay. So, now that we built our combat

depth in multiplayer, we can fairly easily or you know, we can apply this depth to single player. So, when we started working on Titan combat in single player, our first instinct was just to assume everything would work great. We spawn a bunch of Titans, we you know, lay out combat the way we normally do, no problem. However, we ran into the problem of combat fatigue where players would often get pretty tired after a Titan fight. So, to solve that we experimented with what we called weak Titans and that's a concept where presumably we would say that auto Titans would be easier to kill than say piloted Titans. And you know, when we implemented it at first it felt great, you know, I go play a fight and I kill a bunch of weak Titans and then eventually once in a while I'll fight like a non-weak Titan uh and so, you know, people were reporting that they were having fun. Um but you know, we let this sit after

some you know, we let it sit for a couple months and after further inspection, we realized that being able to mow through a bunch of Titans in a row, it really made feel Titans feel, you know, weak um and like they weren't special anymore and they're too disposable. Um in addition, we had all these cool behaviors from multiplayer that weren't being showcased because the Titans would die so easily. So, we scrapped that idea and instead split the difference. We increased the health back to their original values and then pushed the differences in health. So, that's our MP health and this is what the health values look like in single player. So, the light chassis has around half the health of its multiplayer counterpart and the medium chassis has around 80% health. Now, a 50% light chassis health is it's still a decent amount so they can definitely still have some time to use their abilities. Uh at 50% health, I believe fighting a Ronin on average still takes the player

around 15 or 12 seconds. So, this allowed but This also allowed designers to tune fights a little bit better. So, if you wanted a an easier fight, you would drop um a couple of light Titans like a Ronin or a Northstar and we found that good fights usually were comprised of maybe one heavy Titan with a couple of light Titans backing him up. Um but in general, we still found that fights were pretty tiring. So, we needed it we needed to break up Titan fights even further with kind of infantry sections or Reaper sections. So, if you don't know what a Reaper is, he's kind of like a half Titan. He he's got like uh one chunk of one chunk of this health uh and he's half the height of a Titan so he he can die pretty quickly to a Titan. Um So, so these these kind of inter uh infantry sections proved to proved to be good not just to break up the fights, but to sell the scale of Titans as well.

Cuz if I'm a Titan and all I come across are big Titans, I don't really get the sense that I'm this giant robot. I just everything's my size so whatever. So, since we ended up with high-ish health Titans, we generally tried to have Titan fights be like 1 v 1 or 1 v 2 1 v 3 and this allows players to focus on the enemy Titans more since we also found that players had a lot more face time with enemy Titans than in low time to kill games. So, this increased face time means that we had all this time to play with so we used it to give enemy Titans more personality. It was a good opportunity to kind of build a relationship between the player and the enemy who would be fighting. And to do that, we wrote a system called the VDU system that would detect player actions and make enemy Titans respond accordingly. The idea was that when you were fighting an enemy Titan, that Titan's pilot would

have an open communications line with you and he would taunt you and stuff. So, it would do things like when the player damages an enemy a certain amount, he'd respond by saying, "How dare you?" Um or if the player used his core ability, you know, the enemy would say something about that, too. Militia Titan spotted. Chassis number BT-7274. Sorry, but this is your dead end, pilot. All right, let's end this quick.

Maybe you're stronger than I thought. Not by much, though. It's been fun, pilot. SALVO CORE AWAY. ENEMY TITAN DOWN. SO, REAL QUICK, LET'S TALK A little more about our Titan AI. Um again, since players are now spending more time with each enemy, they have a lot more time to notice AI behaviors, so it became worthwhile to invest in a more robust AI system. Um in designing our AI Titans, we decide we decided to go the route of Street Fighter bot fights. So, we like that when you fight against a bot in Street Fighter, um all kind of normal thought processes

that you would go through when fighting a player still applied uh more or less to fighting a bot. So, our Titan AI generally tries to act like a player, and they follow the same rules as a player. They only block or dash once you fired, and having projectile weapons again here will help because they can take advantage of that travel time to make that decision. Um they also follow the same cool down rules for their abilities. So, it's also poss- so, it's also possible to bait certain abilities out of an enemy and then counter with your own attack. Uh nice thing is these Titans are used both in multiplayer and single player, um but single player just adds a layer of characterization on top of the multiplayer AI. The last thing I want to talk about is our AI boss battles. In Titanfall 2, spoilers, players will fight a number of boss characters which are the different mercs that Blisk has hired to help him. Due to time constraints, we didn't have enough time to hand script each boss battle like how we would usually think

of a traditional boss fight. Instead, our bosses just use the same system as our other Titans, but they're just more aggressive about using their abilities. And what I mean by that is that our normal Titans will usually wait a little bit after their ability's ready before they actually use it. But, our boss Titans will use to try to use their abilities as soon as they're available. But, we still wanted to make our boss battles feel more special, and there are a few things we did for that. So, one is all our bosses have these cool intro showcase intro cutscenes that showcase their personality, and with the exception of a few special cases, this is the only time we actually see cutscenes in Titanfall 2. Two, bosses have regenerating shields just like the player. Not only does this make them harder to fight or deal damage to, but it also enables them to act just a little bit smarter than other Titans

since they'll also try to preserve their shield. So, boss Titan will try to retreat behind cover to get to pack to try to get his shield back. Bosses also have doom state, which is an extra layer of health that puts a Titan in kind of an executable state. And we removed this for our normal Titans because again, it extended their health too much. But, it seemed appropriate for bosses to have. Um the nice thing is this also makes it so that bosses are the only ones that only Titans that can get executed, and each boss actually has a special execution animation tied to him. Finally, boss Titans have extra video callouts that they can use, and they have a special pool of just taunt dialogue that they can say if they, you know, if they haven't said anything in a while. So, they'll call out they'll in addition they'll also call out if they're being aggressive or defensive, so they seem just a little bit smarter. Um so, let's see what that looks like. You were not invited to Cain's party,

and that's why you're dead. Well, well, another runaway hero. With an SRS Vanguard class Titan. WOAH, NOW WE'RE TALKING. BRINGING A KNIFE TO A GUN FIGHT. LET'S GET THIS PARTY STARTED, SCRUB. Walk away, scrub. Just walk away. Only a matter of time before I'd rip you out of that cockpit anyway. Target hit. Fox eight. Nice try, love, but I'm no pushover. Time to die, Chronus one. All right, I'm ready for more. Goodbye, you little bastard. Dodge this. He's using his core. So predictable. You should not have done that, pilot.

Who's got core ready, hard shield, too? I won't get me destroyed by some pitiful excuse for a pilot. Watch out! GET SOMETHING NO! VIPER ONE LOST CONTROL. I'M LOSING CONTROL. Going down. Mayday, mayday. Going down. You cannot compete in THIS COMBAT, EVEN NOW PILOT. OVERALL, I THINK GIVEN THE CIRCUMSTANCES, it worked out pretty well. I think the fact that the bosses actually use normal AI makes our boss fights feel kind of less scripted and more real than the traditional ones. Uh I do think there's a lot of

improvement though, and I think there may be a middle ground where we can have our boss Titans use their AI, but designers can script in kind of special events to to sell the spectacle. And you get a little bit of a glimpse of that in the Viper fight in our game. So, the Viper fight is one that was more special scripted cuz the designer for that really wanted to do it, and he's the one that flies around and does a special core ability that doesn't appear in multiplayer. Um so, I think in the future, I would try to push more in that direction. Finally, let's talk about the unexpected problems that we ran into in during development, and you know, some of these problems we haven't even really figured out the solution for, but these are just things to watch out for. The biggest pitfall in the direction we took was that all these different loadouts really increased the complexity of Titan combat. Um by introducing the skill of ability combos and enemy recognition, we created a potential greater potentially greater skill gap between players that played

the game a lot and know all the characters and the players that don't. And this can be partially solved by good matchmaking so we don't ever put two players of different skill levels in a match together, but matchmaking can only really go so far. So, I think more work needs to get put into that. Um other league games such as League or Street Fighter, you know, they do a lot of work both inside and outside the game to give players access to information about their characters. We did a little bit of this with the intro videos for the Titans, but I think there's still a lot more we can do. Something like a training room or or a certification course would be ideal so that players have a safe space to practice using their loadout. Um another thing that added to this was that when we started working on Titans, we primarily focused on making sure players were having good one-v-one fights. And this was all well and good, you know, we built kind of a one-v-one map and we would play test there and just have like really fun one-v-one fights or even two-v-two fights.

Um so, this is all well and good, but it was became really disastrous when we actually played a real match of attrition, a six-v-six match of attrition, since players were having to deal with not just more Titans, but the potential of pilots sneaking up on them as well. So, this means that we actually had to tone down some of the more obtrusive effects that were again working in one-v-one context, but were pretty bad in kind of a six-v-six concept context where anything could happen. Um so, and a good example of that is Northstar's plasma railgun used to apply a really huge knockback effect when you whenever you would get hit. And this was great in one-v-one since it really helped him force his opponent into his into his ideal range, but you can imagine when playing a six-v-six match, you can get get tagged by a Northstar railgun out of nowhere, and then all of a sudden you're shifted like 36 uni- or, you know, 3,000 units to the left, and you don't really know why. So, we had to take things like that out. The other problem that presented itself,

um Sorry. The other problem this presented was that we also tuned health values to be again, to be interesting in a one-v-one context. Uh again, our intention was to make sure the Titan fights, you know, had enough time to feel like a real fight. Again, looking at Street Fighter fights as reference. In practice, however, we found that whenever two Titans would get into a fight in a real match, they would effectively kind of end up going into their own little corner and play out their fight that took forever, and they would effectively ignore the match. Um and we had this one I remember in particular this one event. There was this hardpoint match, and I was a Ti- I got in my Titan, and we're running around. I tried I and I found another Titan, we started to fight, and it was the the match had already finished like the victory screen went up, and we were still fighting, and the evac happened, and everyone left, and we were still fighting. And like we basically just ignored half that match, and so that was we we determined that that was not a good thing.

Um this problem of complexity exis- is exacerbated by the fact that you're in a Titan for about only really 30% of the time in Titanfall. So, this means that even if you wanted to practice using a certain Titan, you would only have you would have to play this pilot game for a while and wait until the next one is available, which is not really ideal. So, right now if you really wanted to practice using a Titan, the oops, the best place to do it Woah. Let's do that again. The best So, if you really wanted to practice using a Titan, the best place to do that is in last last Titan standing, which is a game mode that always starts you in a Titan. But even that game mode, it's an elimination game mode, so you got to wait after you die before you can, you know, do it again. Um in single player, this is the main reason why we switched to the on-the-fly loadout switching for BT. Initially, players would pick up a new loadout from enemy Titans, but we ran into the problem that they might pick out a

loadout that they didn't like or they weren't good with, and then they'd basically be stuck with that loadout. And it takes a long And since it takes a long time to learn a loadout, it really disincentivizes players to seek out new loadouts. They would just stick with the ones they're comfortable with. With the Mega Man style loadout system, players could kind of kind of try a loadout without the fear that they might not like it cuz then they could always switch back to something they're comfortable with. So that at some point at any point in time, they can just kind of dip their toes in and then possibly be better. Uh Also, in single player, this pushes us to create a dynamic dynamic hint system. So what our hint system does is it tries to kind of it detects what you're doing in a fight and kind of takes a lot of data. It tries to figure out how long you've been in a fight and if you're only using your primary weapon in that fight, it'll signal you to use your other abilities. Um Unfortunately, the system isn't super duper smart yet, so sometimes when

you're taking damage, it'll just tell you to use electric smoke. Um really the intention is it tells you to use the electric smoke and then take then go to cover, but people didn't really um know that all the time, so there's a little bit more improvement we can do for that in single player. Oh, and also we should port that to multiplayer. Um another specific thing to high time to kill is the problem that it's difficult to communicate to players where they made a mistake. For low time to kill gameplay, we have a feature called kill replay, which will show the players their final moments before they died. Um this 8-second window is usually enough to be able to show players exactly how they died so they can do better next time. But high time to kill fights could last for, you know, a minute, and showing the final 10 seconds of someone's life may not reveal anything valuable. Um in addition, players will often eject out of their Titan instead of dying, so they aren't even in a mindset to absorb that lesson since they're still busy trying to run around shooting other Titans. Um so this one we're still unsure of how

to solve this, and some ideas have been brought up like showing a feed of major damage sources, but I think more research needs to be put in in regards to how we can present this information. As I mentioned in single player, when we started designing combat encounters with these Titans, our first instinct was to lay out one fight after another, similar to other shooters we've worked on. But we found that even just one even after just one fight, players got pretty tired since Titan Titan fights are a lot more involved than pilot fights. So we had to make a conscious effort to space out Titan fights more and insert infantry combat in between just as a pilot cleanser. The only exception is levels like trial by fire and the fold weapon, which is this. So this is trial by fire, and in this level you're fighting a bunch of Titans in a row, but we have additional mechanic we have friendlies in this level, so they can help kind of take tank the damage from other Titan enemies, and they can also deal damage to enemies, so it helps you

kind of just push through that level. And in this level, this is the the last level, the fold weapon, and in this one you also fight like a bunch of Titans, but in this level we have a mechanic with ship bombardment, and that'll bombard all the enemy Titans and basically turn them into weak Titans cuz they're all now they're all damaged, and you can go mow through all those weak Titans pretty quickly. Um this is generally less of a problem for our multiplayer since there are less Titans on the field overall due to the whole pilot Titan pilot thing. Um and in last Titan standing, we have natural breaks that are the round transitions, so those are also good cool down moments. But I imagine if you were to make a multiplayer game that's like high time to kill all the time team deathmatch, you would probably run up against this problem as well. Finally, the last problem is more of a Titanfall specific problem, and it has to do with regen health. So regen health is great because it makes type you know, it'll force players into this position where

they have to retreat, get behind cover, stop, think about what their next thing is going to be. But for Titanfall, it was bad because we needed to Titans to cycle out. Um we In Titanfall, we were dealing with this problem called Voltron, and Voltron happens when one team gets four or five Titans together and they just, you know, gang up on one Titan and they obliterate all the other team's Titans. And now the other team has zero hope of bringing of of winning the match cuz every time they bring their Titan in, they just get stomped. Um and in Titanfall 1, this was a huge problem because of regenerating shields, it was very difficult for the enemy team to even damage the Titans. So removing any sort of health regeneration kind of helped us a helped the other team kind of bring other Titans down. But again, in single player, we kept regenerating health cuz that wasn't really a concern.

Um so this is the end of the talk and then we'll just recap real quick. So high time to kill negates depth at the atomic level, so try to push designs that embrace high time to kill. For us, fixed loadouts fit really well with high time to kill. They push enemy recognition and ability combos, and it fit that's type of stuff fits really well with kind of the increased face time. Um think about how to add more interesting decisions within and outside that layer. Again, for us, you know, low frequency attacks and permanence were good step kind of first step, but there could be more out there. Be ready to handle increased complexity and watch out for combat fatigue. Finally, I just want to point out that when we started this project, you know, high time to kill was kind of seen as a little bit of a problem child, and there was definitely a lot of temptation to lower Titan health values. But I'm really glad that we really embraced the high time to kill nature of Titans as it made us it really pushed us to create

something that I feel is really unique in terms of shooters in in terms of shooters, and it kind of opened up this new design space for shooters where ability design can really flourish and you can have more room to create more weird and kind of cool abilities. Uh and I I'm pretty excited for what, you know, the future holds for that. So that's the end of the talk. I want to call out just special shout out to these people up here. They in These people in particular, a lot of their work is also represented in this talk. So Mac McAndlish, who was lead designer. Uh Steven DeRose, who is the other guy that worked on Titans. He really helped push Titans to be, you know, super polished and shippable. And then G Sang was the AI programmer who worked on Titans as well. And that's all. Uh I don't know, we have like 4 minutes for questions. Also, Respawn is hiring.

Hey, how's it going? Uh I'm wondering if you experimented with damage regions to offer more choices at the atomic level. So like I shoot off a Titan's arm and it can't use a gun. I shoot off its chassis and it can't use its core. And if you did experiment with that, why did you not take that? Um I don't know if we experiment We didn't experiment with that in Titanfall 2. We experimented with damage regions dealing different amounts of damage, but not like losing your arm cuz then you basically would just be walking and then you couldn't play anymore. Yeah. You would like lose a segment of gameplay that would normally be, you know, important to you. Um with regards to different damage values, something interesting that I didn't actually put in this talk was In Titanfall 2, we reduced the amount of damage uh buffs that the crit point gave you because we found that we were able to put depth in other places and having variance in the amount of damage you can

deal was bad for the guy receiving damage cuz he also needed to predict how much damage he was going to receive, and he can't really control where the guy's going to hit him. So we So reducing the amount of variance just kind of helped improve the predictability of that. Thank you for the talk. Uh hi. I was curious uh with the balancing of all these six uh abilities and Titan classes, I was wondering if you experienced any particular like challenging areas and uh if you have any comments on like any stuff you've seen shake out from the players uh life testing it for you. Um sorry, could you Would you mind being more specific? I don't think I fully understand your question. Okay, I was just uh curious how the uh what your experience was like balancing these six different like Titan classes. Oh, so the question was about um just in general balancing kind of the six classes? Yeah. Yeah. Um so I didn't I wasn't actually the guy

balancing it, just, you know, to put that out there. But overall, when I think when we were making these Titan abilities, it was kind of a weird it was especially in its infancy, it was difficult to figure out how many abilities a Titan should have and which it we There was one point in time where we experimented with sharing kind of maybe they should they should have one slot which had shared abilities. Uh but it turned out it was just much better if everyone had their own unique thing. And then balancing them, you know, you just play test and then if a specific thing was too strong, you would pull it back or if it was too weak, you would buff it. You know, it's all you're just getting feedback from different players and you know, as if you've been in chin stock, what the multiplayer team play tests every day and then every week we have Friday night fights. So, we got a decent amount of feedback from uh from our dev team. And then from the live game, you know, we we look at we look at the Reddit, obviously, you know, that's kind of a slice of our community. Trying to figure out if any meta has has

emerged and then figuring out what we want you know, we we balance the live game as well. You know, we'll we'll change some numbers depending on what players' activities are. Thank you. Great talk. Thank you. Hi. Thanks for a great talk. Uh I was wondering in single player if you had any tricks uh for scaling difficulty without falling into those some of those same traps of bullet sponges on one end or Titans not feeling special on the other end. Um specific Well, I think the once you increase the health, it's So, basically what I found is you can once you start increasing health to bullet sponge levels, wait, is that your question? Sorry. Is your question if I increase the health to bullet sponge levels, what should I do or what can I do to not increase bullet sponge levels? I was more curious if you had any uh more creative ideas than just scaling the health uh since you pointed out that sort of leads to bullet sponges or

Titans not feeling special at the other end. Uh well, there's what did we do? We did kind of shield usage, which is a little bit of health, but it also um it's just a little bit of extra health that they can regenerate. So, it also helps make makes them feel smarter. Um you can play with how often they dash or avoid damage or block. Um when one version of Titan AI when they weren't really tuned where they would use the vortex shield and they were so good at using the vortex shield, they would vortex your bullets and then fire it back at you and then you would catch it cuz you know, you saw them use it. But then then you fire it back at them, then they would catch it right away and then fire it back at you right away. So, you could try to make them just be more aggressive with how they use their abilities, too. Uh I guess things like that. Hopefully that answers your question. Thank you. That's it. I think we're That's time. Uh there's a wrap-up room over there if anyone wants to talk some more.

But I think people need this room. Thanks for coming.