In The Lair Of The Teutonic King - Talking Dragons, PS3 With Factor 5's Julian Eggebrecht

At a recent event for Lair, in which the PlayStation 3’s finer points were put on display for a group of select journalists, Gamasutra spoke one-on-one with Factor 5 president Julian Eggebrecht about the game and its future.

In particular, Eggebrecht highlighted design considerations for the Sixaxis, Factor 5’s discoveries in terms of how to use the PS3’s SPUs and RSX GPU, also hitting on the company’s work on a new entry to the Turrican series.

You're obviously taking advantage of the Sixaxis. Why do you think that hasn't happened so much so far?

Julian Eggebrecht: On the one hand, you have to have the time for it. On the other hand, you have to have the right game for it. Lair was just screaming Sixaxis. As I said in the presentation earlier, I'm a firm believer by now that you really shouldn't force motion control on functionality which it really isn't made for.

So that's why, for example, we switched the ground control back to the stick. Lair just lends itself in so many aspects to the motion control. It actually is better than the stick. For some other game types, there might not be a single aspect which is better with motion controls, so [it would be better to] leave it out.

As I've also noticed with the Wii, sometimes it's a little difficult to know the limits of how far you can turn, for example, in a game with motion controls.

JE: In Lair, we're mapping the level of the controller exactly to the dragon. Basically, if you lean the maximum 90 degrees, the dragon will turn that much. The angle matches perfectly.

We found that very important, because that came after focus testing after TGS. We were watching people fly, and they were having some fun, but they were having more problems than later on, when we made sure that [the controller matched on-screen movement]. Humans seem to have a pretty keen sense for what the angle of something is, and are able to correlate that subconsciously to what's on-screen. It helped a lot.

Did you have to hand-animate the dragons? I saw videos of the mocap that was done for the humans, down at SCEA San Diego – it was really nice.

JE: The dragons obviously have to be hand-animated. There aren't any dragons to motion capture!

You could’ve used a lion or something!

JE: The interesting thing about the dragons is that most of the time they're actually biped. They're kind of dinosaurish. The sole reason for that is that you want to have the front arms able to do something.

We had an early prototype where the dragons were completely four-legged on the ground. It made it way too hard to control. Just spinning on the spot was impossible, so it turned very much into horse-like or tank-like gameplay. That's actually why we raised them up a little and went with more T-rex-like proportions.

From the presentation, it feels like there is a lot of importance and focus being placed on Lair on the Sony side. Sony always used to focus on exclusive titles, and that's not as possible now with budgets people have. How important do you think first- and second-party titles are to Sony's and the PS3's future?

JE: I think they make all the difference. That's why it's important that you have a strong first-party slate throughout the year. The fact that the other games are multiplatform inherently means that most of them probably won't take advantage of the platform the way they should. That applies to both [the Xbox 360 and the PlayStation 3].

I'm sure that there are enough titles which come out on 360 because they also have to be on the PS3, which don't even take full advantage of the 360. It's a two-way street. That's the one thing where you really have to show the strength of your system: with the first-party exclusives.

Secondly, there's this generation's very interesting structure regarding controllers. We've never had that before, that we've had so vastly different controllers. And who else but first-parties will take advantage of that in the way it should be?

It seems like a game like Lair would be relied upon to move system units, too, when you have multiple SKUs and other systems have been out longer.

JE: We've never designed like that. Lair is fresh in terms of genre because it's a new genre - we call it "fight and flight" - and using the controller actually came out of the fundamental design involving dragons.

When the dragon thing happened, the motion control aspect was just a glimpse on the horizon. It might have happened, it might have not happened. I think we just lucked out, but on the other hand we also always try to use every single feature in a new machine simply because we get a kick out of it.

How far do you think there is to go, graphically, on PS3?

JE: It's huge. I think (PS3 GPU) RSX isn't a big secret. What's really interesting is that we're starting these days in the optimization process for Lair to use Cell to do certain things on the graphics side, which you normally wouldn't expect.

I think there is a huge room for growth. RSX, as well as the GPU in the 360, are known quantities. Around the middle of the cycle, most people will have figured out what you can do with them. But then you suddenly have the connection between Cell and the SPUs and RSX, and you can do a lot more with that, graphically.

How are you splitting stuff up to the SPUs? What sorts of things are you finding you're able to stick in there?

JE: We initially started out by saying, "Wow, we're going to do physics," and everybody should do that at least. But it very quickly became a matter of "oh, let's put this on the SPUs, and now let's put this on there." We've got the fluid dynamics, all of the physics including ragdoll, and all of the collisions.

In a game like Lair, if you've got a couple of thousand soldiers running around and hundreds of dragons, one of the big issues you have is a ton of collision checks, which other games simply don't have. It's always been a big issue, and the SPUs are perfect for number-crunching like that. Other things which you can do nicely on the SPUs is to prepare tasks for the RSX, which normally you'd have to do with the CPU, because the GPU really can't do it at that moment.

We've got a lot of things, including army AI. If you have primitive AI for the distant armies, you can easily run that on the SPUs. As the army comes closer and these guys need to get more intelligent, you move the more intelligent army AI code onto the PPU.

So you've got to find these certain bite-sized areas that you can optimize?

JE: Yeah. You need to find all of these systems, and you need to pre-design systems. We found that out rather early on in a nasty way with the collisions. We assumed our collision system wouldn't be done on SPUs, and that was a big mistake. We designed it around general access and not small chunks.

We had to completely kick that out the door and redo it. We realized that we should look out for that in the future, and every single system which would even remotely lend itself to that was designed like that from the beginning.

How close is it to release?

JE: We're releasing in July. The state of the game we're in right now is alpha. Alpha means that we've got all of the levels in final memory, which was our biggest push. Now that all of the features are in, we start optimizing, tweaking, and bug fixing.

Can you tell me more about the progressive mesh? I noticed that things were being generated on the fly, and mountains were just kind of popping up. It's cool in a way, but it probably shouldn't happen.

JE: In the end it comes down to tweaking, and we're probably going to tweak the progressive mesh until the last day. Somebody who is running on a 1080p system and who is really accustomed to look for stuff like that will probably see a little bit of growth here and there. Then again, at least we can do something like that. Without progressive mesh, I don't think that the thing would have been possible, quite frankly.

Are those mountains generated differently each time? What is important about progressive mesh?

JE: Progressive mesh is not only the mountains. Every single thing you can see on screen is running through the SPUs' progressive mesh code. On the landscape, it's basically using more traditional techniques. It is progressive mesh, but we've had landscape generation since the GameCube days. We never had a CPU which was strong enough to do it for the whole world, though. It's really cool.

The big thing on the landscape is the detail level. I don't think any landscape has been done with that level of detail. Back in the day for the GameCube landscapes, the typical approach was to tediously hand-texture everything on the landscape. What the RSX does in Lair is we give the graphics chip a pixel-shader program which is actually a rule set. That makes texturing of landscapes an extremely quick thing.

Why draw from the movie world for your script and score? Obviously they have established good work there, but it's a different industry from ours. What is important about that Hollywood feel versus the game feel?

JE: I don't think that there's a Hollywood feel when it comes to music. I think music is something universal, and I think that we simply don't get the best composers. It has nothing to do with Hollywood, per se. It's just the attraction for someone who writes a good orchestral score, and has a choice between working with movies and games. I still think, sadly enough, that most people choose to work in movies.

The deal with John [Debney] was very unique in the sense that when we thought about using somebody who has mostly been working in movies, I was really afraid of it. On the one hand it was a matter of who can deliver something like John Williams, but on the other hand there was the matter that all of the guys who can do that are probably not interested in games, which means that we're going to get this half-baked soundtrack that doesn't have much to do with the game, and to work with them will be a real pain.

The cool thing about John was when he really started asking the right questions. I think his son had gotten him into video games just a couple of months before we contacted him. He was playing, and he was wondering about the medium. John has an interesting history, because before he became the John Debney we know nowadays, he was actually at Disney and was composing songs for the theme parks.

Award Winning Composer John Debney

He was very much an on-staff composer back in the '70s. So he has dabbled in the media, because the requirements for theme park music are quite different for what you do for a movie score. He actually tapped back into that experience when he was working on Lair.

We explained that we needed to have pieces which can quickly fade from one piece to another; we can't have long, drawn-out tracks. Normally, every composer would just say, "Come on, you guys, I'll compose you something and you figure it out for your game thing." John really said, "Ah, interesting. Okay, then let's figure this out." It wasn't the Hollywood thing with him specifically.

That's good, because a lot of the time when you get Hollywood composers it ends up like that. It's a game, so obviously the music should be somewhat interactive.

JE: You know how interactive our Star Wars games were. Basically the best example of that is probably when you're landing on the ground. There's drums setting in, in addition to the piece. The whole mood is changing. That was actually created on huge boards which we did with John and our music director. They really mapped out for what motion and what branches are possible at certain moments in the game and in the level.

John really looked at it in his score and said, "I can give you three types of that general theme." The general theme is important, because it's this overall arc that goes through the whole game. But some moments you need something where drums can come in and more tension is created, or it can fade down. He really did all of that collaboratively. It wasn't something that we had to suck out of the material at the very end. That was really cool.

So I hear that you hired on some guys for universe creation?

JE: No, the universe creation we did at Factor 5. We hired the guys when the universe was finished. That was the whole point. We did the whole universe and the basic overarching themes and certain archetypes of characters we knew we would have to have in there. It's the approach that George did with Star Wars. Before he wrote the story, he created the universe.

We had all this pre-design going on. We had a universe, and then we reached out for scriptwriters and basically said, "Here, guys. This is your universe to work in. Now create characters and a plot." That was the process.

Is it worth the extra cost to get that sort of extra world creation, culled together from different cultures' tapestries? Will that ultimately pay off?

JE: Yeah, I think so. The most convincing IPs which are being created in the game space are ones that A) have a vision that's strong, and B) go to that sense of detail. The best examples are with BioWare and their games. Of course they're going to these lengths, otherwise you wouldn't get a believable IP. Some of the other guys out there who create successful IPs always go to that length.

You were saying that if the game is really successful, there will probably be a sequel. How do you measure the success of a game like this, given that we're still in a part of the PlayStation 3's lifecycle where not many consoles have been sold?

JE: In the end, you have to see it in an absolute way. It has to break even. If it breaks even, I think for a new IP that's almost the best thing that can happen. The best thing that can happen is that you've got such a huge installed base that you basically go gangbusters in terms of profit. But quite frankly, for Lair, I would be happy if it breaks even, because that would be something huge already.

What do you think it would take for that? One million units or so?

JE: I don't know, but we'll see.

Can you talk at all about any PSN titles you will be working on?

JE: I cannot talk about it. We're working on it, that's the only thing I can say.

Any progress on Turrican?

JE: Yeah. We've been concepting quite a bit internally. That's another universe creation thing. I was looking at Metroid Prime's reinventing of a franchise that had been out there for quite awhile, and we're facing the same thing with Turrican.

There's aspects of the old games where people will feel betrayed if we don't transform them into the next generation. On the other hand, there's other stuff which is simply cheesy, let's face it. I don't think gamers will accept those things anymore. It's a fine line to balance.

It seems like it would be difficult to get some of those things across, because some of those things really have to do with 2D level design. The way you explored had a lot to do with vertical and horizontal scrolling.

JE: Yeah, 2D was all about world exploration in our games, but also about scale. That is one of the things we've transformed into our 3D games, where it's all about scale from macro to micro. I think some of these elements actually do apply, and they're quite different from what you've seen, say, in Metroid, which has a very rigid design.

So, [in Lair], why are all these guys on the ground when there are all of these dragons in the sky that can tear them to pieces?

JE: That's not true! They have kites on many levels loaded with explosives, so if a dragon approaches too closely, the thing blows up and harms you quite badly. They're not completely exposed.

Do you think you're going to eventually get to a scale of actual warfare? Right now (in the bridge level) you've got like ten guys on the front lines fighting, while everyone else is just waiting in the background going, "Yeah!" I know it's got to be difficult because you've got to separate the blue army from the red.

JE: That's the inherent problem. It's not so much the problem of displaying all of these guys, it's how you then interact. You immediately get punished if you start whacking your own guys too much. That's why we kept it separate. It's a tricky design issue.

I know that if we should do a sequel, we want to refine the army combat for sure. It's a tricky case, because you're this monster, a tank in the middle of a bunch of guys, so you would probably kill too many of your own. Having said that, though, there certainly should be ways of getting more fights next to you going, so that they also cleverly avoid you so that you don't accidentally kill too many of your guys.

It's interesting how design scope kind of explodes when you get these worlds so much bigger. When you're getting close to realism, it calls all of your old tropes into question.

JE: But it actually helps the gameplay so much in ways you didn't imagine. You're kind of afraid of it in the beginning, because you're saying, "Oh boy, the old tricks won't work. I don't know what the player is going to do." But embrace it, because in the end, as we've seen in Lair, those levels are the best ones.

There's quite a few plate-spinning levels where you have sandbox-like gameplay, and they are by far the best. They're so cool.