Enemy Territory: Quake Wars Interview
With id Software's Kevin Cloud & Splash Damage's Paul Wedgwood
Quakecon 2006 is over, but before the next one happens we'll very likely see the release of Enemy Territory: Quake Wars, a new online shooter by the guys at id Software and Wolfenstein: Enemy Territory creators Splash Damage. I got the chance to sit down with id's Kevin Cloud and Splash Damage's head guy, Paul Wedgwood, to find out more about this ambitious title.
Finger: DOOM 3 had per-pixel hit detection, and a lot of people liked that feature. Is that going to be in Enemy Territory: Quake Wars?
Paul Wedgwood: Nope. We had the per-pixel stuff for about the first year and half, and it's very accurate, and on paper it makes a lot of sense as to why it would be that you'd shoot at somebody and would want to hit [only] the body. But the problem is that there's actually quite a lot of "negative space" when you look at an animated player that's running along. So to make the bullets do any worthwhile damage over a fairly decent distance, as with an assault rifle or something, you end up having to increase the power too much, and it ends up feeling random. When a bullet does hit the body it does too much damage and then when it doesn't hit the body. So we reverted to a traditional "hitbox" approach of having the head, torso and legs, and it really improved the quality of the gameplay at the time. So we stuck with that.
Finger: And vehicles...?
Paul: You still have a hitbox which represents what you're looking at, so it's not a rough model that actually represents the shape of the vehicle.
Finger: Megatextures. Will this technology mean ET:QW will take up a ton of hard drive space, and are you planning on a DVD release?
Paul: We've always focused on trying to make the megatexture as small as possible, so one of the things to bear in mind is that when we compile the megatexture initially, it's six gigabytes, and obviously this is an unrealistic size to distribute. But we've already got the megatextures to around 400 megabytes each [after compression], so currently there's no reason why it shouldn't ship on a single DVD.
Finger: Are you doing DVD only, or is that not determined yet?
Kevin Cloud: That's an Activision decision, but my assumption is that that will be the case.
Finger: For mapmakers, is there anything they can do to reduce download sizes for a map that they created?
Paul: When you set the compression level on the megatexture, this is something you choose based on the quality versus [file]size, so we have a tool called Megagen and another tool called Makemega which actually makes the megatexture once you've set up all of the properties for it and how you want it to look. When you compile the megatexture, it goes through a series of DCT and DXT compression, and the DCT compression algorithms allow you to get it pretty much as small as you want. But obviously you start to get artifacts and things at a certain level.
So what level designers could do in the community is they could ship their maps with very [low-quality] megatextures for people to sample the gameplay, and then people could have an option of a download for a larger megatexture once they've been playing on a server, have enjoyed playing, and they know that it's a map they're going to keep.
Finger: Let's talk about tournaments. What are you guys adding to get competitive players to really take ET:QW seriously?
Paul: Do you mean from the player's perspective or do you mean tournament administrators and server administrators?
Paul: Well, Kevin's idea back with Wolfenstein: Enemy Territory was to go with someone like Orange Smoothie Productions and actually integrate OSP into the game. For Wolf:ET the two most popular admin & tournament mods have been ET Pro and Shrub ET.
For Enemy Territory: Quake Wars, we have an even bigger team within id Software and Splash Damage, real hardcore multiplayer gamers that have loads of clan and tournament experience. Especially at my company [Splash Damage] - our lead level designer, Matt 'Wils' Wilson, plays competitively and has a really keen eye on what's going on in the community and with what server administrators need. So there isn't a phase of development that we've got to yet, since you've got to have everything done in the game to know what features you are going to turn on for tournaments.
But obviously, things like Stopwatch mode, and stuff that you need to be able to play an accurate and reliable tournament game will be in ET:QW. And things that the server administrators might want to control, like the gained experience points on an unranked server will work in the same way that it worked with Wolfenstein: Enemy Territory.
Kevin: One of the things we want to do is make the game out of the box with the mechanisms people need. But I also want to avoid having a lot of different types of server settings. What we see happening is that when you first join the game, there's no way a server browser could have communicated all these different types of server settings. So if a person gets in he's not quite sure what he's going to get, and we want to avoid that.
Also, what you find out is that over time, these type of competitive mods, they iterate themselves as people make them and make more and more discoveries. And in some cases, it takes years to get to the point where the community's happy with exactly that type of setup.
So our two things are that we want to make sure that we're going to have what's good for tournament level play "out of the box", and we're doing an outside beta, we're getting a lot of feedback from tournament-level players and all that. And at the same time, we're going to support the community, release out the game tools and try to work to get those types of mods going and out sooner.
Paul: When you play on a public server, there should not be this disparity between different public servers, of actual gameplay changing between them because of [server admins] turning on and off features.
Kevin: When you jump on to a server, you need to know whether or not you're going to get a rocket launcher, or if one class is filled up because there's already three of them [on your team], that kind of thing.
Finger: For some new player that's maybe played some Quake but hasn't played something like Battlefield - how does Quake Wars make this a little more accessible for new players? This is a pretty complex game when you first start out.
Paul: There are a couple of things we're doing. The first thing is the easy out-of-the-box [system that tells you how to use the game's various items and vehicles], what it does and how it behaves, when you get into a vehicle, how to control it, things like that. When you jump in it will say, "Have you heard this before? Nope? [Ok, here's the briefing.]"
The big thing we're investing time in, though, is what we're calling the Solo Assignment System. What we've noticed is when people get into the map, their biggest questions are, "What do I do, where do I go, and how do I help the team?" It's not "how do I shoot a gun?'. What we've developed is this system in which player interactions, like you running out in the world and pointing out a target or with your radar picking up targets, or the natural game scripts like what objectives there are to take. This data gets collected and then gets sent over to players as missions.
What happens is the game analyzes who the appropriate people to get these missions based on their class, location, whether they have a mission already, their experience points, that type of stuff. So, if you jump into the game for the first time and you're not sure what class to be, [you pick one], you're not sure what that class can do, and even if you do you may not be sure of where to go. It will actually tell you all that. Get a mission - ok, I'll take that mission - get it set up, it puts a waypoint in the world, it tells you where to go and what to do when you get there.
Finger: Campaigns last three maps on any given server, and you gain XP over the campaign. What does the XP do for you?
Paul: Well, there's two sides to this, really. We have the campaign-length experience points, and we have the persistent stats as well. So to talk about the difference between these two, the persistent stats reflect your status to other players but are not gameplay affecting - except for instilling fear in the enemy when they see you coming towards them because they know you're good and you've got a high rank. And we have a number of solutions where this rank is not just based on the length of time you've been playing or the number of things you've shot out of the sky, but how effectively you've played continuously.
But the campaign-length experience points are gameplay-affecting with unlockables. So, you may have a tool, an item, an ability, or an attribute modifier which is unlocked based on your proficiency as a specific character class. All of these experience points are gathered - everything that can be unlocked during the length of the campaign - based on your teamplay. Simply sitting up on the top of a hill and just sniping over and over and over will only unlock one or two things and it's not going to make a big difference in terms of the rewards you receive. But if you're supporting your teammates, deploying "Third Eye" cameras, deploying radars and gathering intelligence on the enemy, AND sniping people (which is what you should be doing as a Covert Ops or Infiltrator character), then you're going to unlock a whole bunch of cool stuff.
We've only recently enabled this feature, and the present implementation is something like Wolfenstein: Enemy Territory, except that there are more physical items in the current implementation. What it means is that when you're playing a character class, you can go and do specific things that you know will unlock a certain item for you in the first mission, and that can affect your playing style for the rest of the campaign.
So if you're a Soldier for example, (and this is just an example - a lot of this stuff is going to shift around in the coming months) one of his unlockables is a really cool machine pistol when he's demonstrated his prowess with light weapons at the start of the game. He can run around and just get some really good shots off using his pistol to prove how good he is with light weapons and upgrade to that machine pistol before he starts killing people with his rocket launcher or his GPMG. Opening up the machine pistol gives him a good backup weapon when he's running around with his anti-personnel or anti-vehicle weapons. Players can actually make the choice during the first map about how they're going to play, and this creates a path for them, a kind of specialism that they can then pursue for the duration of the campaign.
The next time the campaign resets, playing the same class they might choose to do something slightly different at the beginning of the map, so instead of being the sniper they gather intelligence for example, and if they go on these missions, that will unlock different things that will affect the way they might play out the rest of the campaign.So, the unlockables create class specialisms.
Finger: How realistic is the flight model, and are we going to need a joystick to not crash repeatedly?
Paul: There are two sides to this. In a good game, the controls should be consistent across the different things you do. As far as it's possible, when you run around as a player, then you jump into a vehicle, we want to try and keep the controls effectively the same. But with the flying vehicles, this is where the first point of inconsistency happens, because what you really want in a flying vehicle is not to have your yaw on your mouse. You don't need to have yaw on your analog control; it can be an on-off state. But what you do want to have is analog control over banking. So generally, what you have is the inverse of player movements; you would use your mouse to look up and down, but you would use your left and right keys to turn, and you would use your mouse to bank because you need to be able to control the rate of banking you currently have. If you just have an on-off state for bank, then you can't strafe. You need a much greater level of control in your flying. So when a player first starts playing with that control setup, it feels counter-intuitive, and that's obviously a problem for people first getting into the game. So we can offer two control mechanisms [for flight], one of which is exactly the same as [on the ground], where pushing keys cause your vehicle to strafe, but the potential for that player's flying ability is really limited by starting with that control setup.
If they start with the alternate control setup, they will definitely become considerably better pilots. They'll find within a matter of hours that they're at a point where they're able to start flipping around and strafing backwards and coming backwards down valleys and strafe-shoot at people.
But setting aside the controls which are fundamentally arcadey, they're not super-realistic because a helicopter's incredibly hard to balance in real life, the feedback does need to be realistic so the model for the physics is as realistic as we can get it given current technology and the fact that we're using a network model. But even though we're currently running over a network we still feel it's more realistic than most games. So if you shoot something like a wheel of a vehicle, it understands that the wheel is missing from the vehicle. If you drive an amphibious vehicle with no wheels, if you drag it into water, then it works in water because it's now buoyant and it's being propelled by propellers in the back.
Kevin: And you definitely don't need a joystick. The challenge for the flying vehicles is that if you make them as easy to control as walking around, then you destroy the feeling of flying which is what you really want to get, the feeling that I'm actually flying around and not floating around in a box. I think we struck a pretty good balance. If you jump into the Strogg side and jump into the Hornet, you'll get a chance to see it.
Also, one of the things we do have are turbo thrusters, and basically that gives you the ability to just streak across the map, linearly, and the combination works out pretty good.
Finger: I heard yesterday there's going to be a Linux client, and I assume a Linux server.
Finger: Do you plan on Mac support?
Kevin: We haven't made any plans.
Finger: If your computer runs Quake 4 fine, how is Enemy Territory: Quake Wars going to run?
Kevin: If you meet the recommended specs for Quake 4, [it'll run fine].
Paul: The real determining factor for minimum spec is not so much CPU or memory anymore, it's more to do with the generation of graphics card and their feature sets, and I think around [Quake 4's] generation of graphics card, you still had some of them being marketed as that generation when really they didn't have all of the features. They were still using the numerology of saying, "we're generation 5", but really it was generation 4++ and had a funny kind of letter on the end [of the model name] that designated that, so it's difficult to say, "All you'll need is a GeForce 5" because with the FX cards there were some strange cards that didn't have the full feature set. So it's probably ultimately going to be determined by [your video card]. It could be a slightly higher required graphics card than the Quake 4 minimum spec or something like that. But basically, that's our target.
We actually have min-spec systems that we build in the office to see how things play on them before it goes into the full Activision Quality Assurance side and we get a chance to see how performance is going to be. We force some poor soul to play on it occasionally.
Finger: We saw some recent screenshots of new maps in the works that haven't been shown previously. Can you tell me a little bit about one or two of the maps?
Paul: Our approach to maps, and I think this is something that worked really, really nicely from Return to Castle Wolfenstein, when id Software developed RTCW, for the very first time a multiplayer combat map had an actual plot and a theme and a reason for being. You were pursuing a specific military objective. This was evolved in Wolfenstein: Enemy Territory where we added the radio news ticker that was going and you had this guy explaining the story for the map. One of the great things about really good World War II movies and games and books is that they retell the story of a famous battle of World War II. They recreate this experience so the player, and the Medal of Honor series is a good example, so the player is immersed in the fancy of having fought in World War II.
Enemy Territory: Quake Wars charts the Strogg invasion of Earth around [the year] 2065. In that sense it's a prequel to Quake 2 and Quake 4, but what it means is that thematically we have an entire art style in the game and a direction for the game which is based on the natural disasters that they've undergone as they've gone up to this point. This sets a really nice theme for the map. We change each map based on geographic location, and this determines things like whether it's arid, or arctic or temperate, and then we have full dynamic light outside, a different time of day for each map, and we have precipitation like snow and rain and fog. Then we have the unique plot of the level, the unique reason for being as it came from Return to Castle Wolfenstein. In Enemy Territory: Quake Wars, we pay homage to a lot of story elements that Kevin wrote for Quake 2 (Kevin was the project lead on Quake 2).
For example, the humans are able to lead a retaliation against Stroggos by traveling through slipgates and then by landing in drop pods. And those two things, the availability of slipgate technology to humans, the way they cracked that technology, and prototyping drop pods and using them for retaliatory attack, they both feature as peaks in the Strogg invasion story in Enemy Territory: Quake Wars.
So each map ends up being completely unique, except for the actions that you do to complete objectives which we keep consistent so it doesn't become too complex for players. Placing a heavy explosive charge is always placing a heavy explosive charge and that doesn't change - a Soldier knows exactly how to perform his combat role. But the map having one destruction objective, it could be an underground base under a crazed biodome in an arctic wasteland, and another one could be at the top of a hill where you're literally trying to get through a trench system to a specific location.
There are also generally three or four specific objectives on each map for the attacking team, and they are always unique from each other. The surrounding geometry is used as well, so every map ends up feeling like a completely unique gameplay experience.
Finger: You don't have different size maps for different numbers of players on a server. If you had ten people in a game, how is it still fun?
Paul: The thing that's really changed from Wolfenstein: Enemy Territory, is that before you'd complete an objective and then you moved forward, your spawn moved, and not much else. In Enemy Territory: Quake Wars, we have this concept of a piece of territory that you start with as the attacking team and the enemy team controls two or three other territories. You have a linear progression by securing objectives that exist between the borders of two pieces of territory. So at the start of the map my first objective is generally on or around the border between my territory and the next piece of territory that I want to capture. This creates a natural front line between the two pieces of territory. When I complete the objective, it shifts to the next one, my spawns shift up, I get additional vehicle availability, more area to deploy artillery guns and the like.
So to answer your question, the reason why balance or size of map doesn't come into it too much, as long as you have a single front line that helps focus combat at any point during play, then with even eight people playing the server, that's eight people fighting against each other on opposing sides of the bridges. When you get twelve people together, you might start to see aerial combat. When you've got sixteen people on the server and there's a lot of fighting on the bridge, you'll find that people naturally take up combat roles like deploying anti-vehicle turrets because there are vehicles coming, deploying anti-personnel turrets because there are more personnel coming. The game organically grows as the player count increases to use more of the game's features that surround it, but fundamentally it's balanced as long as you have the same number of people on each side of the team.
We've had games where there's been 4 or 5 people on each side and it's still felt like a really intense combat experience because of the way spawn waves work and so on.
To counter that, with a lot of multiplayer combat games, what actually happens is you have a gameplay mechanism like deathmatch, capture the flag, or capture and hold, and what this means is that, especially with the Capture and Hold game type, you disperse across the map to try and capture individual locations, and in doing so, even with 64 people on the server, and if you had ten capture locations, that's 6.4 people per capture location (which is only 3 people fighting 3 people). So even on a 64 player server that's Capture and Hold, the potential for combat is generally only interactions between three people and three people at any given time; you don't tend to have masses of people all competing over one thing, except towards the end when you've got a lot of points captured. With focused combat, because everybody knows where the combat is and where the front line exists, they can choose to engage in intense combat around the front line, or they can play a supporting role like infiltration that might involve going somewhere else around the map.
Kevin: The objective system is more than just focused [combat]. Because they work out sort of like military missions, once they're accomplished, there's no return from that. So, like Capture the Flag, that's very focused and you go get this flag, you go get this flag, but if I take your flag you can always take another flag back and it's basically like a sporting game. The way the objective system works is that there's a huge amount of intensity and focus on supporting that location. And so the battles work out where when it's five on five, what happens is that those five people just begin to close in more around that centralized objective, to protect that area.
I will say that in your five-on-five match, or six-on-six, that's where your clan matches and your tournament-level matches would be; your players have to be.. GOOD. If there's an engineer who's not setting down vehicle turrets, [there's going to be a problem]. As far as a public server or the average guy getting into the game, that doesn't really know how to play, six-on-six is going to be a difficult challenge, but our balance for public games is twelve-on-twelve.
I'd like to thank Kevin Cloud and Paul Wedgwood for giving us such honest, complete, and enthusiastic answers to our questions. Enemy Territory: Quake Wars is set for release some time in 2007.