A Bizarre chat

Eurogamer: You were quite rushed at the end of PGR 3 and the second half of the game ended up unbalanced. But you seem much happier with this one.

Gareth Wilson: Yes, very much so. We just had a lot more freedom with the game. Starting with a finished product was just such a huge benefit to us. I mean, 360 is fairly straightforward to program on if you've got the final 360. Trying to build it on Apple Mac alpha kits is probably not the best thing. That's what really held us back on PGR3. It was Tron world 'til quite late on, to be honest. So this time around, we had the whole of PGR3 to work with, so if we wanted to try something out we just fired up PGR3 to play it, so that really helped with the prototype. So yeah, much happier.

Eurogamer: Alan [Mealor, lead artist] was saying that the best thing was simply being able to sit back and actually take stock of what you'd been doing.

Gareth Wilson: The biggest improvement in the environments has been the lighting, and a lot of the lighting effects - bar the weather effects, which are new - were in PGR3, but now you've got the completed game you can go back and have a look at how you did everything. You can see it in the PGR3 cities that are in PGR4 - it's the same geometry, but they just look so much better.

Eurogamer: With the handling model, Forza's come out recently and is obviously completely one end of the spectrum, simulation, and the other end is your arcade fantasy racing -

Gareth Wilson: Ridge Racer -

Eurogamer: Yep, and you're somewhere in the middle. How do you come up with a new car's characteristics? Because people do associate your cars with realism, but at the same time there's that element of fantasy.

Gareth Wilson: We make double the work for ourselves, because our physics engine is as complex as any physics engine out there - as complex as GTR's engine or GT's engine or Forza's engine. But we then put a car in with its basic performance characteristics, and it drives like a sim car - it's quite hard to handle and you crash a lot. The next stage is then we have a whole new stage of parameters you get in a Gotham car. So I'd probably say that our actual engine itself is more complicated than other engines. We start with a sim car, and then we make it fun to drive.

That's what we did with the bikes as well. To begin with, in the early development of the bikes, they were more like MotoGP bikes or Tourist Trophy bikes, and that's fine if you're doing Tourist Trophy but we're doing Gotham, so the huge amount of our work was trying to make the bikes handle like Gotham bikes - make them drift, make them endo, make them wheelie, without falling off all the time, make it fairly straightforward to get round corners.

Eurogamer: In terms of introducing the bikes and coming up with a handling model that was Gotham... I can't even get my head round how you would go about doing that!

Gareth Wilson: [Laughs] You have to put in a certain level of control, and with bikes it's predominantly to do with braking and driver balance. In MotoGP they will expose the balance of the rider and the front and the rear brake and then the player has to learn how to use those, whereas in our game we have one brake and the "drifto" button, which is kind of more rear-brake, and extra stuff on top of that.

All of that stuff, we have extra code over the top that makes sure the balance of the rider - that the rider actually moves in a particular way so you won't fall over when he does an endo, or rather it's hard to fall over. Don't get me wrong - you can - but it's not like MotoGP where it's quite easy to fall over, and then we can tune that, so some of the bikes are fairly straightforward so you don't fall off even if you're drifting. Later on, the faster bikes and the harder bikes, they are more simmy in a weird sort of way, so we can control the amount of "simmy-ness" and manipulate that.

Eurogamer: Is it more complicated to make that gradient of skill for a bike than it is for a car?

Gareth Wilson: I wouldn't say so - I'd say the hardest thing about the bikes was to get "the Gotham bike". Once you got the Gotham bike, you could then make variations on the Gotham bike, so you could have the drifty bike, you could have the one that's easier to stunt, you could have the heavy bike, you could have the one with good cornering, but it was getting to the prototype that was most difficulty.

And we had that for PGR. The car which sits in the middle of the vehicle set - something like, I dunno, the Ferrari 599 in PGR4 - which is quite drifty, quite stable, that's the benchmark, and the other cars come off that either way and there will be bookends, so at one end you'll have a super-grippy car like the Mazda RX-8 or a Lotus, and at the other you'll have something like the Camaro, which is an absolute beast and a nightmare to control, but some people love it.

Eurogamer: How do you test things and decide what to change, because you must get an enormous amount of feedback from playtesters?

Gareth Wilson: Yes. We have two ways that we do the playtesting and balancing. We have an internal team, so once the game's getting near to completion we basically hire a load of people - just people off the street [laughs]. People walking past - "hey, do you want to come play this game?"

Ben Ward: They're skilled professionals, come on.

Gareth Wilson: [Laughs] It's generally friends of friends, so it will generally be cousins that are looking for a university gap job between courses or whatever. So we get a lot of people that haven't played the game at all, and we get them to drive or play all of the games and then get them to fill out these great big forms and put it into a big database and analyse trends as opposed to actual data. You have to take a trend. You'll have one tester come up to you one day and say "that platinum is so hard, it's impossible" and then another one will go "that platinum is so easy", but as a trend you can look at 20 people playing it and see that on average that race had more retries than any other race, so it's probably too hard.

That's one thing we do, and then Microsoft are excellent at usability. They have a huge usability lab in Redmond, and you probably read -

Eurogamer: The Wired story about Halo, yep. [Feature story in the current issue.]

Gareth Wilson: With Bungie, yeah. They've got this great big usability lab and they hire people and they put them into these little pods and they have a build, and it's a slightly modified build, and they analyse them behind one-way mirrors and stuff like that. They come back with these quite scientific reports, which will be, again, trends, with good things - because hopefully there are good things - and bad things, which we tend to focus on. It'll say "19 out of 20 didn't understand how to do an endo", and then you'd look at reasons why and work out "oh, we never told them what the control is - duh".

Eurogamer: Is that a real example?

Gareth Wilson: Oh absolutely. There's tons of things. Messaging to the player is probably the hardest part of design, and you wouldn't believe - you can get something so straightforward, and you think "how could anyone get that wrong?" Turn markers are a perfect example. You start off with turn markers of a reasonable size and think "that's fine" and you put it into usability labs and people are going into corners and going [makes a car shape with his hand and shows it bashing into a barrier]. You can see the turn marker, but they see it going the wrong way - there's a very iterative design there where you then revise the turn markers and put them in again and so on.

And that's what we do - we have the internal guys, which is mainly for balancing, and then the external MS team for usability.

Ben Ward: We got forum guys as well, didn't we?

Gareth Wilson: Yeah, we hired a few guys directly off the forum. Like, all the platinums in the game have been checked by Fuzzy, who's our super, uber platinum player. So he plays all the platinums, and his job for the entire game was to play it at platinum.

Eurogamer: I'm quite jealous of that chap!

Gareth Wilson: But it's a good thing to do actually, because once, he was playing it, and he told us the platinums had got easy. And it turned out that someone had made a minor change to make the steering, the hands work clearly, and they had altered the clutch, and the change to the clutch had made every single vehicle in the entire game faster, but no one knew! So having that sort of thing is absolutely essential. A really good safeguard.

Ben Ward: And how long did the balance take this time? It was a long time, wasn't it?

Gareth Wilson: A loooong time. It's such a big game. There are 128 medals in Arcade alone, which is more than PGR3, and then the Career mode is 200 races at least, so it's bigger than PGR2 - it's enormous. It's too big. It's a monster! What have we done!

Eurogamer: You talked a little bit about how the process of designing for 360 has been different this time. I think a lot of people find that interesting - that idea of developers getting better with hardware. Could you talk more about how that actually manifests itself?

Gareth Wilson: With PGR3, we did a lot of - I guess you could call it brute-force development, because we were like "this machine is ten times more powerful at this particular thing, so to get the fidelity we will just do 120,000 polygons". So you do it and it looks great, but when you've released it, you look back and you go "did we really need that many polygons? Actually, couldn't we just do a shader that would replicate that look and halve the polygon count?" That means that you can get the same look in a more efficient way, and if you apply that across the whole of the game, you're saving a fraction of a frame here, there and everywhere.

Reflections on the bonnet are a perfect example of that. In PGR3, we took the whole of the world, rendered it onto the car, and that took a huge amount of work - the cube-mapping. This time around - I don't know if you've seen the bonnet-cam, but the reflections are absolutely perfect. The cars are rendered in it, which they weren't in PGR3; the crowd are rendered into the reflection, which they weren't in PGR3, and it's just because the programmer was looking at it and he went "you know what? I could just do it this way and it would be fine." Because of a particular way the architecture works - I won't get technical - but it stores the last frame in a memory buffer - that's how the graphic engine works - and if we just take that frame and put it on the car, then we don't have to do the obscene amount of work we were doing before. That frees up more CPU to do the weather effects and stuff.

Eurogamer: It's all about being able to take stock.

Gareth Wilson: It is, and there's all sorts of things like that all over the game. And there'll be more to do. Although you can exploit the 360 power quicker than the PS3, without a doubt, there's still more to go on a 360. The next game we do no doubt will look better than PGR4 does.

Eurogamer: Going back to Xbox, how's Live evolved from a developer perspective since you last started working on it?

Gareth Wilson: Live's been such a huge success, hasn't it? The thing that blew me away was the whole Achievements thing. When we were doing Achievements on PGR3, we didn't really know what they were going to be like, so we made quite vanilla Achievement lists, which was okay. But now we've all played 360 and we've all got Achievements, we really enjoyed doing Achievements this time around, so the Achievement list's a bit mental.

Eurogamer: I was going to ask what the "Puzzle" ones were.

Gareth Wilson: Yeah, there's the funny Puzzle ones, there's one about beating someone at Bizarre Creations [or someone who has already beaten them]. We had loads of fun with that Achievement set. The Live stuff - the actual fundamentals of Live hasn't hugely changed, but what has changed a lot is the LSP [Live Server Platform] support stuff, which is the PGR On Demand stuff and that sort of thing. The actual tools you get to create your own custom online stuff have gotten much, much better. So a developer can use Live as a wrapper, but then just do whatever they want.

Eurogamer: It's all gone a bit Web 2.0.

Gareth Wilson: Yep, it's all good. It's really enjoyable working on that stuff. Don't know what else to say really! [Laughs]

Eurogamer: Obviously you've got PGR On Demand instead of Gotham TV. What did you learn from Gotham TV? I spoke to some Burnout guys at E3 and they were saying that they were focusing more on Friends stuff now.

Gareth Wilson: That's a real key thing that we've learned from PGR3 in general. The Gotham TV stuff was popular but in spikes, and it was popular when it was something that meant something to people. When we had the Lamborghini tournament, we had - how many people online at once? It was a Lamborghini tournament to win a trip to Lamborghini. Literally thousands and thousands of people watched that tournament, but when there wasn't anything they knew about the intake was much lower, so when we did PGR On Demand we made it very Friends-focused. You can find out races your friends have done, you can see all their photos, and it's very easy to do that - you just go Friends, Photos and get a list. When you go into your garages, you can put your friends' pictures in your garage.

Time Attack as well - if I've done a cool lap, that I'm pleased with, I can upload it and send a message to my mate and he can download it. All of the Time Attack is Friends-based. Every leaderboard shows your Friends leaderboard first, and if you want to see the top people or your position, then you can totally do that, but we focus on the Friends stuff. Because it's not particularly compelling to be 120,000th in the world, so when we display that number, we do show it, but we also show it as a percentage of the leaderboard - it would say "you're in the top 10 percent" or the top 20 percent, or the top 30 percent, because that has more meaning to someone. But yeah, the Friends stuff was really an eye-opener for us, so I can see what the Burnout chaps are on about.

Eurogamer: It's been a while since Test Drive Unlimited, but it's still the only game to take that approach of a pure racing game in an open world on that scale. Why didn't you take that approach?

Gareth Wilson: I'll tell you why - it's because we looked at free-roam stuff, and free-roam makes a great driving game but it doesn't make a good racing game. Test Drive's dynamics wasn't that bad, but because it's a free-roam racing area, you can't put the love and care into the circuits that you can when it's a pure circuit-based racing game. We have teams of designers working, honing every single corner, putting the barriers up, putting the turn-markers up - that stuff's essential to making a circuit-based racing game.

Eurogamer: Is that why you haven't got the Route Creator in this time?

Gareth Wilson: Yes, because the Route Creator ultimately didn't produce very good routes, because we'd already picked the best routes and put them in the game. So that was one of the reasons - the other was that it just wasn't very popular, people just didn't use it.

But you know, we have the tech - we could have basically just built New York, and that would have made a great driving game. Test Drive Unlimited is a great driving game - when you drive up to a junction you almost feel like stopping and just letting the cars go past and then driving along - but not as a racing game. If you miss the checkpoints you go flying off and before you know it you're 200 metres the wrong way and all those sort of things mean it's just not a very nice experience when it comes to actually racing.

I never found myself in Test Drive repeatedly playing the same challenge to shade a second off my lap-time, whereas in Gotham certainly that's the key in the Arcade mode. It's that replayability and learning the course. And that's why as a company it's very unlikely that we'd do a free-roam racing game. We might do a free-roam driving game - for a Driver or GTA, free-roam is king, but they're not racing games, so I think that's why Test Drive came a bit unstuck.

Also I think it was too early as an online racing game. I think if it had been released now it would be much more popular. Ahead of its time.

Eurogamer: Another Live-y sort of question - you did quite a lot of downloadable content for PGR2 and 3, and I assume you're going to do more of that -

Gareth Wilson: Yes, absolutely we are -

Eurogamer: But when are you going to start doing that and what sort of things do you have in mind?

Gareth Wilson: We're working on it right now, I don't mind telling you. We'll do DLC - I'm working on it right now. I can't tell you any more [laughs]. It'll be cars and bikes probably.

DLC's got two really interesting things to it. We do make a certain amount of money from it, but that's not really why we do it. We do it to keep the longevity of the game up. If you look at the sales for PGR3, it's been a really, really slow-burner. It's just gone on selling and selling, and it's still selling now. And I think some of that is down to the DLC, because people comment, and they enjoy the DLC that comes up, and they keep hold of the game a bit longer, and we get articles on your site saying that new DLC's out. Particularly stuff like the free stuff we did - the marketing we did with that just kept it going. We get stats for Live usage, and every time there was a DLC pack it went up.

Eurogamer: Finally, do you expect to make a PGR5?

Gareth Wilson: Christ! I've not even finished PGR4 yet, Tom! Come oooon. I can't tell you that!

Eurogamer: I left it there so openly! "Do you expect"!

Gareth Wilson: [Puts on silly voice] Of course we will be working on another racing title [laughs].