I'm sure it's not a piece of cake, but what would you expect the new sound guy to say? "Oh yeah mate, I can churn out 5 or 6 cars a day, it's relatively straightforward" Of course he's going to say it's hard, makes the fans appreciate it more and makes his boss happy too.
Actually, he pretty much does say that!
The recording and making of loops are the easy parts. This particular sound team will have a quick mosey around the car, figure out where to put microphones, and then choose which ones to put there. Then they spend an hour or two recording what they need. Within a few hours of getting the recordings back to the studio, they have samples.
No, what's hard is recreating the
continuum of sounds of the real car from what are effectively snapshots. It's like trying to make a movie out of a limited range of stills (except that movies are linear, and games shouldn't be IMO). Add to that the technical challenge of making the samples sound right at all pitches (it's really not just a matter of changing the rate they play back at, at least it shouldn't be if you want them to sound good), and of not destructively interfering and of not popping etc. This is all pretty hardcore sample theory / filter design stuff, unless you want to do it trial and error.
Then there's the fact that instead of having a naturally varying timbre, you have a fixed, "timbre-stamped" sound all the way through one portion of the rev range until the next sample is faded in - that's not how real engine sounds work, they are infinitely variable (implies infinite samples!), but it's how all engine sounds are reproduced in games (except LFS and all those early synths.) So there's a bit of faffing trying to get the variation to sound natural.
Then once you've got that working OK, you need to place it in an environment and with hundreds of other sounds and get it all to mix nicely and run nicely and not kill anyone's eardrums. That's really, probably, the most difficult part. Imagine a problem like the global illumination of a game scene, only you need to consider each colour of light separately, because each one behaves differently.
Most people cheat, and use the spatial character inherent in the recordings (of whatever environment the car was in) and apply that to the game, so the car is perpetually in the same acoustic environment no matter what it does in the game. We can do better than that. Indeed, I believe pCARS is trying something ambitious with environment modeling, although its samples are still heavily coloured (which will interfere with the modeling), and it won't use separate intake and exhaust sounds (but it does use dedicated interior samples, which is a good idea as the interior doesn't change anything like as much as the exterior).
Point is, other games have far better sounds. Other games with far, far smaller budgets. I believe what's holding PD back is the 1200 cars. They could make the 200 new cars with premium sound quite easily I'm sure, but they won't, because it'll make all the other cars sound like crap and they won't have that. I think it's an all or nothing deal.
The car count is a major obstacle, especially when you add in the modified sounds required as well. But they've been working on the new system for years. I was suspicious of the apparent lack of attention the sound samples had received since GT4; why were they still using the same samples, and why did new cars to the series have recycled sounds from other cars? Obviously, the classical conclusion, ad ignorantiam, is:
lazy. Looking at the rest of the sound approach (the only relativistic sound engine in the business!
), and the way they were tackling other systems, it just didn't make sense to me.
One of the great benefits of a simulation approach is that you don't need to have separate samples for modifications, as such, because you can just feed the relevant changes into the model, and the right sound should just drop out. These modifications can be much more fine grained and more diverse than just three extra options of exhaust. I've been having great fun over the last couple of years "inventing" new sounds for certain engine types, and just playing around trying to match, texturally, certain cars' sounds. I can imagine a whole "community" settling around such a varied customisation feature in a game, like there is for photomode and livery editors etc.
Anyway, just because other games sound better than GT5 at the moment, doesn't mean there's nothing more to be had. Games could sound a lot better, and advocating old techniques because "they work" is not going to get us anywhere new.
I promise to cook up a few lo-fi (I mean it, you'll have to polish your eardrums to hear what's going on) demos to show you all the kinds of inherent flexibility a totally new approach could yield, but only once I've rebuilt my computer in a month or two's time...
(By then, PD may have shown us what they've been up to, and I can crash into an über-depression as I realise how far behind the curve I am!)
