Building better worlds

Building better worlds


Building better worlds

In the unlikely event of today's consoles still driving the game industry in 2015, their reign will have endured for almost a decade – an unprecedented period of suspended animation for dominant videogame hardware. With the PC marching stubbornly on and handheld devices fast catching up, it's the licensed engine that's evolved the most. Games light enough to run on Xbox 360, flexible enough to port promptly to PS3, scalable enough to satisfy PC, and adaptable enough to spawn phone and tablet tie-ins cannot, by and large, afford to solve that conundrum from scratch. Hence the runaway success of Unreal, the bold claims of CryEngine, and the quiet, poised growth of Unity.

But the first two, in particular, are getting restless. The last few months have seen chest-beating demos in Epic's Samaritan and Crytek's CryEngine For Cinema. The future is calling, and they're keen to answer.

"2015 is a long time in terms of hardware advances," notes Carl Jones, Crytek's director of global business development. "The sheer processing power that we're going to have at our fingertips will be immense, and the parallel power of the GPU opens up new rendering techniques to us that we haven't really been able to use in realtime before. By using a combination of them, you're going to get much more realism in areas where we don't have it holistically yet.

"If you look at LA Noire you see amazing character faces because they've gone for a bespoke solution for those faces. But the power isn't there to match that quality in the environments and such. In the future you'll be able to do the same quality for everything, so you won't get that 'texture shock' of some things looking high quality and others not. Anything you see now in movies, you'll be able to see in realtime by 2015."

A developer's eye view of CryEngine 3

For Crytek, in particular, there are practical connotations to the hyperbole. Its aforementioned Cinema demo speaks of nothing less than complete game and movie industry convergence, where the products don't just look alike but are built together, sharing resources, energies and, importantly, release dates. "The creation of linear media is a very slow and frustrating process for creatives," explains Jones. "We don't get the chance to see the results quickly enough. We've all seen movies where things just don't look quite right, and normally that's not a bad creative decision but the results coming back so late that the creators have to live with it.

"Even in the best movies and the best games, the disconnect results in bad crossovers. With an approach where you're using the same technology, it means that you can do these things in parallel. Both sides understand the technology and can have more of a say in the creative process. What we should see is better crossovers, and less rush one way of the other. Generally, one side has to rush to meet the schedule of the other, and that always results in less quality.

But beware, he says, the urge to make "gaming's version of Avatar."

"Someone joked with me this week that we should ask for infinite RAM in consoles because that would solve all our problems. But the issue with having more power to play with is that it requires more content, so the removal of that bottleneck means you need to be more efficient in how you create it. Because with all the power in the world, no one's going to be wanting to make games with 700 people working on them, spending years and years and years to make the number of assets you can take advantage of. So yes, we'll see a major difference in the quality of the graphics on show, but it's really going to depend on the tools being in place that make the development efficient."

Silicon Knights' Too Human

What happens if the equilibrium isn't there is no great secret. Remember Frame City Killer, Namco's open world epic for Xbox 360? Probably not, because the game was never finished. Managed to forget Too Human yet? Silicon Knights boss Denis Dyack hasn't, because after a torrid development history the game jumped to a reportedly unfinished and undocumented Unreal Engine 3, stumbling its way through ridicule, infamy, and finally into court. As much as enabling new and better things, the job of tomorrow's engines is to stop this kind of thing happening again.

"We've always had this philosophy that anyone using a game engine should be able to see the changes they're making in realtime," says Jones, "and we have this realtime pipeline that we continue to enhance. With CryEngine 3 we introduced deferred lighting, and could add as many lights as people wanted to a scene. But very quickly the work of adding those lights became a pain, so we made sure we had ways of auto-generating them.

"For the artist, specifically, the challenge is allowing them to use the tools they're familiar with and making the pipeline into the game as realtime as possible. We previewed it in our Cinema video where you have a live link-up to Maya, etc, and we're pushing that further. So our artists will be able to generate anything they can really imagine. We're confident that the hardware they'll be building it for can support it, and then it's just about us making sure there's no bottlenecks between the two."

What that hardware will be, exactly, is hard to know for sure. Jones foresees – and welcomes – a future where tablets, phones and consoles become increasingly PC-like: CPU next to GPU, with increasingly similar architectures. Things are, he reminds us, going that way already. "The details are different, the operating systems are different, but fundamentally that's what we're looking at. And as we potentially move towards servers and clouds, it may be that the device itself becomes less relevant, and ultimately you'll be running on a low-end PC."

But are tomorrow's engines anticipating such a future? Not to any great degree, it seems, because the infrastructure just isn't happening. The bandwidth, yes. But the latency? "It's not going to be solved any time soon," says Jones, "because unfortunately it's only really an issue for gaming. The latency levels that are required for 99 per cent of consumer demand are delivered by what we currently have. You don't need better latency for movie streaming, communications… But gaming desperately needs it, and I guess we're not a big enough force to make the companies that control it improve it. So yes, in principle we're all looking for a time when the necessity for hard storage isn't there any more, but the ability to do it online isn't there, either."