GameSpot AU: When GameSpot spoke to you last year, you mentioned that Crytek had done three years of research and development into consoles before work on Crysis 2 started. What sorts of things were you looking at?

Cevat Yerli: If I wanted to make a linear experience in New York, where I break certain things at a certain time, then that's easy. But if I wanted it to be systemic experience, where it's a sandbox and the game reacts and ties things together like physics, AI, graphics, and animations in a systemic way so they can interact with one another in an unpredictable fashion in the game, then it puts a much bigger burden on technical requirements. That can cause a huge demand on processing and memory, especially with new kinds of geometry formations. If I break a wall, it creates new geometry. If I deform a car, it creates unique deformation. All those things occupy more memory as we go.

If you're on a console with limited memory, then you have to deal with optimisations that allow you to make these kinds of games. On a PC, what do you do if you need more memory? Well, you just ask for more memory. Memory is not an issue on a PC. For me, I wanted to translate that type of interactivity in a live world with a sandbox nature before I could commit to a game on the console. Our console research has shown more and more progress with systems that can interact with each other. Those systems were getting more and more optimised, and we even optimised the PC version, so ultimately we were able to pull it off.

If I wanted to create an explosion in a game in New York where it's just linear, we could do it--every game engine can. It's just a matter of quality difference. But in our case, it's not just a quality bar, but the unpredictable nature of our sandbox gameplay. You may actually cause 10, or 15, or 20 of those explosions, or none at all, and the technology has to cope with it.