GDC '08: Crytek revisits Crysis

CEO Cevat Yerli joins lead designer Sten Hubler and producer Bernd Diemer to dissect highly decorated FPS; emphasizes need to stay true to the original vision but retain flexibility.


GameSpot may get a commission from retail offers.

SAN FRANCISCO--The 2006 Electronic Entertainment Expo offered gamers a number of reasons to be excited about the coming year. However, trumping much of the noise coming out of that show was a solitary tech demo from Far Cry developer Crytek for its upcoming EA-supported shooter, Crysis. The Frankfurt, Germany-based developer had already established itself as a technological powerhouse with Far Cry's stunning CryEngine, but crowds were not prepared for the lush tropical environments and jaw-dropping breakable foliage on display.

As impressive as the showing was, it came very close to not happening. That and other tidbits were presented to a bleary-eyed crowd still reeling from Thursday's night's festivities in a postmortem session for the game kicking off the final day of the 2008 Game Developers Conference. Helming the session was none other than Crytek CEO Cevat Yerli, who was joined by Crysis lead designer Sten Hubler and producer Bernd Diemer.

As is how postmortems go, the Crytek trio first laid out what had been learned from their previous efforts, in this case 2004's Far Cry. (Far Cry's subsequent sequels were handled by Ubisoft's Montreal studio.) First and foremost, the team was proud of Far Cry's peerless visuals and the technology powering the game. Yerli felt these features were complemented by the game's sandbox design and responsive human artificial intelligence.

However, despite its glowing critical reception, Far Cry had its problems, with Yerli specifically calling out the game's "B movie story," bad jokes, and lack of a quicksave option. Yerli blamed problems with the story on the fact that it was shoehorned in near the end of the game's development, while the lack of a quicksave feature was purely a design decision, calling the move "egotistical." Yerli also noted that the game's difficulty wasn't balanced properly and that the artificial intelligence for the nonhuman wasn't up to standards, due mostly to the fact that the enemy types changed drastically throughout the course of the game's development, from dinosaurs to aliens and ultimately mutants.

The goal for Crysis, then, became to turn all of the weaknesses from Far Cry to strengths, while at the same time improving upon everything the team did well. For Yerli, this all began with his original vision, which was to take the lush, warm tropical paradise of Far Cry and encase it in ice. "The Vision" also included maximizing players' ability to express themselves through the game by offering a variety of play styles. In so doing, he wanted players to outsmart the game by forcing them to think before they shoot. Lastly, Yerli's vision called for realistic sci-fi, namely an alien invasion that is as believable as possible.

In 2005, the process of actually making Crysis began to coalesce. One of the team's top priorities at the time was to "create the best-looking game ever," and to do so, Crytek signed on Venice, California-based Blur Studios to mock up a prerendered video using assets from the game to set a benchmark to aspire to. "The goal was to get step-by-step closer to these visual benchmarks," Yerli said.

One of the first steps in accomplishing this goal was to take a "business trip" to Tahiti, Yerli noted, with Hubler chiming in with, "We're coming from Germany, so we really don't know what a real tropical environment looks like." Real research was key, the team noted, and it was important to get a handle on how standing in a tropical forest and looking through the trees actually feels. "To make the best-looking game, we had to capture a believable world," Yerli continued.

Then came E3 2006, and with it a proving grounds for the work that had thus far been channeled into the game. Yerli revealed that the most striking scene shown off of the game--when a soldier mows down a clump of trees in the forest using a minigun--only made it into the demo due to a little luck with the timing. It had taken nearly six months to get only one tree to break apart at an acceptable 300 frames per second, and an equal amount of time to be able to reproduce the effect on any object of choice.

Yerli felt breakable vegetation was one key aspect coming out of the show that went on to define the game. However, it presented severe problems with the game's AI, namely the fact that if a player is able to completely obliterate an object in the game, then the enemy looks a bit silly hiding behind the now nonexistent obstacle. Using that as a starting point, the team underwent an extensive AI overhaul, addressing a variety of issues ranging from animation to open-ended gameplay.

The challenge, Yerli noted, was achieving "human" AI and not "perfect" AI. "Perfect AI, anyone can do. Human is a more difficult challenge, because you have to replicate humanity," he said. To illustrate how Crytek overcame this obstacle, he showed a video of North Korean soldiers pouring over a stone fence, with most landing on their feet, but a few tripping and falling.

The other key concept of Crysis, which wasn't even fully realized at the time of the E3 showing, was the game's distinguishing nanosuit. The nanosuit was born out of Yerli's realization that even though the core shooting was fun and the company had hit its visual bar, Crysis "had nothing to bring to the table yet in a meaningful way." Initially, the nanosuit was to gain power over the course of the game, with the fully featured suit playable only available in the last five minutes of the game. That was a problem, of course, and necessitated a reimaging of the feature.

For the nanosuit, Yerli insisted that "customization be king." By that, he meant that the suit should allow players to a wide variety of choices and ways to express themselves, whether that be through run-and-gun, "Rambo-style" gameplay, more tactical stealth, or a combination of the two. Outsmarting enemies also played heavily into the design of the suit, and the team explicitly wanted to avoid the FPS staple of "run in and shoot them before they shoot you" gameplay. They also wanted players to be able to experiment with strategies, make mistakes, retreat, and then try again.

The suit went through a significant iterative process, where the team returned time and time again to play testers, who invariably reacted in "unexpected ways"--which, as the team noted, is the nice way of saying the testers were idiots. The suit's visual interface underwent some of the most significant overhauling, with the few iterations involving catering heavily to the micromanaging crowd.

These complex systems--which, according to Diemer, were more likely to result in "death by chicken" than anything else--were eventually streamlined into binary suit powers (that is, strength or speed) that could be accessed quickly and intuitively. As the game had been in development for quite a while by the time the nanosuit began to take shape, it caused a good deal of friction between its proponents and the level designers, who essentially had a good deal of work negated. While painful, Hubler noted that it aided the team in realizing freedom in a sandbox level design.

Concluding the postmortem, Yerli noted the importance of keeping the vision and sticking to the quality bars set in the beginning. Yerli also noted that the vision should only act as a seed for the development team--"a vision is developed by a team; it's seeded by a visionary." Closing, he stressed that visions that are forced will inevitably fail, and that it is necessary for a team to buy into that vision to maintain passion and inspiration.

Got a news tip or want to contact us directly? Email

Join the conversation
There are 73 comments about this story