20 comments

» MEDIA (2) A facial render from Crytek's upcoming FPS game, Crysis Beata Beatrix, by Dante Gabriel Rossetti.

Gamers are fond of comparing screenshots, so I'd like you to look at two images. The first is a screenshot from Crysis, a FPS currently in production by FarCry developer Crytek. This image is an impressive demonstration of the level of facial detail that Crytek's new engine will be capable of:
http://upload.wikimedia.org/wikipedia/en/5/57/Crysisface3d.jpg

This next portrait might not be what you expect. It is a canvas by Pre-Raphaelite artist Dante Gabriel Rossetti, entitled Beata Beatrix:
http://www.ibiblio.org/wm/paint/auth/rossetti/beatrix.jpg

I understand that a side-by-side comparison is complicated by the fact that the two portraits were created for very different purposes: Crytek is looking to show off the capabilities of their newest technology, which must render many frames of continuous video in real-time, while Rossetti only had to create a single-frame representation of one moment of rapture. Still, I think it would instructive to consider a few questions as we look at this portraits.

Of the two works of art - if you'll permit me the phrase - which looks more realistic? The answer, clearly, is Crysis. Crytek is able to achieve a granular level of detail that was simply beyond Rossetti. Crytek is working on a modern canvas of high-resolution pixels that are each a third of a millimeter wide, while Rossetti is unable to accomplish any detail finer than the width of his paintbrushes. The meticulous, computer-generated attention to minutiae in Crytek's portrait is evident in every perfectly shadowed crease of the man's skin, every wrinkle, hair, and pore. Rossetti, on the other hand, has to make do with broad brush strokes and swaths of color.

Of the two works of art – there's that “art” word again – which is more aesthetically appealing? The answer, equally clearly, is Beata Beatrix. Obviously, Rossetti has a number of advantages here. Unlike Crytek, he's only interested in creating a single frame, so he can attend to peripheral details like composition and arrange props around his subject to create specific effects. The subject is back lit and a shimmering bloom effect, suggestive of a halo, is visible around her hair. Prominent lines in the portrait – the gnomon of the sundial, the stem of the rose, the curve of her arms and curling of her fingers – all lead towards Beatrix's head, further emphasizing it. More importantly, Rossetti knows how to work around the limitations of his tools. By using swaths of a lighter tone on Beatrix's face, he gives her a healthy glow. To be fair, Crytek's aim in their screenshot wasn't to create an aesthetic image; it was to render a very realistic-looking face. At this, they succeeded.

Now our final question. Of the two, which portrait is most human? In other words, which figure arouses the most empathy in us and which artist best maintains the illusion that we are looking at an actual, breathing person? To me the answer to this question is just as clear: Beata Beatrix, again, without a question. Crytek has photorealism in their corner, but Rossetti has style in his. And when it comes to creating a figure that an audience can identify with and respond to, style will win every time.

I think these distinctions we are making – between realism, aesthetics, and style – are important, because far too often they are conflated and used interchangeably in the gaming press. When a game is technologically impressive, this is described as “good” graphics when in fact the graphics are simply realistic. Many developers, meanwhile, sink millions of dollars into a robust, state-of-the-art graphics engine and pay hundreds of artists to model, texture and bump-map every square inch of their game-world. But ultimately, what makes a game visually memorable is not realistic graphics; by next winter a new game will have already scaled even greater heights of photorealism. The true test of a game's graphics is its sense of aesthetics, its sense of style. A strong aesthetic is why World of Warcraft's characters have become iconic and recognizable in any context while Everquest 2's generic avatars and inhabitants would be difficult to distinguish from those in the many fantasy games released every year. Even today, gamers reminisce about the iconic elegance of the very cartoon-style units in Heroes of Might and Magic 2, while the ever-increasing detail and complexity of the successive units in Heroes 3, 4, and 5 have elicited no such passion. Similarly, I can still fondly recall of the colorful, decidedly low-tech vector graphics of Darwinia years after I've played it while the other RTS games I tried that year all seem to sort of bleed together visually in my mind.

The problem with a game whose only goal visually is to resemble reality as closely as possible is that every other game is already copying from the same source. Even if, one day, every single game manages to perfectly recreate reality on our computer screens, that would represent the horrifying prospect of every single game looking exactly the same as the next. This is already happening today. There's nothing stylistically to distinguish the look of CounterStrike:Source from Rainbow Six: Vegas. They're both carbon copies of reality, and thus carbon copies of each other.

To truly stand out and resonate visually with an audience, to create a world that gamers will enjoy inhabiting, will identify with and even believe in, the game absolutely must have an unique perspective, an idiosyncratic aesthetic all its own. It must not be content simply with representing the same reality that countless other games will be similarly copying; it must also interpret reality visually. In a word, it must have style.

The problem of excessive faith in photorealism is becoming an increasingly urgent one for the industry, not only because of ever-escalating development costs, but because gaming graphics are quickly encountering a phenomenon that is known in the robotics field as the “uncanny valley.” The uncanny valley is the observation that when as a robot – or any other facsimile of a human being – becomes increasingly human-like and realistic, our response to it will become increasingly positive until the level of realism reaches a point where we grow unnerved and repulsed. When something grows too realistic, we start paying attention to the small details that are “off” and the figure begins to look grotesque. For example, consider a classical Greek statue and a figure in a wax museum. Despite the fact that the wax figure is a far closer imitation of an actual human being, the Greek statue is majestic epitome of the ideal human form while the wax figure is a glassy-eyed, eerily skinned aberration that frequently appears in horror movies. Computer graphics have already started to descend down this uncanny valley. There was something unsettling about the pallid faces of the characters in Doom 3 even before they became zombies. The flesh looked extraordinarily real, but not quite actually alive. Fortunately, this sensation complemented that game's particular mood and theme, but similar effects are appearing in plenty of other games – the realistic-yet-hideous faces of the characters in Oblivion, for example, or the strangely vacant, glassy eyes of Half-Life 2's characters.

I don't mean to suggest that building powerful graphics engines are futile endeavors or that advanced technology automatically represents some kind of aesthetic deficit. It may just seem that way sometimes because developers who are working within the constraints of dated graphics engine are more likely to use ambitious art design in order to compensate. The games that do manage to marry advanced graphics and a compelling aesthetic styles become truly distinctive visual experiences. These are games that use bleeding-edge graphics to create stylish game worlds – the Unreal Tournament series will always be known for its over-the-top hypersexualized characters, and I think anyone who played Homeworld when it was first released will remember what a revelation those graphics were, as much because of their gorgeous presentation as for any technical merits. It would be nice, however, for game graphics to explore artistic styles beyond the strictly photorealistic more often. After all, few people play games to be reminded of reality. You might as well depict a visual universe with some style.