FREE counter and Web statistics from sitetracker.com
collision detection
content | discontent
send me yours
May 17, 2006
PS3 games: Plunging deeper into the "Uncanny Valley"










I've written several times about the "Uncanny Valley" theory -- the idea that as computer-generated depictions of humans become more and more photorealistic, they look creepier, more ghastly, and more cadaverlike. The concept is simple: When we look at a cartoon-like drawing of a person, like Charlie Brown, our brains fill in the missing information, and the cartoon seems warm, cute, and lifelike. But when an animated version of a human becomes incredibly close to being real, we start focusing instead on the tiny details that aren't right: The slack skin, the not-quite-dewy-enough eyes, the stiff body movements. Paradoxically, the more realistic the human becomes -- the worse they look. Sure, enhanced graphics look terrific when lavished on static things, like scenery or smoke or bullets. But the human face? Our video-game graphics aren't up to it -- and, if you believe the Valley theory, may never be.

The Uncanny Valley effect has become painfully, itchingly obvious in today's video games. Whenever a game comes out with cartoonish and stylized humans -- like the anime-style Final Fantasy series -- they look wonderful and lively. But whenever the game designer gets obsessed with being "cinematic" and "superealistic" and producing "cutting edge graphics", woof woof, meow meow, the results are just unwatchable -- as with, say, the "lifelike" characters in Half Life 2 that cavort about like a corpsetastic army of zombies.

After hurling themselves against these shoals and crashing again and again and again and again and again, wouldn't you imagine that game designers would learn their lesson?

But no. The advent of the Xbox 360, the Nintendo Wii, and the PS3 have all got them whipped into a fresh new lather about creating "photorealistic" humans. Thus it was that I came across the promotional trailer for Heavy Rain, the sequel to the terrific game Indigo Prophecy that's slated for the PS3. I clicked on it, wondering ... hmmm, are the PS3's graphics finally so good that the designers have climbed out of the Valley?

Nope. They've trudged in ever deeper. Check out this clip, in which a young girl does a "casting call" and delivers a long monologue into the camera. Prepare to scream and scream again. Seriously: It's goosebump-inducingly bad. Her lips attempt to smile, and pull back in some unholy rictus of a grimace; her skin slides like dead sheets of atrophied flesh along the surface of her bone structure; and her eyes -- my god! Her eyes! It's like looking over the edge of the flat earth into an endless infinite howling darkness, unto which an anvil could be tossed and fall for forty days and forty nights and not yet reach the inky awful depths of her soul.

This wouldn't be so bad if the designers were actively trying to create some eldritch, sephulchral nightmarescape straight out of Goya's Black Paintings. But no ... they're trying to create a spunky, cute, realistic girl.

God almighty, these people must be stopped. This stuff is hideous beyond description, and I describe things for a living.


(Thanks to Jonn Wood for this one!)

Posted by Clive Thompson at May 17, 2006 02:14 PM

Trackback Pings

TrackBack URL for this entry: http://www.collisiondetection.net/mt3/mt-tb.cgi/1487

Comments

I don't know, Clive. Ever since I read about the Uncanny Valley here, it's been one of my go-to cocktail party gaming vignettes.

But I have to disagree with you on this clip--I think it's pretty good. They nail the body motion, the eyes, and the cinematographic techniques that trick you into thinking you're watching live action (focus, light, etc.). Right before the lightning flash, they even have the skin right for a while.

OK, the mouth is horrible and scary ("rictus" is a perfect descriptor). But I think they've got a lot of components closer than ever. Still corpsey-creepy, but this clip convinces me that in the next couple of years we'll see a game that gets it right.

Posted by: ScottS [TypeKey Profile Page] at May 17, 2006 5:05 PM

Hmmmm. One can certainly have different views of the success of this stuff: At some point, it's a purely aesthetic viewpoint, and you may well find stuff pleasing that I find awful!

But personally I think in-game humans have been so zombielike for so long that we have adjusted our standards downwards. Something that seems incrementally, slightly better is hailed as a breakthrough. But to any casual observer whose aesthetics haven't been slowly and punishingly ground to dust by 10 years of having the game industry produce wretchedly awful human-face in-game animation, while proclaiming it "groundbreaking" and "photorealistic" and "lifelike" ... this stuff seems just like what it is, which is to say, really, really, really sub-par animation. I mean, this crap wouldn't be tolerated on the most poorly-fudned of low-budget Saturday morning cartoons.

Posted by: Clive [TypeKey Profile Page] at May 17, 2006 5:12 PM

That clip would be great if at the end, a tentacle came out of he mouth, reached out through the computer screen and slapped us all around for a bit, Cthulu style.

Posted by: MoXmas [TypeKey Profile Page] at May 17, 2006 5:13 PM

I would just like to point out that the Lovecraftian next-to-last paragraph made me giggle. Almost Tycho-like in it's intensity.

Is it strange that when I finished watching this, the first thing I thought was "Clive Thompson has got to see this"?

Posted by: Jonn [TypeKey Profile Page] at May 17, 2006 6:26 PM

To me this is a perfect example of the Uncanny Valley. And I think the 360/PS3 are going to be going down to the river, so to speak, quite a bit over the next few years.

The Wii? Not so much. Because I think the Wii is made for more stylized games that it won't be as much of a problem. Just my opinion however.

Posted by: Karmakin [TypeKey Profile Page] at May 18, 2006 11:44 AM

Jonn, Moxmas, heh.

Karmakin ... I'm pretty sure someone could use the Wii's processing power to create a perfectly ghastly Valley-plunging pseudohuman. But you're right -- Nintendo tends to aim towards more cartoonish, and thus more emotionally convincing, characters.

Posted by: Clive [TypeKey Profile Page] at May 18, 2006 1:39 PM

Some communities will favor persons with faces and motion patterns similar to these generated persons. "Wow, cool, he looks like that game character". Some people will be popular because they are easy to model in certain development systems.

Posted by: gurka3 [TypeKey Profile Page] at May 18, 2006 4:15 PM

The recent CGI film, Final Fantasy VII: Advent Children, impressed me in that it largely seemed to avoid the uncanny valley effect. While the characters were obviously slanted toward extreme realism, they were also stylized. The stylization was relatively subtle and yet made such a large difference.

I do think that in video and computer game terms, everyone has become far too used to terrible plastic puppet technology; it probably is true that when something slightly less terrible comes along, it blows people's minds as "the most realistic thing ever!"

Posted by: Kaijima [TypeKey Profile Page] at May 18, 2006 4:31 PM

Most stars today receive a big doses of Botox, Photoshop, and plastic surgery, which makes them seem uncanny as well. We are used to a weird aesthetic, and it gets more evident as time goes by.

I agree that this generation of consoles will probably have the worst looking "humans", but by the time it gets too strange a seamless-smooth transition will have occurred in our psyche.

Posted by: lorbus [TypeKey Profile Page] at May 19, 2006 7:22 AM

Maybe it's just me, but what bothered me most about the clip - much more than the eyes or the skin - was the designers' inability to syncronize mouth movements with speech. This is also an area in which we have mentally enabled low quality craft and provided the link ourselves. But in this clip, it seems, the problem is not that it's so close that we notice the tiny difference, but that the differences are not tiny. Everything else is so close that the speech stands out.

Posted by: GKoutnik [TypeKey Profile Page] at May 19, 2006 7:28 AM

MMhbwwuuahhaaahhaaahhAAaaa!

You have seen nothing... NOTHING!!!

http://www.dearannemovie.com/

Apart from being creepy this is over the limit of "Kitsch" that can be tollerated by the time-space continuum before the universe collapses in a cosmic burp...

And excuse me for the clumsy "me poor italian" attempt in Clive prose...

Posted by: Mario [TypeKey Profile Page] at May 19, 2006 10:20 AM

I agree that attempts to make video game humans more photorealistic usually end up producing more 'creepy' results than abstracted characters - essentially cartoons. But I disagree with the overall theory that extends this thinking proportionally - the more realistic, the creepier. It seems that there are merely some limitations in budget, time, and rendering power within the video game world that limit these human characters. I have to assume that there will eventually be some breakthrough, as there practically has been in the movie industry. Look at the quality of digital doubles. Look at King Kong! Not a human, I know, but he has all the qualities of real life expressions.

Adam

Posted by: ajberni [TypeKey Profile Page] at May 19, 2006 12:43 PM

What Clive is saying is actually the eternal question between simulation vs. reality. We used to ask it about theater vs. movie...

At the same time, we have to realize how much video game are new. The technology is not yet stabilized. We are not yet in its golden age.

Posted by: Nicolas Toper [TypeKey Profile Page] at May 19, 2006 5:21 PM

Trackback, from Grand Text Auto: Dry Water in the Uncanny Valley, and more

Posted by: andrewstern [TypeKey Profile Page] at May 19, 2006 5:25 PM

"Uncanny Valley"? Clive, you sould take a look at the book "Understanding Comics" by Scott McCloud (a great reading, after all).

According to Scott, cartoon-like characters are warm because people who's reading the comic gets identified by the cartoon. Cartoons are good for getting deep inside a story.

That's what makes Peanuts, Garfield, Tintin, and Dilbert readers coming back for more.

BTW - Maybe you may find interesting "The future of comics", also by Scott McCloud, for future stories ;-).

Posted by: IƱigo Gonzalez [TypeKey Profile Page] at May 24, 2006 11:34 AM

I can see where you're coming from Clive. Indeed, this trend toward more and more realistic animation is falling a bit short of the mark. However, I'm not so sure I agree that we'll never get "there". Things have come a long, long way toward genuine animareality. Nobody complained that Mario didn't look "quite right", but as we get closer to the real deal, it's true the differences stand out. Give 'em all some time and we may just be surprised.

Posted by: digital_blue [TypeKey Profile Page] at May 24, 2006 1:27 PM

Excellent conversation here, folks!

Inigo, indeed, I've read all of McCloud's books, and in fact referred to him specifically in my original column on this subject two years ago! He's amazing.

It is possible that we'll crawl out the other side of the Valley. But for what it's worth, it's by no means certain that we'll have better games for it! Games aren't created by graphics; they're created by rules that constrain our behavior in interesting ways while guiding us towards a teasingly difficult task ... which is why one can play a game with one's eyes closed, looking at nothing (as with word games you can play with someone else.) The big problem posed by the effort by game publisher to "climb out" of the Valley is that they're putting all their energies into the graphics battle, and sparing virtually no innovation on the gameplay.

Posted by: Clive [TypeKey Profile Page] at May 24, 2006 3:45 PM

I think one of the problems with the design is that the human characters are made to be too "pretty"; the skin is too flawless, the eye orbits are too round, the lips are too shaped. By contrast, the "ugly" characters (the fat men with wrinkles and age spots) are much more successful and realistic looking; the flaws create a more tangible landscape of the face to which [real] people can relate. I think one of the fundamental difficulties in the design of game characters is that the designers can't (or perhaps haven't yet) capture the shadows and slight flaws that make the human face more lifelike...or perhaps they've avoided the flaws because they feel flaws make the character less attractive.

Posted by: royalgeekg [TypeKey Profile Page] at May 31, 2006 10:31 AM

Dude.

In the middle distance shots, she was totally convincing. Enough detail was lost to the low resolution that the mouth didn't stick out like a sore thumb.

In the close-ups, the mouth... yeh, they need to work on that. I mean, the proportions are in the range of normal humans... I've seen popular actresses who have really wide mouths like that. But there wasn't anything INSIDE her mouth, the edge of her lips were like the edges of a wound. If they made the mouth smaller, and did something about the lips, I think I'd have to say they could make it to the other side of the valley.

Posted by: Leechman [TypeKey Profile Page] at October 9, 2007 7:55 PM

Post a comment

Thanks for signing in, . Now you can comment. (sign out)

NOTE: If you posted a comment and you can't see it -- try refreshing your browser.


Remember me?