The Uncanny Valley refers to a chart created by Japanese roboticist Masahiro Mori. He noticed that, when building robots, as he added human-like features to a robot ordinary people would respond more positively than if it were some abstract collection of steel and wires. However, as he added more human-like features the familiarity with which people responded began to crest and then take a huge drop into unsettling territory. The valley itself comes from this chart:
Basically this was what he was referring to:
You can see that this robot has all the physical features of a human being, but it lacks true human-like movement. Notice the way it shudders and sways with every major movement, and the way it's mouth seems to only open and close, very unlike the human mouths broad range of expressions. It is clear that the woman in that video is a robot, but the way that the likeness of a human has been tainted makes it even seem a bit frightening.
In recent years, as graphical fidelity has increased to staggering levels, the video games industry has made many attempts at complete photo-realism when it comes to human beings. Two noteworthy titles are Heavy Rain and L.A. Noir. The former billed itself as an adult drama story with realistic characters and an engaging plot. Here is one of the early videos the development studio released showing early animations of one of the main cahracters:
Here the thing that stands out (to me anyways) is the mouth. When she smiles and her teeth jut out like some kind of grotesque undersea creature... it makes me shudder every time. Granted this was early in the development process, but it illustrates the kind of hurdle that the Uncanny Valley presents.
L.A. Noir is as of yet unreleased but seems to have a little bit more success, here is a trailer for that:
This game seems to have a bit more success with the facial expressions, but this is a result of some new and advanced facial capture technology, which we can readily assume is quite expensive.
So how do developers normally get over such hurdles? Well a very common answer is to use stylization over photo-realism. Cartoon-like styles help make more endearing characters while also helping accentuate emotions. There's a much more readable visual language at work. My favorite example of this is the Ace Attorney series. Take a look:
In conclusion, the Uncanny Valley is really only a problem of approach. Absolute realism is a lofty goal, but if the pursuit of it is to make more "believable" or "endearing" characters, the effort needed to avoid an uncanny disaster would be much better spent on designing a less realistic looking character with a lot more actual character.
-Sergio L. Barrio
It's often true when they say simple is better. You look at the sparsely detailed anthropomorphic animal-beings of comic writer Jason (he only goes by one name) and I couldn't imagine the stories having the same emotional impact with detailed human art. Maybe we need things to be exaggerated for them to seem more real? Maybe if an artist could capture the complete essence of man, physically and mentally, we still just wouldn't be as interested.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteThis is very interesting. I think that the more human-like a character is, the more something off and nonhuman-like stands out; thus, causing a sense of uncanniness. This goes along with Freud's claim in "The Uncanny" that uncanny things are usually familiar. If something is off with something that is not human like, such as a cartoon character, then we attribute that to them being fantasy and don't think of them as uncanny.
ReplyDelete