The uncanny valley
I know that I suffer from what's called "The uncanny valley"
The uncanny valley is a hypothesis that when facsimiles of humans look and act almost like actual humans, it causes a response of revulsion among human observers. The "valley" in question is a dip in a proposed graph of the positivity of human reaction as a function of a robot's lifelikeness.
So movies like the Polar Express, Beowulf and even to a certain extent Shrek all creeped me out because the actors were almost human looking.
Although, for the record, Angelina Jolie transcends this for some reason.
Must be the lips.
But oddly, I find this even creepier for some reason...
posted by The Sum of All Monkeys at 10:30 AM
2 Comments:
This problem is vastly underestimated, and is also nothing new - the difference is between a cute cartoon you might doodle on a napkin and an ambitious but failed portrait.
The cartoon is a low information symbol of a face. A person reading the image responds to the symbol. The information requirements to succeed as a symbol of a face are very low- even infants respond to a happy face. It's the association between the symbol and our experience of faces that provides the meaningful recognition that gives us pleasure.
In a failed portrait, large amounts of detailed information, and all their interactions, are presented - the complexity piles on exponentially: details of texture, shape, light, shadow etc. As the complexity rises, what you might call piles of garbage information - bad description, incorrectly considered relationships etc, also rise. As you build good information and present it in ever increasing quantity, you also, if not highly experienced, present a lot of bad information that contradicts what you are trying to describe.
Hence, a crappy portrait that was a lot of hard work.
This juxtaposition of lots of good and lots of bad information about a face creates a yes/no response in someone reading the image - a huge dissonance, and hence creepyness.
A good portraitist spends much, maybe most of his or her time removing bad marks, bad paint, etc that contradict what is essentially TRUE about the portrait.
Removal becomes an essential descriptive strategy - discovering what is essential to a personality expressed in a face -and balancing that with observed characteristics that too are essential- is as difficult a problem as exists in art.
The computer techniques by contrast tend to just pile on more info, good and bad. And you get bad color, bad lines, lagging, and general mushiness which makes uniform what should be specific to the person's image.
In the best digital animation, the technique is barely distinguished from traditional painting. It' a huge amount of hard work, a person or many people spending time making hard decisions about what is important and not, what belongs and doesn't, about what is specifically descriptive.
It boils down to someone becoming completely aware of the figurative image they are creating - if that isn't done, the work is likely to be remain unconvincing despite the sophistication of the software, or even because of it.
I think that we learn from pretty early on (at least we did before CGI) the difference between an animated face and a live-action face. Getting in between that boundary is likely to cause some discomfort.
Post a Comment
<< Home