I found a video this week describing how the LA Noire team are using advanced technology to make their characters as realistic as they can. You can watch it here.
However, when you see the demo scenes, it feels just plain weird. They can spend millions on the latest technology, I will never confuse their characters for real human beings! Why is that? It’ s a principle called the Uncanny Valley.
The uncanny valley is a theory coming from robotic engineers, that basically states that a character (or a robot) has more chance to be liked if it has some human features. However, the more the character is made to be realistic (a fake human being), the more our brain identifies every details that is wrong about this character and makes us feel uncomfortable.
The same is true with any media trying to replicate human beings. Video games are not an exception. This theory explains why we are so quick to feel familiarity to characters with exaggerate features, like cartoon characters. When a character is stylized, our brain is happy, because it understands perfectly the patterns shown and will fill the “holes” that makes that character unrealistic. Think of the Gestalt psychology theory, for instance. However, when the character is close to be photo realistic, the brains becomes really picky about the details that make this character a fake. This is truer when the character is in motion.
Movies, on the other hand, don’t have these problems. Actors are real human being and they are captured perfectly by the camera. Animated movies often use really stylized characters to avoid the uncanny valley.
What Rockstar Games team is trying to do is scanning the face of real actors and transposing them on artificial bodies that moves thanks to motion captures made separately. The idea seems good, but unfortunately, it looks exactly as it is: a face glued on a rigid fake body. And no matters how hard they will work on it, it’s just almost impossible to get through the uncanny valley. Unless they find a technology that will scan perfectly the actors in 3d and makes 3d models out of them. I’m sure this technology will exist eventually, but we’re not quite there yet. So until then, why not letting some space for imagination?



























