Hey All,
This is a great question and the topic of the Uncanny Valley is one that was brought up at the Game Developer's Conference (GDC) yesterday. Arisilde did a great job touching on what it actually is, but let me go a little deeper so there's a clearer understanding.
All animals on the planet - including humans, have a discerning eye when it comes to identifying things that are like themselves. Statistically, people pair up with partners who are about as attractive as they are, and people are very keen to notice visual anomalies in others - even subtle differences in the symmetry of someone's face.
This is an evolutionary adaptation and isn't something we do consciously. Even babies can recognize a human face from that of an animal's once their eyes clear a bit.
What the Uncanny Valley is, is really the subconscious unease we feel when something looks like it's pretending to be human. The closer something comes to looking human, the more critical we become of the subtle disconnects. And while relevant to movies and games, it's real focal point is on robotics and the creation of machines designed to resemble humans.
Robotics aside, I'd like to say this is one of the reasons why games tend to use stylized graphics instead, but the reality is we're just now really starting to experience the Uncanny Valley in games. Up until recently, real-time graphics lacked the performance capabilities to generate life-like characters in real-time. However, it's something that we've been seeing in movies and cinematics for almost a decade. Perhaps the best example was the Final Fantasy: The Spirit Within movie - the first attempt a feature-length photo-realistic movie.
That all said, Chronicles of Elyria, while maintaining a stylized look that is closer to reality, isn't attempting photo-realism. It's still a stylized game. To be photo-realistic would require significantly more complex shaders than we're using, better global illumination, higher res textures, etc.
So while the Uncanny Valley is definitely a concern in robotics and modern media, our ability to create photo-realistic characters in real-time that trigger the UV-response is still a ways away.
Of course, facial animations still suck in most cases and causes it's own kind of nausea. :-)