The video above is a collaborative work between James Portnow and Daniel Floyd discussing the Uncanny Valley.  It is based off an article by James Portnow, the full text of which can be found below.

The Uncanny Valley...some of you may not have heard of it, many of you may not know it by this name, but it’s one of the most important concepts in character design today.

Over the course of this article we’ll explore the Uncanny Valley, plumb its depths, and discuss its ramifications for the industry.
 
What is the Uncanny Valley?

"I have noticed that, as robots appear more humanlike, our sense of their familiarity increases until we come to a valley. I call this relation the "uncanny valley." - Mori, Masahiro (1970). Bukimi no tani (or The Uncanny Valley)
These words came from a famous Japanese roboticist way back in the 1970s.  He had noticed a strange trend.  At first, as he made better and better robots, people liked them more.  The new robots hinted at human characteristic, they were like big, bolted, awkward children and everyone loved them.   But as he continued to make improvements, as he added synthetic skin and rudimentary facial expressions, he found to his surprise that people no longer liked his robots.  Oh, other members of his field were impressed with these new robots and everyone acknowledged they were an “advance”, but being around these new robots made people uneasy.  No one “bonded” with these increasingly realistic androids in the way that they did with his earlier more primitive designs. 

This observation led Masahiro Mori (the robotisicist in question) to come up with the theory of the Uncanny Valley.  The premise is simple:  if something is clearly not human but has human qualities, we find those qualities endearing, but if that something becomes an imperfect simulacrum we find it disquieting and revolting.

So, How Does This Apply to Videogames?

Graphically, modern “next-gen” games have advanced to the point where we have to address the problem of the Uncanny Valley.  Many proponents of high end graphics say we have already crossed the valley and are firmly on the other side.  This is a myth.  It’s much easier to say that the problem is behind us, that we solved it accidentally at some unnoticed time in the past, than to address the fact that hundreds of millions of dollars spent on graphics have not brought us to a solution.

I’ll give the counter argument:  we have not yet crossed the Uncanny Valley. “Photorealism” alone will not break down the empathetic wall. 

Let’s look at some characters:  Mario (human caricature), Master Chief (devoid of facial features and expressions) and Final Fantasy characters (stylized humans).  These are some of the most recognizable characters in gaming today, yet none of them are photorealistic.  In fact, if you make a quick mental list of videogame characters, say the first ten that come to mind…how many of them are photorealistic humans? 

When a character is clearly not human their human characteristics stand out.  Mario becomes more cute and funny, Master Chief becomes more stalwart and heroic, Final Fantasy characters become more earnest and expressive, because those are their most human characteristics. 

We forgive these characters much because we don’t expect them to live up to our experience of human beings.  We don’t require them to move perfectly or to react to danger like a human being and, when they get stuck on a two foot wall we’re willing to excuse it as an annoying idiosyncrasy of the medium rather than being torn out of the experience.  Most importantly: these characters invite us into the fantasy which they exist in by requiring us to use our imagination.

Other Media

Before we move on to talk about photorealistic characters, let’s talk about the Uncanny Valley in other media.  Film is probably the easiest example.

What’s the difference between a film with poor acting and one with great acting?  One sits firmly at the bottom of the Uncanny Valley while the other lies somewhere far up its right-hand slope. 

And what films can get away with bad acting?  Action films where the main character is not intended to be “human” (this includes characters like James Bond).  The best action films end up at the peak on the left hand side of the valley, with characters far enough from everyday humans that they can charm us with their human traits (Bond, Indiana Jones, Ash Williams).

In part, this is due to the fact that the medium of film is largely constrained to using literal human beings as its central characters.  Videogames suffer from no such restriction.  We can use the physicality of our characters to emphasize whatever psychological effect we are trying to achieve.

So What Are the Problems with Photorealistic Characters?

The main problem with presenting the player with a character that is almost, but not quite, human is that all the areas where the character falls short of being human become glaringly obvious.

moscalloutThis doesn’t mean we shouldn’t aim for realistic characters.  It just means we haven’t gotten there yet./moscalloutI’ve seen the same player play a hyper realistic game, get stuck on geometry and say “what the h*$%, that’s so stupid, I’d never get stuck on that” and then play a Mario game and have the same thing happen without it phasing them in the least.

This effect occurs wherever we approach human realism.  Many of you have heard your art teams talk about sub-surface scattering (the light that bounces around under the skin and then gets reflected back out), people notice if that level of detail isn’t present in a human being.  This holds true across all fields...animation, voice acting, you name it.

So Where Does that Leave Us?

This doesn’t mean we shouldn’t aim for realistic characters.  It just means we haven’t gotten there yet.  As I see it, this leaves us with two options when creating characters:

1. We can make next-gen photorealistic models and then try and increase fidelity by removing little glitches, smoothing out animation, perfecting voice acting and assigning human motivations to our characters.
2. We can try and perfect the craft of exemplifying human characteristics in “non-human” characters.

Number one seems like the duty of large developers.  Two seems like the purview of everyone.

Parting Shots

Both these routes are dangerous.  Delving further into creating believable humans is expensive and can easily fall flat.  On the other hand, creating characters that are human enough to make their human traits really resonate with us, while still avoiding the Uncanny Valley, is a fine line to walk.

So in the end it’s not that spending money on graphics is bad.  It’s necessary.  Graphical fidelity should be pushed to the limit. It’s just that graphical fidelity does not equal human fidelity, which is a problem that the industry has yet to admit.

Crazy Talk

I’m going to leave you with one last conjecture. I’m going to go out on a limb and say that, for videogames specifically, the peak on the left hand side of the valley can be as high as that on the right.

I’m out of space, so I’ll just let that hang.  I’d love to hear your thoughts on the idea as well as your arguments for or against it. 

As always, if you have any thoughts or question or insights into how to cross the Uncanny Valley, you can reach me at jportnow@gmail.com.