The Uncanny Valley

April 18, 2012 1 Comment by 3DBlog

Have you heard about the Uncanny Valley? This is not a scary movie or a horror story by Stephen King. This is a phenomenon that was first identified by the robotics professor Masahiro Mori in 1970.

In layman’s terms this means a kind of glitch in the human brain.

When you see an object for the first time, you experience kindly feelings to it if it is humanlike. The more humanlike its appearance is, the stronger your feelings. You are indifferent to electric iron or a toaster, because these are just machines that satisfy your needs.  There is no similarity between the toaster form and human body.

But when you see the WALL-E robot in the computer-animated film you are overwhelmed with tenderness and sympathy. Why does this happen? Just because you cannot tell WALL-E is a piece of inanimate iron. You can see emotions in his eyes.

You probably like the RoboCop or the characters from the “Despicable Me” film. They are close to humans but still have a noticeable difference. Gru’s legs are disproportionately thin and that’s funny. The RoboCop hides his eyes and we can’t discern if his eyes have a glassy stare of the lifeless robot.

One would think if an object is maximally close to human, then kind feelings should almost be at maximum. But that is not the case! In fact when an object is a robot or a 3d character and it looks almost as if it is a human, it provokes an utter aversion. You can see it on the graph below.

There exists an abnormal zone of enmity that is called Uncanny Valley. Most of the people find humanlike objects scary, creepy and eerie, for example, like this android.

So now the 3D artists and animators have faced the problem of making the graphic objects as human as possible but not too human.

The first issue is symmetry. We can tell you for sure that you’ve never seen the person with the left and right eye identical in size and shape. Asymmetry is quite natural for the biological organisms. That is why the 3D modeler should be very careful with the symmetrical reflection. It seems to be a simplest solution: to build the left part of the face and then make a mirrored copy. But this option doesn’t work. An experienced 3D modeler always adds slight asymmetries to the object. It can be a wider eye or a higher position of one brow or asymmetric freckles. Even the shape of the face and body should have some slight defects.

The second and the most important is the movement and motion of the object. The animation of 3D objects is built on the sequence of slightly noticeable movements. It’s very hard to re-create the movements of real people with all body signs and involuntary movements. Even blinking! Besides the body pose and location should always correspond to the character’s state of mind and temperament. For example, the movements of an aggressive character should be steady and confident.   If the head and shoulders are slumped down, it would not accurately reflect the character.

Creating your own character is a long and diligent process. Fleshing out your character is even harder than building a wireframe model with the textures.  The perfect recipe is patience and a talented artist.

One Comment

  1. Frank S says:
    Thursday, September 27, 2012 at 6:38am

    I have never thought of it. But now as I have read the article it seems that I begin to understand some feelings I’ve experienced in the past


Post a Comment

Your email is never published or shared. Required fields are marked *