Something went wrong. Try again later
    Follow

    Uncanny Valley

    Concept »

    When an animated character graphically looks realistic, but shows poor, unrealistic animations, the result can be very awkward. These instances of almost-human-but-not-quite can appear stilted, off-kilter, or even eery. This is the Uncanny Valley effect.

    Shouldn't the familarity be on the X-Axis and likness on Y-Axis?

    Avatar image for godwind
    Godwind

    2924

    Forum Posts

    345

    Wiki Points

    0

    Followers

    Reviews: 0

    User Lists: 0

    #1  Edited By Godwind

    I was looking at the page, but shouldn't the two axises be relabeled?

    Avatar image for ricosrevenge
    RicosRevenge

    8

    Forum Posts

    574

    Wiki Points

    0

    Followers

    Reviews: 0

    User Lists: 0

    #2  Edited By RicosRevenge

    I think I understand what you mean, but the whole idea is that the x-axis represents something's "human-ness," with "not a person" to the far left and "a person" on the far right. Familiarity in this case means "people like it," and the traced line indicates how much people like/dislike the thing as it progresses from "not a person" to "a person."
     
    I dunno if this is really the best response, but I saw that nobody replied to this and figured you deserve at least some explanation.

    Avatar image for mrwizard6600
    MrWizard6600

    21

    Forum Posts

    0

    Wiki Points

    0

    Followers

    Reviews: 0

    User Lists: 0

    #3  Edited By MrWizard6600

    yeah its a poor choice of words to describe the two axis.
     
    as you move outward on the x axis the object becomes more and more human unti you have an actual human at 100 (han solo or whatever). So at 0 you can have anything, a mug, dirt, whatever, and you have no empathy for it, or as the chart would say, you have no familiarity with it. As you move further and further up as objects become more human, for the most part your relation to them increases linearly, or even exponentially. For every extra human characteristic you add the object becomes more human, to a point. At that point, or range, suddenly we stop focusing on th e characteristics that make him/her/it human, and start focusing on the characteristics that make them not human, which can often be unpleasant (or at least I'd imagine being dead is unpleasant). 
     
    But again, its a poor choice of words. You could swab the two descriptions and it would still make sense. The X axis is meant to represent the characteristics that make people people. I don't want to go all psychological on you but a thought process, free will, 2 hands 2 feet, etc. If you create an object with ~70% of those features, rather than focus on the fact that this thing /him /her has 2 feet, 2 hands, a head, and a free will, we focus on the fact that it doesnt have a proper face or whatever.
     
    ps: OP nice rack.

    Avatar image for grumbel
    Grumbel

    1010

    Forum Posts

    12

    Wiki Points

    0

    Followers

    Reviews: 99

    User Lists: 2

    #4  Edited By Grumbel
    @Godwind said:

    " I was looking at the page, but shouldn't the two axises be relabeled? "

    I am a little late to this discussion, anyway here it goes: I consider the Uncanny Valley graph to be basically useless, as the X-axis is completly arbitrary. There is no objective or even proper subjective way how you qualify "human likeness", thus you can basically place your observation wherever you want on the graph to make the theory work. So for example take a Half Life 2 characters are they on the left side of the valley or on the right side? Are they more "human like" then characters from a cutscene from a early Playstotion1 game, say Final Fantasy? How do they compare to the characters of Polar Express? You can basically place them where ever you want on that graph to proof your point, without much of a way to verify or reproduce your placement objectively, i.e. you are fitting your observations to the graph, not the graph to your observations as you should.
     
    Now that of course doesn't mean that computer graphics or robots can't look uncanny, they often do, but not because they fall into some mystical valley, but because they get stuff wrong.  If you motion capture a face and sloppily apply them to your CGI face so that all the data is of by a few millimeter things will wrong and uncanny, because those few millimeter make the difference between a normal expression and a weird  grimace. More complex graphics provide of course more points of failure then cartoon graphics, but they don't get uncanny all by itself just because they are more advanced. You can have rather simple graphics that look uncanny and pretty complex ones that look just fine. The important part is simply to keep the balance right, if you include additional detail in the graphics, make sure that that detail works correctly, as when its wrong it will end up looking uncanny.
     
    Over the last few decades in game graphics there basically never was an uncanny valley, graphics overall overall  just got better and more realistic with each generation. Did they look uncanny every now and then, sure, but you have uncanny games on the Playstation 1 and great looking ones on the Playstation 3 and visa verse.

    This edit will also create new pages on Giant Bomb for:

    Beware, you are proposing to add brand new pages to the wiki along with your edits. Make sure this is what you intended. This will likely increase the time it takes for your changes to go live.

    Comment and Save

    Until you earn 1000 points all your submissions need to be vetted by other Giant Bomb users. This process takes no more than a few hours and we'll send you an email once approved.