Forgot your password?
typodupeerror
Science

The Tipping Point of Humanness 272

Posted by CmdrTaco
from the isn't-that-uncanny dept.
sciencehabit writes "Robert Zemeckis, take note. Using videos that morph the face of a baby or man into a doll, researchers have figured out at what point we stop considering a face human — and start considering it artificial. The ability, the researchers say, is key to our survival, enabling us to quickly determine whether the eyes we're looking at have a mind behind them. It may also explain why so many people hated The Polar Express."
This discussion has been archived. No new comments can be posted.

The Tipping Point of Humanness

Comments Filter:
  • "It may also explain why so many people hated The Polar Express."
    It may also explain why so many people loved the LOTR trilogy (Gollum).
    • I didn't hate The Polar Express, but it wasn't exactly very memorable.

      The simulated Jeff Bridges in Tron: Legacy is pretty decent, though they should have put a little more work into getting his mouth to move naturally when speaking. Any single frame looks just like Jeff Bridges sure, but when it's all put together the effect is still a little stiff.

      • Re:LOTR (Score:5, Funny)

        by TheLink (130905) on Thursday December 23, 2010 @11:56AM (#34652674) Journal

        Any single frame looks just like Jeff Bridges sure, but when it's all put together the effect is still a little stiff.

        I've noticed a similar issue when watching Nicholas Cage and Keanu Reeves movies.

      • by Moryath (553296)

        Actually, the effect would have worked just fine if they'd only used it for Clu. After all, Clu was supposed to be a corrupted, visual simulation.

        The problem is, they used the effect for that first "Sam as a kid" scene as well, which just didn't look right.

      • by trum4n (982031)
        I think that was the point. He isn't Jeff. He's CLU. He is a simulation.
      • by Guspaz (556486)

        I'll agree. It wasn't the visual design that did it for me, it was the animation.

        Looking at the CG face, it *looked* perfectly human, but it didn't move very naturally, and that moved it from perfect into uncanny valley for me.

        From behind-the-scenes imagery, it looks like they used a multi-camera head rig with visible-light tracking points (not many cameras, though). There are some shortcomings to this sort of system, and it seems like on top of that they over-interpolated the motion data...

        From what I've s

      • After Tron: Legacy I told my son Clu's face was very Oblivionesque. I can't decide if they didn't spend enough money/time working on it or they were simply screwed by being up against the uncanny valley.

    • Exactly; Gollum was one of the most expressive and real CGI characters I've ever seen.

    • by denzacar (181829) on Thursday December 23, 2010 @12:35PM (#34653066) Journal

      LotR is based on a seminal work of fantasy literature for all ages, read by generations of readers over the decades. So it is fair to say that it already had an established fan-base.
      It also featured a whole lot of "real people" actors, most of them of a rather high caliber.

      Polar Express is based on a 1980's children's book, based around a character created by Coca Cola's marketing division.
      A character that has since then grown into a symbol of consumerism like no other.
      Oh, and the animation sucked.

      Also, one features a HUGE universe and loads of heroic battles and quests, while the other features... well... public transportation.

      • by StikyPad (445176) on Thursday December 23, 2010 @02:21PM (#34653964) Homepage

        The image of Santa as a plump man was popularized by "A Visit From St. Nicholas ('Twas the Night Before Christmas)" in 1823:

        He had a broad face, and a little round belly
        That shook when he laugh'd, like a bowl full of jelly:
        He was chubby and plump, a right jolly old elf,
        And I laugh'd when I saw him in spite of myself;

        And the modern icon is generally credited to Thomas Nast, [osu.edu] circa 1880, upon which Haddon Sundblom based his Coca-Cola ads a full 50 years later. At most, Sundblom popularized the red suit, but he was quite an artist in his own right, so calling it a "character created by Coca Cola's marketing division" is both giving their "marketing department" too much credit, as well as doing a disservice to Sundblom. It's more accurate to say that Coca-Cola's advertising used to consist of actual art.

    • Re:LOTR (Score:5, Interesting)

      by Dan East (318230) on Thursday December 23, 2010 @12:35PM (#34653068) Homepage Journal

      I've watched the Polar Express at least a dozen times (our 2 year old is really into trains at the moment). I think there is a problem with that movie besides just the rendering. Tom Hanks' acting was motion captured for practically every character, including the lead boy. Hanks has very specific mannerisms, and I don't think they were ideal for portrayal by CGI characters. For example, he does this stiff-necked type expression, where he keeps his neck stiff and rotates his whole torso part way towards the subject, then he looks at them out of the corner of his eyes (usually with a sort of bewildered expression). It's a very stiff motion, almost like he's wearing a neck brace. He does that several times in Polar Express, and it simply exacerbates the problem and makes the characters look more rigid and artificial. If anything they should have been slightly more articulate and dynamic to compensate for the negative expectations people would already have over the CGI.

      I think it's also an issue of scaling. You can't just take an adult, have them pretend they are a child, scale them down to child-size, and have it look exactly right. The proportions and mechanics are wrong for one thing, besides the fact that kids naturally move and act like kids, and adults naturally act like adults.

      Originally Hanks was even going to voice every single character, and voice affects would be applied to change the pitch, etc. In fact, there is a trailer floating around in which Hanks voiced the lead boy's voice, and it sounded horrible. I'm glad someone with enough clout was able to step in and convince Hanks otherwise without stepping on his toes too much. I think Polar Express demonstrated the same problem with the motion capture - everyone is drastically different in their motion and mannerism, and if you use one person to portray a dozen different people, then they will all look unnaturally similar. If your CGI movie has 5 lead roles, you need 5 people to act for motion capture, and 5 people to voice, and they should be cast just like for any "normal" movie. It's that simple.

      • Re:LOTR (Score:4, Insightful)

        by DragonWriter (970822) on Thursday December 23, 2010 @01:44PM (#34653636)

        I think Polar Express demonstrated the same problem with the motion capture - everyone is drastically different in their motion and mannerism, and if you use one person to portray a dozen different people, then they will all look unnaturally similar.

        I don't think that's a fundamental problem, I think that's a problem with choosing the wrong actor. There are actors that are very good at creating distinct characters through motion and mannerism, and there are actors that have a more limited palette of motion and mannerism. Either -- depending on how well they portray the feeling of a particular story -- can be great actors at playing one role in a piece, but only the former are likely to be successful playing multiple roles (unless they are supposed to have eerily-similar mannerisms) in the same piece -- whether its live action (as in a live "one man" show, or on film using camera tricks) or done via motion capture.

        (The same thing that is true of motion and mannerism is true of voice acting; there are actors that can define distinct characters through voice alone, and actors that can't, and either can be good for one voice role in a piece.)

      • Re: Polar Express (Score:5, Informative)

        by Animats (122034) on Thursday December 23, 2010 @01:44PM (#34653638) Homepage

        The Polar Express seemed to have "rubbery" motion capture. I used to see this problem at trade shows like SIGGRAPH. The electromagnetic motion capture people would have a stage with a live dancer wearing sensors at her joints, and screens showing the CG character driven from the dancer. The CG character always moved worse than the live dancer. If the dancer did a hard stop, the CG character would show much less abrupt deceleration. That's because the electromagnetic systems were noisy, and had to be low-pass filtered.

        There were also alignment problems. The hand positions were usually off. Metal in the area would distort the fields slightly. Around 2000 or so, errors of several inches were still common. I asked one of the demo dancers to touch her fingertips together, and the CG character was off by the breadth of a hand. The Polar Express animation had a similar slightly-off look.

        This got better once motion capture started using multiple cameras at much higher frame rates than the animation. There's still some noise and filtering is still needed, but the noise is up at a few hundred Hz and the filters have higher cutoff frequencies. By the time the motion is downconverted to 24FPS, the effects of the filtering have disappeared.

  • Much as I enjoyed Tron: Legacy, young Flynn/Clu was just wrong enough to seriously creep me out. I think it was because some parts of his face didn't move right when he talked and smiled (cheeks and eyes).

    • For me, it was his mouth that seemed way off. Somehow, the motion of the lips moving either seems to be off-sync with the sound, or exaggeratedly tight-lipped.

    • by hellfire (86129)

      Although the movie wasn't a tour de force, I thought the "creepiness" of Clu's facial expressions was actually a plus, because he was supposed to be creepy, he's the bad guy! The theme was the attempt by Flynn at perfection, and you can see that perfection isn't perfect in Clu's face. However, it probably would have been a better idea to use makeup and a little digital retouching in the initial scenes for young Flynn. Obviously Tron: Legacy was trying to show what current technology could do compared to

  • by Anonymous Coward on Thursday December 23, 2010 @11:35AM (#34652426)

    Yo mamma's so fat, the recursive function calculating her mass density had a stack overflow.

  • It has a few obnoxious scenes, but seeing The Polar Express in IMAX 3D was one of my favorite movie experiences. They really nailed the 3D aspect of it.
  • Missing dimension (Score:5, Insightful)

    by ColoradoAuthor (682295) on Thursday December 23, 2010 @11:38AM (#34652464) Homepage

    Time. Consider our everyday conversations: "Ooh, he's creepy. He keeps looking at my stomach." "Look me in the eye and tell me that." "Watch that customer in the Jewelry department--he's got shifty eyes."

    Examining static images of faces has limited (some, but limited) value. When we look at eyes, don't we immediately calculate *what they're looking at*? Much of our assessment of the character and intentions of people and animals seems to be based on how the eyes move.

    • You can't access character by watching a persons eyes or body language. That doesn't stop people from trying of course.

      • by jfengel (409917)

        Or as Shakespeare put it, "There's no art/To find the mind's construction in the face." That quote leaped to mind when I read your post.

      • by mcgrew (92797) *

        Did you mean "assess"? Otherwise your statement makes little sense to me. If you did, some people are REALLY good at assessing the average person's character by watching body language. I'm not good at it, but some are.

        Of course, nobody can assess a sociopath''s character except a trained psychologist administering a test for the disorder.

        • Of course, nobody can assess a sociopath''s character except a trained psychologist administering a test for the disorder.

          Not necessarily so. The desire for political office, or any authority for that matter, is a pretty sure sign.

      • by Shotgun (30919)

        But you can assess intentions. I wrestled in high school. Later as a coach, I could confidently instruct wrestlers to watch the opponents eyes, both in order to get an idea of their intentions and to never broadcast your move.

        Someone's intentions give hints to their character.

      • You can assess behavior and intent, however, which is much more important in the short term.

      • by TheLink (130905)
        You may not be able to do it 100%, but are you sure you can't do better than random?
    • "Ooh, he's creepy. He keeps looking at my stomach."

      I'm actually looking at your breasts. You might consider wearing a bra once in a while

  • because it wasn't that great of a film. The 3D wasn't terrible, but most folks I know really didn't care for the story.

    • by Reece400 (584378)
      I watched it last night on TV and the story was lame, but the kid's facial expressions were also damn creepy at times.
      • it's like they rolled a porcupine in botox and slapped the characters in the face with him; the mechanical monkey in "Toy Story 3" was more lifelike.

    • Well, that's happens when you expand thirty or so lines of text into a ninety minute movie...
       
      But seriously, the 3D *was* that bad. When characters opened their mouths, you could see the inside of their skulls, etc... etc...

  • by macraig (621737) <`mark.a.craig' `at' `gmail.com'> on Thursday December 23, 2010 @11:53AM (#34652648)

    My cats look directly and intently at my face every day, and it's obvious from the circumstances that they recognize that a mind with intent is attached to those eyes and they're eager to figure out what that intent might be (and whether it might adversely affect them). This is not at all a behavior exclusive to primates, much less humans. Presumably that means my cats would have hated The Polar Express, too. They're already annoyed by Tom Hanks' nasally voice.

    • by jfengel (409917)

      Face-watching is certainly not unique to humans, but the "uncanny valley" presumably is. Or if other species have a similar response, it's a lot harder to test.

      We might well learn something if we could get a cat to respond to The Polar Express, but getting cats to do ANYTHING reliably in a behavioral study is a pain in the ass. You'd be better trying it on dogs.

      For that matter, it would be interesting to see where the uncanny valley arises in observations of other species. Does can you get that creepy ef

    • You need to consider that scent is as almost a large part of a cat's perceptual universe as sight. Even the best CGI or robotic cat simulation won't have the scent.

      • I suspect the uncanny valley response in humans is an early warning system against communicable illnesses, we react adversely to anything which doesn't quite fit into the way a person is meant to look or behave both of which could indicate infection before more overt signals can be seen such as more obvious skin discolouration, sweating, physical reactions such as coughing and sneezing etc.

        If my suspicion is correct then the only other animals which would experience a strong uncanny valley effect would n
    • by 140Mandak262Jamuna (970587) on Thursday December 23, 2010 @12:41PM (#34653114) Journal

      My cats look directly and intently at my face every day, and it's obvious from the circumstances that they recognize that a mind with intent is attached to those eyes and they're eager to figure out what that intent might be (and whether it might adversely affect them).

      Dogs you mean. No other animal can beat dogs when it comes to reading the human mind. Most animals don't even know to look at where we are pointing at. Dogs have evolved with humans for the last 30,000 years. I posted earlier the theory about dog-human interaction could be the one that led to sedentism that was the precursor to the domestication of plants and agriculture. Someone asked for references. See Nicholas Wade's book "Before the Dawn" for a good over view of "The Great Leap Forward". (But the main thrust of that book was building inheritance trees of the Y Chromosome, the mitochondrial DNA, DNA of the body louse, the tree of languages etc and showing how they all agree with one another and gives us clues about fixing crucial dates before the recorded history. For example lactose tolerance and cattle domestication in west-central Europe about 8000 years ago. Or the correlation between horse based civilizations and Indo-Aryan language family. )

    • by DocSavage64109 (799754) on Thursday December 23, 2010 @12:55PM (#34653236)

      My cats look directly and intently at my face every day

      Maybe your cats are just waiting for you to pass on so they can eat you? Actually, I do wonder what they are thinking at such times... maybe something as simple as love.

      • by Nadaka (224565) on Thursday December 23, 2010 @01:24PM (#34653476)

        They are attempting to calculate the best approach to getting fresh food in their bowl based on your mood and their own. Do they howl until you give in? A little mewl and a flick of the tail? A pur and flop next to the bowl so that you notice it is empty while giving a belly rub? That little head butt thing that says "you one of my people and thats cool with me"? Do they sit on your dinner plate? Do they walk up to a glass of grape juice, look you in the eye and then knock it onto the carpet?

    • by mcgrew (92797) *

      Cats love to try to stare you down. It's a game to them (and they hate losing).

      Cats are far more intelligent than anyone gives them credit for.

  • by Drakkenmensch (1255800) on Thursday December 23, 2010 @11:55AM (#34652668)
    The concept of the uncanny valley is a well known one: http://tvtropes.org/pmwiki/pmwiki.php/Main/UncannyValley [tvtropes.org]
  • But, but ... they *weren't supposed to look alive*

    I suppose the story got a little fouled up in the editing, but here is a clip from the director's cut [whatisdeepfried.com]

  • I can agree (Score:2, Interesting)

    by Nyder (754090)

    I'm an eye person. It's what I love about chicks the most (but their soft curvy parts come in a close second).

    But besides that, i can read people by looking in their eyes. Hard to explain, but if I make I contact with you, I connect to you somehow and can figure out what's going on inside.

    That being said, it gives me the creeps as most people really, well, suck.

    So if i'm looking at your cleavage when we talk, don't be offended, you don't want me to read what's going on in your head.

    (you might think i'm

  • by Arancaytar (966377) <arancaytar.ilyaran@gmail.com> on Thursday December 23, 2010 @12:07PM (#34652798) Homepage

    Yeah, I guess it'd be pretty important if the zombie uprising ever happens, or the world is taken over by sentient dolls.

  • by seebs (15766) on Thursday December 23, 2010 @12:29PM (#34653014) Homepage

    I'm autistic. I don't seem to have the automatic distinction between things with minds and things without minds. In fact, I can occasionally forget that other people have minds, briefly. For instance, a couple of days ago, I was pinching Beloved Spouse's cheeks, and I suddenly got fascinated with how the various components of the face are connected and deform each other. I started messing with this. Suddenly it occurred to me: There is a person experiencing this, and it may not be a preferred experience. But there you have it; for a good four or five seconds, I had completely forgotten that my spouse was a sapient creature. While staring directly at said spouse's face.

    I can't think of an occasion on which this has been any kind of survival problem. (My spouse is very forgiving.)

    I suspect that it's useful to get this stuff automatically, but it also produces all sorts of strange buggy behavior when we find things that trigger the "that's people" grey matter but which aren't actually people.

    • Wait, doing shit like that means you are autistic? I'm not sure that's right. Then again, I am occasionally seized by the certainty that the relative densities of thoughts in my skull produce surface-like tensions that can be bounced upon like a trampoline and also drown me... Anyhow, are you sure you aren't just a mental illn'iss'ist like the rest of us?

      Relating back to the thread: based on my experience in a human skull our main concern should be not making any one thing both complex enough to fail as u

      • by Doomdark (136619)
        Wait, doing shit like that means you are autistic?

        No, diagnosis by doctors confirms you are. Just because he didn't say he has been diagnosed does not mean he hasn't. I would assume implication is that there is the diagnosis. Or are you thinking he is just bragging about being autistic?

  • by Quiet_Desperation (858215) on Thursday December 23, 2010 @12:32PM (#34653040)

    There's simple artistic concerns as well beyond the math. Zemeckis and others need to sit down and understand why Pixar's hand crafted, "super deformed" characters come across as orders of magnitude more realistic than high tech attempts to directly dump humans into the computer. I've always felt if you want realistic humans just use actors, and CGI everything else.

  • How can they say (Score:4, Insightful)

    by bugs2squash (1132591) on Thursday December 23, 2010 @12:58PM (#34653266)
    that the cut-off is at the 67%/33% mark ? After all, one end of the scale is fixed, a picture of a real human, but the other end is not fixed, should they have drawn the line at 99%human/1% lego brick ?
    • by Legion303 (97901)

      The other end of the scale is a picture of a real doll that looks similar to the initial real person head shot, from what I gathered.

  • Other cultures have different rules on eye contact. In the second part of the study when they measured where people looked the most, the people were looking at the eyes. It's hardly surprising that if you make the bits that people focus on look more artificial, they think the whole model looks artificial.

    If they had done this test in a different country where people don't make eye contact then the results might have been different.

  • Looking at the example side by side comparisons at the top, I think I was latching onto the wrong changes. I think they were trying to push the geometry around a little, but what *actually* caught my eye was the ones on the right look like they have a lower poly count and lower res textures. Was that actually what they were trying for? If not, they would have made a better test by similarly de-resing the originals.

Save gas, don't use the shell.

Working...