@@alamrasyidi4097 It's especially weird because the last 5 or so all look perfectly human, it's just that he has a different beard. I guess a 70s pornstache makes you more sympathetic somehow.
@@alamrasyidi4097 Yeah I'm just pointing out that the entire graph makes no sense whatsoever, the first 4 are all stuff of nightmares, "cute robot" isn't how I'd describe them. "Cursed doll" or "psycho murderbot" would be more fitting.
From the final graph, we can deduce that anime style robot are the optimum point in human likeness and likability, and thus all robot in the future must be build with anime waifu in mind to achieve the optimum point.
Wow. This is something I’d just accepted as fact without questioning it deeper. Until watching I didn’t even notice that I’d done that. Makes you wonder how much else of what “is just known” may also not be on solid foundations. Of course as you point out at the end, that doesn’t make what you’re saying automatically correct either. What a wonderful nudge towards keeping an open mind and critical thinking. Great video, great message.
I'm in the same boat. I think that its easy to accept explanations that sound like they make sense, but its difficult to actually rigorously go through things. I also think its difficult to think of alternative explanations when you already have an explanation you accept. Or it could be that even that which I stated above is just a commonly accepted explanation which may not be entirely accurate XD I guess the best we can do is to take a scientific approach. That a theory is only as good as its predictive power(especially when it can predicts an unexpected result). But what predictions can the notion of the uncanny valley make?
Spoiler: We know pretty much nothing about anything. The more we think we know, the less we actually know. As an example, anesthetics. We have no idea HOW they work, just THAT they work, and we as a species use it millions of times per week to perform life-altering surgeries.
but it still seems like he did not understood what uncanny valley really is and his examples of why uncanny valley doesn't work are "cherry picked" too.
Nothing is on solid foundations, I’m sure the people who thought of the uncanny valley didn’t portray it as truth, but just as a hypothesis. It’s everyone else who took it as truth
Yum, chocolate & tomato lasagna! LOL, it's still uncanny. Also pizza + cake = bread, not lasagna, lasagna is (supposed to be) a delicate pasta sheet layering and closer to puff pastry if anything. Instead both cake and pizza are made of rough dough.
The Uncanny Valley could also be described as lopsided focus of design. For instance if you only improve the likeness of the eyes or teeth but leave the rest as robotic. Or concentrate on physical appearance only while having unnatural movement.
Or just anything that seems out of place. if you've ever seen the "teddy bear with human teeth" photo, it's quite unnerving and yet nobody says "it's because the bear now looks so close to human that it is making us notice it's not a human". It's because the bear looks unnatural and disturbing and is causing a different part of our brain to be active that isn't usually active.
He just said that the uncanny valley isn’t real but is also real but isn’t real because of why we thought but it is real and also that he may be wrong but the uncanny valley is real not though but kinda but real but no
@@cakiepop2038 The point of his explanation is that, though the uncanny valley can be perceived as real, it isn't because what we're using to view it is relatively made of unrelated examples. TL;DR He said uncanny valley untrue because you're trying to compare apples to oranges.
Fightre Flighte That makes absolutely no sense. So something isn’t real just because we wrong about how we originally perceived it? We knew space existed before we made telescopes my dude.
@@cakiepop2038 Not necessarily. Keeping in mind he was open about how he could be wrong. Others have also commented from areas of expertise in robots that both approve and deny. Also, yes. We thought space was a bubble around us, originally, and that we were the centre of the universe. Then realized we were wrong, and that the bubble around us is our spherical atmosphere, and that we aren't the centre of the universe. Before realizing we were wrong again, because we are, literally, the centre of the observable universe. Just because we see something, perceived it incorrectly, but saw trends that look related to each other; that doesn't mean it's right or wrong off to bat. That means it's up to deliberating if it is or isn't. So, are you going to compare apples to oranges, or are we going to talk about why I like granny smith apples and you like golden delicious more because the datum we're searching through is entirely subjective in the first place. If you think about it, that's all this is. Humans judging subjective subjects off subjective datum, to find subjective results that we want to categorize as correct and accurate, and in a graph that everyone can understand and disagree with.
I somewhat remember this coming up in my robotics classes. What happened was one dude was building lots of different kinds of robots, and he noticed that the closer to looking like a human he made them, the more averse people felt towards them. He reasoned that since humans aren't horrifying, there had to be some point where his accuracy would start trending back up and people stopped being so averse to it, leading to the uncanny valley. I don't think it had quite the level of universality of now until computer graphics started to really improve. There were hundreds of artists and game developers trying to make better and better looking people, but the more accurate the skin and flesh and such looked, the bigger the requirements for making it move accurately was. And if you didn't have quite the right blend, you tended to fall into this really creepy zone where it almost looks human but it doesn't move like one. The more human looking something is, the easier it is to make something that looks distinctly not human.
I was thinking about this the entire video. It's good to point out the flaw in generalizing the concept, but I always knew the uncanny valley as a computer graphics thing, focussing more on the idea that an increase in fidelity (due to improving technology) wasn't enough to continually improve the perceived "quality."
Good point. Likeness vs movement. If you increase the likeness, you must also increase the fidelity of the movements. You should also avoid anything that makes it resemble a dead person (yellowish and inflexible non-transparent skin, stiff expressions).
@@bicycleninja1685 Makes me think of those robots in the game kenshi. They would try to act like a human, even though they were clearly machines with the skin of a dead human attached, so that they could lure you into their camp and skin you alive
I don't know. I kind of feel like this video's creator is getting hung up on semantics. He seems to be picking at the word choice rather then the fundamental idea. On the left side of the graph you have a symbolic representation of a human. The upswing happens as the representation gets better, the uncanny valley occurs as we transition from trying to hint at the human form to trying to duplicate it. The far right side is perfect duplication, landing in the uncanny valley neccessary means mistakes, and human form with viable mistakes *do* look diseased, sickly, or like they are a threat. All his initial points are true ... about the uncanny valley! I just think the maker of this video talked to some people who phrased things in a way that stuck in his craw, and now he has some idea in his head about the uncanny valley which is slightly off from reality ... ironic.
He said it in the video at 4:10. It's not what roboticists claim uncanny valley is. What you're describing is what he believes what uncanny valley should mean.
@@dropmelon But that *is* what the roboticist's uncanny Valley is. As OP said, it's about phrasing. The "classic" uncanny valley is if a thing approaches human likeness, we like it more. but beyond a certain threshold we see it less as "thing looks human" and more as "human looks like thing" and once it crosses yet another threshold it actually fully duplicates a human and therefor is indistinguishable from a person. even the notion of "are they uncanny, or are they creepy" the answer is yes they're creepy, but saying "they're creepy" is actually less descriptive than saying they're uncanny. Why are they creepy? because our brain sees them as too close to humans, but disfigured Since the only real world reasons for someone to look so horribly disfigured is disease, injury, or death (more disease) we have an instinctive aversion to it. The entire concept behind this video can be applied in reverse to reinforce the concept of the uncanny Valley since in the end "its creepy" doesn't mean anything. "it's creepy" means "this makes me think of bad things"
@@techno_tuna I'm still not clear how that is actually distinct from what the uncanny valley already *is*. To me, you just described the classic uncanny valley again. In logic I think this is called the "distinction without a difference". He doesn't like the terminology/the words have gotten twisted up wrong in his head. So he invents a new way of saying the same thing that makes more sense to him and sets it in opposition to the old way, but he isn't actually saying anything *new*.
For me, the main takeaway of this video is that we may be drawing a trendline that doesn't actually exist, or is unjustified. That we are mixing data that really belongs to different categories and plotting them on a graph and drawing a line that looks like it fits. Kind of like with Simpson's paradox. Specifically at 13:48 he notes that he suspects that there are actually two categories *attempts at trying to create human likeness* and *attempts at trying to elicit a positive emotional response* and that we are putting them on one graph which assumes we are plotting likeability vs how human-like they are.
Some people, myself included, already understood uncanny valley like that. "It's hard to make a human without making an ugly human". But a lot of people genuinely believed that it's something akin to a mathematical graph.
I mean, part of the point of the video was he could be wrong too. Don't just take away the opposite. He does kinda discard my definition of uncanny valley for one I disagreed with right at the start.
@@horacefairview5349 Yeah, he just kind of argued mindless points while trying to structure them in a way that would make it easy for people to side with him. Especially people like me that knew less on the subject. Luckily I know just enough about specifically the uncanny valley that I can say that his points were very much invalid and proved nothing but him having a misunderstanding of how the uncanny valley worked
Another problem is that most people treat the uncanny valley as something it isn't: a rule. The uncanny valley was a scientific observation of tendency. As things become closer to real, there is a point we start hyperscrutinizing the flaws, until those flaws are small enough that we don't actively pay attention to them anymore. This is the basic idea, but the simple phrase about things looking closer and closer to human without actually looking human, is the way that was chosen to to explain it in vulgate, because (exemplified by this very thing ironically) the average person is exceptionally bad at grasping the intended meaning of a statement, in contrast to the meanings of individual words chosen as part of the phrase. This is why you can call your friend an absolute bastard, and people who know you understand its a term of endearment, while people who don't, will ignore the fondness in your voice as you say it, and latch on to the words you chose. They aren't looking for context, only dumb surface level detail. On the other hand, to say a lot of these things tauted as looking close to human but not, to me, don't really look like humans at all. And on that point, I certainly agree with you that the observation is applied in a very hamfisted way. There are cases of real uncanny valley dissonance however, and they can even happen WITH real humans... I don't have an example I can give, because it's very subjective to what an individual's mind is looking for in the details.
It happens with real humans because of behaviour as well. Consider overthinking, pretentious, pseudo-intellectual people who keep being rejected and don't understand why. It is because they are not really behaving like a human, in the eyes of the majority of humanity.
The uncanny valley is not a scientific observation of tendency. The uncanny valley is a subjective, immature perception of facial appearance and was coined by only one person named Masahiro Mori.
As an artist its always been confusing, if you rotoscope a video of someone dancing you get stiff inhuman motion, even though it is literally traced from real life. On the other hand, if you animate it with slightly squishy and stretchy limbs, it feels perfect (as long as the emotion in the movement is there). The fix for it has always been, if you are going realistic, go 100% full realistic, no... go so realistic it looks more realistic than real life. Anything less and you HAVE to stylize it, simplify it (fortnite it), but not really...
I feel like you're too hung up on semantics. I haven't seen the uncanny valley taken as a literal, mathematical gradient, but rather a general explanation about a particular aspect of likeability. Your explanation in the latter half talks more about intent rather than effect, but your example of the dribbling basketball still shows that the robot that is closer to human characteristics but not perfect is less effective than either extreme. That would be the general thought behind the uncanny valley, because it's not trying to be a hard rule about all psychology related to likeability, but more of a simple explanation about how some particular things tend to be generally seen as creepy. Creepy, by the way, is as descriptive as uncanny, so if you use one you can use the other. That said I did enjoy the bit about the functionality of copying existing human characteristics vs designing with the mechanics of the functionality itself in mind.
But his whole video is about the semantics of uncanny valley, youre criticising the very meaning of his video. He fully recognises your points and dismantles it
@@GimmickBox39 Except he's arguing against the entire concept as a whole using purposely exaggerated examples. No one thinks that the graph is mathematical proof, it's just a visual to show the concept of the uncanny valley. He argues against the overall idea by ignoring the concept as a whole and simply grabbing random internet images that fit his argument. Not only that but his own arguments fit well within the concept of the uncanny valley. It's semantics, nothing else, there's no dismantling of anything here.
@@carcosian No, he's using the graphs to build upon the notion that there is no actual logical explanation for the uncanny valley, which is just another step in the overall path toward his point. He's not saying that "Robots don't look strange." He's saying that the _reason_ they do look strange is NOT, as the uncanny valley theory purports, due to being "similar but not quite right"; rather, it's due to it being "Similar but horrible." It's creepy because it's creepy, not because it's "not quite" realistic enough. It's _that_ specific notion that a lot of people have taken for granted as a fact, including myself. He's saying that there's no correlation between the level of realism of something and the level of creepiness, as you and I might have thought, but instead, the "creepy" examples are just examples of _badly done_ humanoids, specifically.
You're breaking things down so far that you're missing the forrest for the trees. Something almost human, but not quite, is unsettling. But it's just one of the MANY ways a thing can be unsettling. Not THE way, but A way. The uncanny valley is not supposed to be a natural law of "creepyness" that informs the the creep-level of all things creepy. It's a description of one SPECIFIC kind of creepyness. Something can look perfectly human and still have a creepy expression, and just as creepy, but different, is something that looks almost human but isn't.
I believe this video is putting forth that ‘almost humanness’ isn’t a creepy trait, and that the relationship we see at the right end of the chart has mainly arisen for other reasons related to the approach/intent. Things that are ‘almost human’ yet distinctly off, but not unsettling, include most of the morphs in the vid and many things in painting/portraiture/sculpture. I feel like what the uncanny valley phenomenon describes is that when you approach realism there are simply so many more things to screw up and it takes much more work to make sure they are all appealing. Bound to accidentally do SOMETHING a creepy way. In results and as a warning to artists I think this is the same as we’ve always perceived the uncanny valley, it would just be describing what’s creepy differently. Not exactly a sea change like the video makes it seem.
"Something can look perfectly human and still have a creepy expression, and just as creepy, but different, is something that looks almost human but isn't." well then it is totally wrong here is a real human who is creepier than any of thsoe creepy robots ;) chicago.suntimes.com/2018/11/16/18468730/metoo-founder-tarana-burke-blasts-the-movement-for-ignoring-poor-women On the other hand, anime characters are as far from humans beings as it logically can be and they even sometimes defy laws of 3d logic. but they are more attractive than any real human being. the top of attractiveness is a combination of real and anime. when you remove all unnecessary detail from real human and make it almost anime character or when you add enough detail to anime character just to make it look like a real human
The uncanny robots are missing crucial characteristics that both cute abstract robots and real humans manage to have. That's the whole point of the video: missing crucial characteristics is what makes the robots creepy, not exactly having too much likeness but not enough.
@@MutohMech To reaffirm your point; take the anime character being highly rated despite falling into an area where it should be "uncanny". It was rated the same "human likeness" as the monstrosity at the bottom, yet it was rated highly positively.
Was this something that needed deconstructing? To my knowledge the Valley is mainly a rhetorical shorthand for "This CGI is unpleasant to look at because it missed the mark". It's not like it's being taught as gospel anywhere or causing actual harm to anyone's artistic process 🤷♂️
@@DKannji are you implying that educational material is worthless and should not be created because a lot of people already know this? That's just wrong, and I wouldn't say "a minority" anyway, I'm under impression that most people have this misconception. I've seen people giving courses on both game design and painting teach their student this misconception. I've seen artists confounded in their creative endeavor because of the misconception. This is the reason why stuff like this is worth discussing. Just by scrolling through comments right here you may find many people for whom it was a new idea, and those are only ones who write it down.
@@stw7120 To put words in my mouth that I imply something that I didn't imply is disingenuous. Honestly curious of what people thought this way as I've seen a very small minority think it fact. Also, the people of the vocal minority are "teachers" and other propagators of such misinformation.
@@DKannji whole discussion's started by "Was this something that needed deconstructing?" - but sorry if I was too focused on Internet-culture habit of "taking sides". Thing is - people who discuss the Valley are usually only ones who gets over that "objective fact" thing, because it needs thought - nearly any source that introduces people to the Valley is at least implicitly presents it as a natural law. If teachers say it is - students believe it is. Habit of doubting what your teacher says is quite rare. As I see it supermajority of people who are aware of the Valley think that's a fact and a natural law, because it is presented in every simple, introductory material without giant disclaimer NOT A NATURAL FACT BUT A MODEL FOR EMERGENT PROPERTIES on them, and most people don't want and don't need to doubt other's research. Only small handful of people out of everyone who's aware of the concept actually understands how it works - but those are usually the same few people who have intelligent conversations about the topic - the "vocal minority", in their own right, not by being the loudest, but by being only source of quality you prefer. It is easy to take things as facts. And our brain is really, really likes easy things.
This seems to be a fairly elaborate straw man, that takes something a little like the original version of the uncanny valley and then ignores most of the further work done on the problem which developed better hypothesis and refined the concept to then say that the uncanny valley doesn't exist and we're really just looking at different aesthetic fields. The main problem I have with this is that you need to have a very selective view of the literature to have confidence in the conclusion that was drawn here; to illustrate consider the four working hypothesis in "The Uncanny Valley - Does it exist?" Brenton, Gillies, Ballin, Chatting, and consider which of these the studies used in the video actually prove or disprove. I think they're obviously still all on the table.
Also the bit about emotional expression is dismissing reactions to still images (most of what he talked about) based on their failure to mimic human motion despite them being compared to other still images
@@GimmickBox39 attacking a false perception of something and claiming it is that thing is the literal definition of a strawman argument. He didn't have to make the strawman to use it.
The elves look a little creepy, but mostly in the "strange adult" way. They're not cute or friendly-looking, which is off-putting for a Christmas movie. Everyone on the train is fine.
Title is woefully misleading. If anything, you've just explained why the Uncanny Valley is real, but people (specifically the ones making those silly graphs) don't seem to understand why or focus on an abstraction that is too broad to quantify in an exact manner, despite being an axis in said graphs.
You literally just explained why it shouldn't be called the "Uncanny Valley." There shouldn't be a valley; it should be seperated into multiple graphs because the abstraction of "human likeness" is too broad. The ideas of the uncanny valley aren't wrong, but graphing it as a "valley" is.
@@kirbykir Not really, no. You can see in the last study graph he showed that there are peaks and valleys, just not necessarily the "One" big valley, and you can even get uncannyness from other places too, but the concept of an Uncanny Valley is very much still valid and applicable, it's just not what people usually portray it to be.
@@HSnake5 The point is that people portray it as one big valley, the whole point of the valley is that they can just say "it's creepy via being in the dipping area of the valley, lolololol." "The uncanny valley" implicates that there's "One Big Valley(TM)" rather than a graph of more complicated concavity. The ideas of the uncanny valley aren't necessarily wrong, but it's like how people hear the term "big bang theory" and imediately extrapolate that it's an loud "start-up" explosion rather than a sudden expansion of spacetime.
@@HSnake5 so we agree that the "valley" part of the name is misleading?
4 года назад
I have 3 observations: 1. We are designing our animations to resemble babies: big heads, big eyes, rounded faces, etc. in order make them less creepier. Even if that style doesn't resemble adult humans, it makes them more trustworthy. We perceive the same with real adult human beings: we feel less threatened by adults with baby faces, as (maybe) you can appreciate in 7:20 where the last picture in the row is not the least threatening. 2. What about face masks? I have the impression that a person with a face mask is perceived as less trustworthy and more threatening precisely because of the face mask hiding human details from the face. I thought that was an uncanny effect. 3. What about facial warrior makeups... I mean paintings? Facial painting designed to scare enemies. How does it fit into all of this. Is it just art? I'm confused. The video was awesome and it really made me think... about how I have been mislead by trending labels. Shame.
1. That's not entirely true, we like characters with bigger eyes because it's easier to to read emotions on them. Eyes are the most expressive part of our face. It's easier to connect with people you can read because it feel safe, the big eyes play on that. That's also why clear colored eyes are more liked. For the big heads and other stuff it's just an aesthetic. Compare comic art to manga art, what is liked is totally different but if you take art from 20 years ago, they both drastically changed. It's like a trend, what is seen as cute or beautiful change all the time. You can observe this throughout history too. You can see the trends change or come back over the decades, it's really interesting. 2. We find things creepy when it can be associated with bad things like diseases, death, madness, suffering. If something look like a broken human, a burned human, a malformed human, your brain will reject it. 3. Makeup made to scare people is a prime example of why "uncanny" is wrong. Those makeups whole purpose is to create features people don't like by reproducing something not liked. It's not creepy or scary for no reason, it's created for this. Analyzing how they were made can give you a good idea on what features and trait the human brain don't like.
I think facial warrior paint is as much for the user, as it is against the enemy. Just as it is beautifully written into a story in Lord of the Flies, the moment you perceive yourself as different, you start to _feel_ different as well.
And the approaches of "making it look right by directly copying it" produces the linear graph and the "stylistically make an abstract representation" should be a fairly distinct distribution with its success somewhat independent of how closely it exactly resembles the literal thing. So I think it can be explained by the logic of the video.
No I think it's the same effect. At one end of the graph you have "symbolizes water." Resemblance to actual water not required, only that we know what it means. At the other hand you have "looks like water." Resemblance to actual water is the actual goal, and failing at it... just fails.
I haven't noticed this with water effects but I think this would be due to the same thing described in the video. An artist would animate water with the goal of having it pleasant to look at whereas someone who simply tries to replicate the physical properties while neglecting artistry (say a programmer with a knowledge of physics) isn't going to achieve the same level of visual appeal as an artist unless they get very very close to perfect. His point is that there is an uncanny valley if you only goal is accuracy but it doesn't have to be this way. If visual appeal is the goal you can improve realism without ever entering the valley.
the uncanny valley model is not a scientific model, simply a good rule of thumb for artists who struggle with the thin line between "creepy" and "cute". any artist who has tried on purpose to make something cute will know how frustratingly easy it is to make the cute thing creepy instead. this theory has more to do with a dichotomy of creepy and cute, using a theory that relates these qualities to human resemblance, than to a dichotomy of human and non-human.
2:31 you are using a visual detail (RED paint) to describe a difference in taste. Uncanny valley deal entirely in the visual aspect. Paint has a distinctly different taste from tomato sauce. Your analogy is closer to comparing the R.O.B. style robot to human form. The closest you could probably get using food would probably be diet sodas. Either way this was a bad analogy.
@@SolomonDragon Except, you literally reaffirm the point he was making with that analogy. "Your analogy is closer to comparing the R.O.B. style robot to human form." That is literally the point he's making. The point he's making is that we're comparing unrelated terms as if they're related, and drawing conclusions based on that. In his analogy, the fact the paint is red is the unrelated term. It doesn't matter that the paint is red, thus closer in aesthetic to traditional pizza sauce than perhaps garlic sauce. What determines whether or not the pizza will taste good isn't the aesthetic of red, it's the taste. He's stating that the uncanny valley concept ignores important information that determines how pleasant we will perceive a robot's aesthetics to be, and instead focuses directly on how closely the robot mimics the human form.
His analogy is only being used to explain a simple point: There's a bigger problem than the look of the pizza. There's also a bigger problem with the uncanny valley than the general notion that the robots are "really human, but not quite human enough."
So by cherry picking data to counter jokes around a shorthand version of a rule of thumb, you managed to validate its existence whilst asserting the opposite. Amusing.
@My Brand Haha, maybe you should pay attention when you watch the video next time. And also read up on what the Dunning-Kruger effect is so you can avoid exemplifying it.
@My Brand For me to "dig my grave deeper" I'd have to be wrong, though. Not that you've indicated any willingness to engage in good faith, so whatever salves your melodramatic soul. ¯\_(ツ)_/¯
this kinda just feels like. being hung up on semantics rather than actual useful criticism. ultimately uncanny valley is a shockingly good and easy, understandable way of portraying to designers _of all kinds_ where they might be going wrong. saying its just sex robots seems like its in incredibly bad faith. this philosphy is used in painting, game design, character design, sculpting, 3d modelling, literally ANY kind of art that cares to show humans (or any kind of animal for that matter. i think there is a throughline of cartoon dog- OH JESUS FUCKING CHRIST - well painted dog) they all have to follow this fundemental. if you want an example of the uncanny valley at work just look up portrait painting on instagram, your eyes will be opened.
Are you a designer? I can confirm that this uncanny valley thing is real nonsense, from my own experience. The mistake is that the human likeness axis mirrored in the end. the most attractive humans are not humans they are cartoons or other artificial products of heavy makeup or photoshopping. Real humans are pretty unattractive and comparable to those creepy humanoid robots. those people you see on magazine covers and on TV are not real humans. To make human or robot attractive you have to abstractive it into anime character so the real uncanny valley graph will be not a valley but hill somewhat resembling the gaussian curve
I believe his argument is that there's no valley, there's just two different upward slopes. The hill commonly seen on the left side is for faces that only trigger the ": )" part of the brain, while the other line triggers the "face recognition" part of the brain. They shouldn't be put together, because they're different philosophies.
I was SO anticipating a jumpscare with some creepy robot face you found, especially when the music went all weird at the end. I’m surprised and glad there was none.
So... wait.... is it wrong? Because by the end it just sounds like it's right but just with different definitions and grouping things via design goals.
@@MouseGoat the point is that human likeness is not the factor that causes creepyness. trying to recreate human likeness is the factor that causes creepyness.
@@knifetoseeya It must be a factor. The thing he can claim here is that random changes will not make it creepy, well thats true only specific changes that are on the eyes or expression will creep in and put our brain in confusion
@@MouseGoat "I am unconvinced", well to be convinced you have to think. He is refuting a central claim, not the existence of the ability to draw graphs. Yes, when you throw a bunch of points on a graph, you can get a value BUT (and this is the important so pay extra close attention) this is not caused by "increasing human likeness increases creepiness". This is caused by "two approaches are taken and if you do the stupid thing of averaging them then you get stupid results". Ultimately he's saying stupid analysis gives stupid conclusions. Averaging two completely different design techniques (stupid) gives the "uncanny valley rule" (stupid)
It's absolutely a General creepiness factor, plain and simple. It's not about morphing and finding the ceepy one in the progression. You are trying to define creepiness: the UV graph highlights that section where decidedly off-, dead- or vacant-looking representations of living things with faces are. People anthropormorphize, we are wired to do so. We don't need a graph to plot creepiness, but it kind of lends some intuition behind the idea that real humans and (well done) cartoons are both preferable to near-misses, or what looks amateurish, like an almost good portrait.
Interesting but I think your making the mistake of the Uncanny Valley trying to describe why it's happening when I feel that it has only ever described that it happens. i.e. There never needed to be only one reason as to why. It's a subjective generalization. but I bet your not the first person to take it as a definition as to why. So I assume you have seen some wackadoodle on youtube claiming they know everything on the subject and getting it way wrong. But still. though I think your premise is incorrect, I still think that video made some interesting points.
I disagree. Isn't the right side of your graph at 13:50 showing the essence of the uncanny valley? Faces that are looking very human are likeable, faces that look _almost_ human are not likeable. Thats exactly what your graph shows. About the not so human faces: Nobody claimed that abstract faces are automatically likeable. Obviously, something doesn't have to resemble a human to be creepy. To me, the uncanny valley only claims that it is very hard to make a almost human face look likeable and very easy to make it unlikeable. It doesn't claim that real human faces or abstract faces *have* to be non-creepy. Your deducting into the wrong direction.
Actually, they have. The one that leaps to mind is Scott McCloud's wonderful "Understanding Comics", in which he argues that the highest level of abstraction is the most broadly appealing, since it yields the greatest number of possible interpretations for the reader. Therefore, a happy face is more generally relatable than, say, the Mona Lisa. I'm not certain I buy that; I don't personally relate more to two dots and a line than I do the captured nuance of the human visage. I'm only noting that some folks in the business of drawing humans *have* made that claim before.
I think you're missing the point, but unintentionally stumbling onto it. He doesn't ever state that there isn't a point in human likeness where faces are unlikable. His point is that the way we compare the data is inherently flawed by oversimplifying what we're looking at. His point is that it isn't a simple sliding scale from "not human" to "human". There's a reason why, at the same point the anime robot scored incredibly high the horrifying monstrosity scored the lowest. Those two robots are solving fundamentally different problems, and one of them succeeded. The group on the right are specifically trying to make a robot look like a human. The further they are, the less human it looks. The group on the left are specifically trying to make a robot that creates a pleasant emotional response. It doesn't matter where on the line their "human likeness" falls, they're solving a different problem.
@My Brand the uncanny vally has to do with a robot looks and movement since robots dont move like a human does if you put a human face on it it becomes creepy. but if you dont and you make it have circle eyes and a happy mouth it becomes cute. its looks vs movement this can also be a thing in games when a human character looks human but the face doesn't move right. and you are not providing an argument like the others in this thread so no you are the one thats simple minded
I believe i can sum up the valley in one word "mimicry" We have a natural fear of thing's that are inhuman presenting themselves as human. Almost all cultures have demons or monster's that do exactly that because of that innate fear. So seeing anything and i mean anything regardless of design or intention that seems like it's something else masquerading as human triggers the brains protective systems and makes us say "something is wrong here" The uncanny valley is the effect not the rule of why it's creepy. Something mimicking us or even mimicking an animal is seen as predatory by the brain and the uncanny valley is the brain knowing something isn't right but not being able to find what
Yes! Also sincerity. We've evolved socially to notice imitations of emotion in others, like psychopaths and sociopaths. They're mentally quirked and dangerous. If the mouth shows teeth bit the eyes don't smile, it's a threat mask, not a smile. This is why characters like ET or Johnny 5 from the movies that show sensitivity and sincerity are characters that audiences can fall in love with. We can identify. Even dogs and cats that have human emotional range. But something can resemble a person very closely, and lack the emotional range or expression, and it will be rejected. Hence Zuckerberg
This is a solid refutation if somebody says "There is a mathematical function such that we can derive human affinity for a face [Y] from the realism with which that face imitates a human face [X]." It is a poor refutation of the almost universal view that very realistic, but not perfectly realistic, reproductions of the human face are unnerving.
Except that's not a universal view. It's opinion based and statistical, but the video clearly demonstrates that even the statistical model doesn't actually contain a valley, it contains pockets of unrelated design intent that have their own linear scales from less to more acceptable.
@@z-beeblebrox I don't know if you read that comment quite for what it was saying. The comment was agreeing that there isn't a truly graphable point where things get creepy but that the more common usage of the phrase as 'somebody trying to make something human and missing the mark and the result of being creepy is pretty damn common.'. in which case the valley is more of a metaphorical tryed to make it human and came up short sort of valley.
@@YayapLives I think the common usage and view is "You start out with something abstract and likable, but as you try to make it more human-like theres the uncanny valley where it's weird". At least thats how I've always seen it used. The point in this video - as far as I see it - is that there's no valley created by an attempt to make an abstract thing more humanlike. It's just that there are either human-like things or abstract things with human features. And the abstract things try to be cute and likeable. Putting them on a graph is then what creates the valley, rather than it being an actual consequence of trying to create something more humanlike.
Yeah, the third, fourth and fifth ones are fairly creepy, while the other ones are perfectly fine. The other examples also cut away a huge part of the face.
For me most of them starts in the valley, so to say you dont see them go down... yeah no shit. But the third somehow went from one creepy to another creepy. still dosen disprove the valleys existens, its not weird to me the brian freaks out when something is close to human but not really, but we do have different tolerance levels.
There's something very appealing about a good graph. I'd heard a lot of people claim the uncanny valley doesn't exist but then never present a decent substitute theory for how we feel or an explanation of what is really going on. Thank you for making this.
Basically the uncanny valley is an erroneous result of looking at the data in fewer dimensions than it truly should be analysed in. "Human likeness" is too vague to be measured by only one dimension like the uncanny valley is normally explained on. The video isn't really arguing against the data from any research on the topic but instead the method of statistical analysis used to explain the data, particularly that there are two distinct categories that aren't directly related
True, but remember the fallacy fallacy: just because an argument has one or more fallacies doesn't mean the idea itself is wrong. Like when scientists in the 18-1900s tried making a model of the atom. It wasn't accurate, but it didn't mean atoms didn't exist.
This all just reminds me of folklore stores about people encountering demons, fairies, djinn, etc who just have something off about them that tips them off that they're not human and so they know to run away and not be tricked. It also makes me think of the descriptions of the men in black from Mothman Prophecies, where they all had something not quite right about them.
No discussion of the Uncanny Valley should be attempted without bringing clowns, and how peoples reaction to them differs, into the discussion. If you don't understand that, you will never understand the Uncanny Valley. And yes, the same thing goes for the injured, disfigured, and humans wearing facial prosthetic make up. Our reaction to those people is part of the Uncanny Valley effect. Also, if we can read the emotions of cats and cartoons, the emotional connection is not the issue. Again, check back into why some people are creeped out by clowns and others are not. That is where you will find the answers to the Uncanny Valley
@@wisemage0 I differ in opinion here because prosthetic facial features are noteworthy and tend to be unappealing in terms of sexual attraction for reproductive purposes. I mean look at someone who had a bad facelift. It's not mean it's just truthful. And nobody here is trying to justify discriminating against someone who's been in a fire. That's just cruel. But we're saying objectively looking at a bunch of AI and human features, at what point does the uncanny valley theory pan out?
What about something that looks perfectly human until it moves? Pre CGI horror film puppets had that problem. They looked great on the front of Fangoria until you saw it thrashing about in the film. Same for some realistic looking CG characters like Senua from Hellblade. I found it a lot of CG characters if you cover the eyes the skin and hair looks realistic but the eyes seem to have no soul. Is that because of the animation?
That is entirely animation. So much stuff that you don't actively notice is hidden in real peoples' facial movements. It's hard to replicate that, even if the textures and models look perfectly lifelike.
umm naa, i got: I can't see it so it must be unreal, also please ignore this proof of it and look instead on this made up disprove that don't really relate to the subject, and also may actually also prove that its real.
The converse of your assertion that "it's not the (almost) likeness; it's the creepiness" is that it wouldn't matter how non-like a human a thing with random changes is, it should be equally creepy to the thing that's randomly changed from a human. The uncanny valley effect is not "oh now brains can't take it." It's that these artificial changes on something too close to looking human stop representing abstract ideas about things, and start representing deformities. Yes, a lot of the examples resemble "creepy humans," but some are just alien. Take Alita from the live-action movie. She was the only character given anime-like eyes, and it IS unsettling. She is otherwise a friendly, pretty girl, with very human expressions even in and around her eyes. But the eyes, which are "cute" in a more abstracted representation (e.g. a cartoon), are wrong on her because she looks too human to have such features be healthy. Your "hair color makes the witch unattractive" thing argues from a truth: that it's not the hair color. But it fails to justify why anybody would claim it's the hair color. The "more specific" reasons why the uncanny valley looks so creepy tend to be all related to the flaws that make it inhuman. That's the point. Whether those flaws are from a human losing his humanity or from an artificial form getting enough right to trigger "that's not right for a human" as a reaction, the 'near human'-ness of it is important. Even your "but highly refined cartoon" argument fails because the "highly refined cartoon" would look creepy if you started making NON-RANDOM changes to make it look more human. I applaud your attempt to plaster over the gap that represents the Uncanny Valley with a life-like statue, but that is not more human-like than the highly-refined cartoon. As evidence, compare to equally life-like wax sculptures that are painted meticulously to look like a living person, which almost inevitably do fall into uncanny valley. "The likeness increased at a steady rate" in the version with the ken-doll like faces in the middle, "but there wasn't any sort of valley effect." Actually, there is; the poor real human got rated more eerie than things less human than him. :P Even your explanation of "here, people are trying to recreate human life, but over there, they're not, so it's totally unrelated" is you projecting an explanation based on your assumption that the uncanny valley is NOT real, so therefore there must be differences in how these were designed that explains it. I'm sorry, but the uncanny valley effect is absolutely real. You fail utterly to do the one thing that would actually prove otherwise: fill in the gap with something between the inhuman peak side and the human peak side of the valley that is at least as non-creepy as the inhuman-peak side.
I think I now agree with both him _and_ with you. The uncanny valley exists, but it's not a matter of "almost human but not quite"; rather, it's "looks human but deformed."
11:26 - Simple and beautiful piano music accompanying a charming lady getting self conscious and smiling bashfully, hard-cut to 11:34 - Cold lab noises accompanying abyssal body horror writhing in total mockery to all that is good in the world.
Not sure if I missed the point of the video but literally, they look creepy bc they seem almost human but something just isn't right, and that's what the uncanny valley is..
4:02 the middle one. That one. That's the one that makes me feel off. The ones on the left are funny and the ones on the right look like people but that center one just feels off
Man, you just nailed down exactly what was that vague feeling I always had about those "uncanny valley" explanations 😆 It always felt weird to me because I watch a lot of semi-realistic CGI, and I know it doesn't *automatically* trigger that feeling. Only hastily-made or amateurish CGI does that. This is going to be my go-to explanation video from now on 😁
You have completely changed my mind and thoroughly explained not only why I was wrong but also why I used to believe it. And you did it in under 20 minutes. That is an incredible feat.
The flying machine example is one of the best things i have ever heard : we're not ready yet to make fully functioning bird-bot , and a plane could do the job 1000 times better ...
@@davidegaruti2582 The point for militaries to research bird-bots is to make plane detection more difficult. If the enemy radar or whatever cannot tell birds from planes, they cannot effectively shoot down planes with out shooting down all birds (almost impossible and very bad for the eco-system).
@@y.z.6517 Yeah my points still stands : airplanes and helicopters have showned themselves to be extremely efficient flying machines capable of being versatile and effective in many circumstances and sizes , meanwhile bird bots are being developed just as a situational tool to trick radars , Wich might soon get outdone when stealth tecnology goes far enough ...
@@davidegaruti2582 As far as stealth goes, it's like armor versus missile. Stealth gets better at one point, and then radar improves, and better stealth is being researched. However, at some point, there will be little stealth can further improve, whereas radar will just have better and better resolution.
As somebody who went to Graduate School in the US for fields of Psychology, I commend you for taking the same academic approach to this that I was trained in grad school to think about any psychological theory. You look into a "well established concept" that "everybody knows is true", break it down, really analyze it to see what problems it has, find what parts of it might actually work, and come up with a very plausible theory that explains some human biases, such as cherry picking data or using incredibly vague definitions to define the idea to begin with. And part of your theory, about how part of this effect may be because of how some people choose to design the AI and not how it naturally is, can be summed up as "correlation is not causation". And while a lot of people know that phrase, APPLYING it is really hard. This is a really neat video and I'm glad you made it. I agree with you; after seeing your argument, I no longer think uncanny valley is a real thing, but there is SOME element of truth to it.
I understand the points the video is making and I agree with most of them, but I still end with belief the uncanny valley exist. In the last graph the "irobot" face is exactly 50% human likeness and 0 likeabililty. In my mind it serves as an important baseline. It was not created to be emotive. It is devoid of character or personality. It is not trying to represent and individual but rather and idealized/generalization of humanity as a species. Yet it scores as more likeable then 75% of robots trying to appear more human. It takes the top robotics in the world with a cost of perhaps +100,000 dollars to surpase it.
It seems to me the difference between uncanny and not isn't a matter of "almost human" but a matter of "bad art." In other words, they're creepy because they're creepy.
So, he just explained how the uncanny valley works and... never really refuted it. Idk man. I haven't disliked a video in a long time, but this video is just semantically beating around the strawman. /:
I really enjoyed your outro -- saying that you could be wrong or this could be just part of the story. I feel like maybe there's a kernel of truth in the uncanny valley we haven't uncovered, but also the graphs and ways people have tried to find "proof" are a bit contrived and overstated too. I appreciate you challenging the dominant narrative and providing an alternate perspective while still having epistemic humility.
I think there is another factor not mentioned in the video. We have expectations based around abstraction, or in other words, we have different standards for cartoony things compared to realistic things. I can draw a circle with a friendly face and it looks "cute" but if I add more details like if I give it realistic features, it becomes an unsettling contrast. Ren and Stimpy uses this for gross-out humor, for example. The same works in reverse, where if something is realistic but combines cartoony elements, it becomes an unsettling contrast. It's perfectly fine to have plastic eyes on a cartoony toy but plastic eyes look wrong on a realistic robot. Its not an axis of likeability to realism, it's that the level of realism for each feature of the robot must be consistent with all of its other features. Humans are drawn to notice contrasting elements, like realistic human hair paired with unrealistic rubber skin.
Excellent. This was an intriguing subject & a stimulating discussion. ~ ~ ~ As an artist, I can confidently state that clearly expressing subtle emotions through art (sculpture/oil painting) is excruciatingly difficult & can come down to remarkably miniscule tweaks in contours, shading & even colour tones. I've had clients tell me paintings of loved ones seemed too angry, or kind'a sad (despite them being very close to the exemplar photographs in execution) & fixed them by making tweaks to the shapes of the eyes & mouth, or the overall colour palette of the painting. Sometimes features are even ever so slightly exaggerated so as to elicit the wanted emotional response (not unlike how stage actors may exaggerate postures or gestures) & this is where training & experience really shows the difference between a talented amateur & an expert professional artist. Accurately representing the complex character of an individual in a still image or sculpture is at the pinnacle of artistry - a place where very, very few artists ever reach in their lifetimes... The human face has both miniscule & macro muscle movements - some lasting only microseconds, yet are brains are capable of perceiving all of these movements. Just look at yourself in the mirror & try to see how many variations of a smile that you can come up with: • Smile for a baby. • Smile at a girl on your first date. • Smile at the -bastard- coworker who just got the promotion you were expecting. • Smile in response to a stranger's smile. ...There must be thousands of increasingly subtle variations & most will include muscle movements in other regions than just your mouth. How many times have you seen an insincere smile, where their mouth moves appropriately, but the muscles around their eyes do not? Some of the most obvious "Uncanny Valley" facial features which I've personally seen, have been in 3D animations, where despite enormous effort there's a lack of bridging micro-movements - causing a disturbing disconnect between the intended emotion & the projected emotion. Looking at some of the greatest classical 2D animators from over the past century, you can see how they've managed to express emotion in a extraordinarily accurate way, despite only using subtle movements of lines. The similarity to the human face is sufficient for our neural facial recognition centers to see them as a "real" face, & they're not so dissimilar as to cause other more pragmatic neural centers to reject them. This takes a special kind of genius and years of training & experience to achieve on a consistent basis. ~ ~ ~ I suspect that with increasingly sophisticated AI analytics the "Uncanny Valley" will be bridged in a very short time, perhaps in less than a decade - both in animation & in robotics. And in the process of developing androids/gynoids with ultra-realistic facial features, we will learn important facts about our own brains, minds & psychology - some that may have applications (implications) in medical science & elsewhere...
I would say that we are just missing an axis, that the "uncanny" effect is a three-dimensional effect. Let me explain with video game graphics where (I'd argue) a similar effect exists: Take game graphics on a 3d Graph: X: Stylised-Realism; Y: "Technical Quality"; Z: how "good-looking" it is Stylised games (see Nintendo) have an easier time to look good than games trying to emulate the real world. For example, Mario Odyssey looks stunning but compared to next-gen trailers it is lacking in technical quality (I don't say it low, the effort of effects is absurd in Mario Odyssey, and it's a miracle they got it to run on the switch hardware) but (In my opinion) looks better than the Halo demo. The Halo demo just falls behind in visual appeal due to problems with LOD, while alone with the lighting has a better graphics quality than any Nintendo Game. [Im sure the final game will look awesome, it looks like they just didn't have the time to optimise the demo, but needed to get it to a presentable framerate, the main problems where just LOD and a too aggressive culling algorithm]. At least in general More Stylised games look better with the same technical effort than games which try to look realism or maybe a better way to say is, styalised graphics are more forgiving. And looking at how games have aged it seems that with technical improvements games on the realism side are judged harsher than stylised games. (see Half-Life 2 and Paper Mario TTYD, I'd say Paper Mario has aged better) The thing with "human-likeness" is we don't have a way to measure the "quality", to split the x-axis into two separate values. Especially on the more Human-Like side (with the linear relationship), it seems more like a difference of "quality" resulted in higher Likeability.
I'd argue that appeal beings to go down with too much style and detail. You've seen that "realistic Mario" image, right? I'd say that there is still a balance between style and detail.
I think specifically when it comes to stylized vs realism the biggest limiting factor is the literal capabilities of the technology. There's a ceiling for how realistic you can make something while making it look good for each technology, and it starts to look bad once you start trading the look for realism, or at least it looks bad in hindsight a lot of the time if you make the most realistic thing of your time people will just be stunned by the realism.
It's been a while since I've started questioning the Uncanny Valley deep inside, though I kept spreading its idea and taking it for granted. Thank you for this video!
Watching this video just made me believe in it more... Obviously it's not impossible to make an aesthetic near-human model. It's just hard enough that it should usually be avoided.
Thank you for this video! It's awesome, and I'm glad to see you back making videos again. I think it really is a lesson in critical, scientific, and statistic thinking. It makes me want to be a statistician or data scientist/machine learning programmer/AI programmer (a.k.a. sort of a statistician lol). It also made me wonder if your job requires you to interact with a lot of statistics / data and correlation finding. Anyway, thanks a lot!
The pizza with red paint already got me. There is food out there you can make without tomato sauce. Don't mess up your dish with something that clearly is no tomato sauce just because you don't know (yet) how to make that stuff.
Best Uncanny Valley video I've seen. 7:18 Uncanny Valley is not about how realistic it is. Uncanny Valley is about reminders of death, disease, corpses, and untrustworthiness. The eyes are the most important and have the biggest effect. If your eyes are bleeding, you will be more alarmed than if it's your mouth, nose, or skin. So you will also be more alarmed if you see someone elses eyes looking diseased than their mouth, nose, or skin. To make something uncanny valley make it look dead, diseased, or untrustworthy. Corpses and very unhealthy people have sallow clammy pale or grey skin, bulging eyes, dead looking eyes, and may look emaciated or bloated. To make something untrustworthy make it look less like a baby, so less cute and less neotenous. A smile without the eyes smiling or with droopy lower eyelids or with agreesive looking eyebrows is especially uncanny, because we automatically sense this as an unfriendly and insincere smile, so it looks very untrustworthy. Women are more neotenous than men. This is why makers of androids rely heavily on neotenous designs for trustworthiness, so they rely heavily on robots that look cute or that look female.
Thanks for making this. It has been the only video I have been able to find that was critical of Uncanny Valley, because the internet just loves the concept for some reason, despite it relying on unquantifiable criteria, and being easily disprovable using examples. Creepy things are creepy, but even there, a lot of this is framing/highlighting, like the liminal spaces fad. Some of the pictures for that are so unremarkable that it makes me question that concept as well.
...honestly, this video seems to me like it's just being kind of pedantic about definitions. You cannot deny that there is a general tendency for robots to be less likeable before they become more likeable again, and you even propose a difference between parts of our brain that do basic facial recognition and take in detail, which is like, the entire point. The uncanny valley happens when the detail-oriented part of your brain kicks in but you haven't put in the effort to get it right, because it's much harder to make something accurate than to make something with a face. It's just a rephrasing of the whole "inhuman with a few human features" thing (i.e. the face) versus "human with a few inhuman features" (i.e. the detail-oriented part of your brain finds things wrong with the "human" it's trying to parse). So really, the uncanny valley graph as it's ordinarily presented seems to go from "inhuman detail" to "human detail", not from "lack of detail" to "lots of detail". Reality isn't suitable for stick figures, since it's not a blank canvas. We're working with actual materials in a 3D space, so we can't just "remove" all unwanted details, unless we work with very specific materials. There'll always be a certain amount of detail. We cannot perfectly control how 3D objects appear, unlike the way we can manipulate 2D images. The problem with the face morph is similar: it displays an idealised process that we can never realistically go through, progressively replacing all the wrong details with correct ones. I also very much question the way you break up the robots into groups. It seems _highly_ arbitrary, and yet it perfectly lines up with your conclusions. It kind of seems like you're working backwards with that one, and as such I can only dismiss that graph as irrelevant. How do you know who is and is not aiming for perfect human accuracy? Why do you even lump the "imperfect accuracy" robots in with the ones on the left rather than on the right, when for both groups, it's the goal to create something that looks human. And that's also true for some of the robots on the left side along the bottom. It kind of seems like you should include those ones with the group on the right simply based on the importance they place on human likeness. But if you do that, guess what shape the trend line of the group on the right takes on? That's right, it's the uncanny valley again. You dismissed the other graphs for being based on cherry picking, but isn't that exactly what you're doing here? In fact, the study you sourced that graph from DOES conclude there is an uncanny valley, so you're entirely misrepresenting the study, just superimposing your own ideas about the uncanny valley on top of it. I'm not a fan. Of course it's not like I've got any data to really back up what I've said here. It's a hypothesis based on the questions I found myself asking over the course of watching this video. But at the same time, I don't think _you_ have really sufficiently proven your point either.
Pedantic for saying the uncanny valley, is not because of human likeness, as is said by every supporter of it's existence, but because two different types of machines, with one being the sole reason for why there is a "valley". "It seems highly arbitrary" "for both groups, it's the goal to create something that looks human" "simply based on the importance they place on human likeness" not even the same color as any human skin, not having half of the characteristics of a face, sometimes just lighting up eyes and a mouth, *single eyeball robot* ...are you shitting me
The problem is that you're basically arranging things by what the intent seems to be and then assessing them as steps on a gradient when they're not. The thing that looks weird is being placed there because it looks weird and anything that isn't as human like but still looks really weird is thrown out. That's where the cherrypicking comes in.
@@nathanlevesque7812 If you ignore everything that looks uncanny that doesn't strive for precise realism then you're cherrypicking your sources. You are ignoring samples like, say, a skinless robot with eyes and teeth that might look more uncanny than something that looks more human and doll-like despite it being less human in appearance. How people percieve things is complicated. There's more to it than just whether or not something is approaching human realism.
@@nathanlevesque7812 That's why this is cherrypicking, you can't really ignore the things that don't fit into your ruling or you're going to come to a conclusion that doesn't fit with reality. It's why these graphs never really show much of a range in the gradient, because there really isn't one. It's mostly just individual cases of things not working between two very different approaches. The situation where something's only three-quarters human and only kind-of uncanny isn't really as much of a thing, so you just end up with a pattern that doesn't really exist. If this feels like it's not cherrypicking it's because it did its job and cherrypicked examples that feel loose enough that you can draw that conclusion while ignoring everything else that didn't really fit with the idea.
It's great you've uploaded! You've been one of my most favorite content creators on this site and I always try my best to share this video with all of my friends on social media. How your channel doesn't have a million subscribers is beyond me. Keep doing what you're doing, Jesse. I'll always be a huge supporter of your channel and I will continue sharing as you continue to grow.
I was more under the impression that the Uncanny Valley referred more to "The sweet-spot between familiarity and unfamiliarity", rather than simply "human likeness". Although, I do understand your point and why you focused on this topic of "human likeness". You bring up many good points and dismantle the issue well.
I just subscribed to this channel. 50 % because of the content of this video and a title of a recommended video on the right and 50 % because the tone of voice the narrator's using.
I think the subgenres of body horror and cosmic horror can tell us a lot about why we are averse to things that are not quite human (and also things like teeth on pliers). Perhaps its the way we identify people that are diseased, or people that will act harmfully in illogical ways, and we react in a way that makes us take notice and not want to interact.
First off, I love your presentation and I can follow your reasoning. But after mulling it over I feel like your issue with the uncanny valley is more of a semantic issue. See, uncanny doesn't aim to replace "it looks creepy", it aims to describe why. as you said, do they look like uncanny humans, or do they look like creepy humans. The answer is yes, both. The item looks like a creepy human but not because they aim to replicate a creepy human, but because our failure to replicate a human leads to items that look more like a disfigured human.
But there shouldn't be a valley. If anything there a graph of an increasing relationship between "number of errors" and "uncanniness," not a valley. Leave the cartoonish stuff out, it shouldn't belong in the same graph.
That study at 6:00 is kind of dumb... that isn't really what the uncanny valley is. That is, it isn't making an artificial human more lifelike. It's just taking 2 images overlapping each other and increasing the opacity. That isn't a real-world example. Also - a bronze statue of a human doesn't contain the near-realism of a human that a robot or CGI character would, so I wouldn't say that a bronze statue qualifies as existing in the same realm of realism. However, a wax figure does. Why? Because bronze statues don't represent real people, but wax figures do. Uncanny valley also does exist when it comes to characteristics of real people - particularly when it comes to mental disorders. Someone who is clearly mentally handicapped (such as non-verbal Autistics) would be seen at the left edge of the spectrum. Neurotypicals recognize they have severe cognitive impairments, and so they don't expect neurotypical behavior out of those individuals. On the other hand, someone with high-functioning Asperger's (like myself), is constantly expected by society to act normal (because that's how I appear). People engage with me as if I'm a normal person, I try my best to fight through the anxiety and have a normal conversation, but after a couple minutes my aesthetic comes off as unsettling and "off" to most people. So, I'm almost normal, but not quite. Hence, I exist an a sort of "uncanny valley." A lot of people with Asperger's report feeling as if they exist on this plane. This phenomenon also exists with people who have severe PTSD (the thousand-mile stare) and sociopaths. There are lots of real people who come across as "unsettling." The problem with throwing a funny-looking guy on that chart is that it's conflating humanistic realism with just being "funny looking." You can be funny looking and STILL appear as an authentic human. So, the "cherry-picked" chart is really comparing apples and oranges in a way.