Uncanny Valley/Cuteness in relation to pixel-art

[Cyangmou]

The deeper I get into realism and the observation of humans the more and more I ask myself how the concepts of the „uncanny valley“ and the concept of „cuteness“ apply to pixel art and in which ways they efectively affect the styles and the outcome of our pixel art.

By the limits pixel art gives us as a medium, we are also limited on what we achieve.
There are limits we an overcome, like skill problems, then there are problems we can't overcome (restrictions in terms of colors, size, animation frames) and then there are some elegant „solutions“ how to imply something in a problem we can't overcome, that it makes sense to us and feels at least ok.

I definitely think those problems heavily define the way how our pixel art looks and how it's crafted.
Some people say pixel art realism is nearly an impossible thing to achieve and indeed it's hard. Especially since we have to fight the problems of the uncanny valley with big sharp dots.


There is a lot to observe if we get really deep into those things and a lot to say, but to give a short kick-off those videos are a good source to get a basic understanding and maybe to start a discussion on the subjects:

Uncanny Valley:
https://www.youtube.com/watch?v=CNdAIPoh8a4

Cuteness:
https://www.youtube.com/watch?v=3Ji0bvwXAvI


some further thoughts for the interested ones and some theories/explanations and ideas:

Another thing which hit my mind is, that „cuteness“ might be quite a modern approach to things, especially inspired by Japanese culture. At least I don't know of any artwork before the second worldwar which just relly clearly emphasizes the baby schemata as much as modern media like videogames and anime do where some productions just build on the „daaaaaaw-factor“.

I don't want to discuss how wars affect the society and how the Second Worldwar affected Japanese people, that's just a side-note.
However, a lot of the pixel-art defining games were made in Japan (not only, and there is other stuff as well, but it's kinda a majority) and a lot of those used the „chibi“ approach which just emphasizes cuteness.

The big point for me is that this cuteness also seems to heavily correlate to a non-scene audience and a lot of people who start out with pixel art just copying those „established styles“ they know of the roots of popular and successful game series.

The chibi approach definitely has some advantages like:
it's easy to craft something decent, it's effective, it's established, nice readable on lower resolutions and so on,     
makes it attractive to a lot of people

Some of those people try to add more detail to those things, but by adding just a tad more realism and structure to the „chibi art“ we face the „muscle-baby“ problem, Helm brought up 6 years ago
http://wayofthepixel.net/index.php?topic=7301.msg85004#msg8500k4

And there are also quite a lot of other notable distortions which make it look „creepy“, which I also think is directly related to the problem of the uncanny valley.

I really wonder why realism has such a strong impact on us.
A lot of people love realism, triple-A studios love realism and hardware designers love realism too. And I am also loving realism (although I also appreciate other approaches to art and I am not sure why I exactly love it).
Nintendo gives a shit about realism and focus on their style (basically they throw out tons of principles) and some of their 2D games really look timeless and now better as their concurrence back then (which are now mainly perceived as „cheap“  looking 3D realistic styled games).

Spoiler

Zelda Wind Waker (2002)


Burnout 3 (2003)

But now there is the „Sword&Sworcery-approach" to pixel art as well, which features „spagetthi people“ and a much higher amount of realism as chibi-art does.
This style creates empathy through completely different tools and also is an attractive direction to go.





And it seems to get another „established“ style directions, as a lot of indie games are using it by now, popular examples are Irkalla or Terra Incognita.

Terra Incognita:
http://www.curious-expedition.com/

Irkalla:
http://forums.tigsource.com/index.php?topic=35320.0


And this style seems to be equally effective as the early featured chibi style and this might be why it finds so much audience, since it really enriches the pool of what one can easily achieve with pixel art.

[RAV]

Reminds me of the differences between Western and Asian recognition of facial expression, where we place more importance on the mouth region, whereas they more into eye/brow region.

Something I noticed about the Sword&Sorcery style is how it creates more distance to the character, as if to allow it privacy. This concession makes it more real and genuine to our Western taste, than showing up close what often looks pretentious in its stylized expression. Also words like mysterious and melancholic come to mind. And it tickles the imagination of the viewer about the emotion, with a freedom of interpretation. By minimizing the face further from what would have been too simplified for subtlety still, it doesn't get in the way of telling by other means, similar to leading the eye, leading the mind. And we may notice more, the world itself has character, things a soul, the scene says. A more practical and immediate advantage is that of animation; these graphics usually impress most through their fluid and complex movement, of which they have a great many, to draw their life from, realistically achievable in the gritty reality of development. The greater range of gesture with the bigger limbs may outweigh in expression what you can do with that bigger chibi-face,

Though these strengths be mentioned without absolute preference. Depends on overall goal and design of project.

[ErekT]

I can't remember a time ever when I saw pixel art that tipped over into uncanny. But that goes for all hand-drawn or painted stuff I've seen. Don't think it's an issue at all for pixel art really. The reason 3D graphics suffer from it is because the computer does half the drawing for you. It's in a sense mechanical. Movement is mechanical (frame-tweening) and surface rendering is mechanical, with all its complex algorithms to simulate surface shading. It's the same reason those robot thingies look freaky while a stone sculpture doesn't imo.

[RAV]

The uncanny valley occurs mostly when the model looks close to real, but the animation looks unnatural in a subtle way. And this is very much the reason you don't see this in classic pixel art:
It's almost too much work already to pixel a static character portrait in all its uber realistic detail... but who then animates that shit? The problem is not so bad if it's not realistic animation.
By frame, that is. Of course the newer fancy assisted-animation methods, like tweening, actually can introduce this problem, since they allow for detailed base material to animate with (warp/stretch)
But since most of that content is rather stylized comic/manga fantasy stuff, it doesn't occur often there either.
It helps too, to be selective in what of it to animate, close-up human face animation is most prone to the problem.

[Cyangmou]

I don't think that the uncanny valley actually only appears in animated stuff. Or at least, I refer to the "uncanny valley" more as "feels odd".
Of course it's really really subtle for realistic stuff, the fact is that it's there.
It's also there in lots of paintings, lots of other art. Maybe not the "uncanny valley" as it is shown with robotics, but the "odd feeling" definitely arises
 
I suppose we have the same problem there as if we are asking a person without understanding of art what looks off in a somewhat decent piece, and the person says "umm it looks odd, but i dunno why", but artists with an understanding actually can put the problems in words.
Means if we aren't well enough trained to see what's off, it's impossible for us to point at it.

If we suppose that every style is just a simplification of realism were we leave out details and squash and stretch proportions and alter colors with a fixed scheme, it would definitely be interesting to know how the odd feeling comes up.

Muscle babies are definitely uncanny. There are some other areas what also looks uncanny, but mostly it's related to characters.

Fact is that it just appears in relation to characters, we feel an empathy for.
I suppose that's because our perception is from nature just really well trained to see the slightest nuances of facial expressions.
Means if a tree or a stone actually don't looks like a tree or a stone should look, we won't recognize it, since we don't have an emphaty for stuff like this.

But actually the problem of "uncanny" or "ugly" also can be related to some TV-shows. I actually know some people who say the Simpsons are butt-ugly. On the other hand I know people who dislike to look at overly cute anime stuff. I think the same principle of empathy there takes place in the "design of a style".

And well, I suppose the "uncanny valley" (robotics wise) addresses the core problem very directly and effectively why we like or dislike styles.
It's just an assumption on my end, though.

So yeah, I agree with all of your first post RAV.

@ErekT: stone sculptures can look creepy too. I think it's the same as with bad painted portraits. You see that something is off. You know that's a painting, and you see that it's a painting, but the feeling actually should be about the same as with the robots (it's just not that strong).

[RAV]

I know uncanny valley as a specific kind of creepy stuff, "real" people that move like puppets, or remind us that in the end we're just flesh puppets. But, hrrm, yeah, fuck definition hick hack, I know what you mean, and that creepy stuff is a larger problem worth talking about in any variation, we might as well just call it all uncanny valley, the term makes sense in more than one way.

I do remember some "realistic" pixel portraits, especially those trying to stick close to photo reference, that looked unintentionally odd or creepy.
If they weren't traced they had many little anatomical problems, and if they were, it was something about the translated colour / shading.

One thing I heard is, that if you make something that's close to what people have seen a lot in real life, it's especially bad if it's off from that. For this reason, modeling/animating fantastical stuff like an Orc face, or a Dragon walk, you can get away with a lot more of "sloppy mistakes", that's basically you guessing how it would work, since no one knows how else it had to be.

So it makes me wonder if seeing the photo references to those pixel portraits actually worsened the uncanny effect.

[Mr. Fahrenheit]

I'm wondering if andylittle's portrait of Woody Harrelson sparked this topic? It definitely gives me the heebie jeebies. When I look at it more closely the things I notice that look off to me are the sharpness of the wrinkles in his forehead, the lack of any real shadow on his head from the hat, the slight out of perspective look to the hat, the "stringy" method of shading the artist uses, and then the lack of a real overall shape to his face.

I'd say its a good example of a pixel portrait that is in the uncanny valley.

[RAV]

But you know what else is interesting to discuss? Not only how to avoid it, but nasty ways of intentionally instrumentalizing it, for horror stuff.

Maybe a greek stone statue is not uncanny for everyone, but once you've seen a horror movie have one turn head and look at you, you're fucked up for life whenever you see one.

So it goes both ways, I guess: real life makes "wrong" art uncanny, but art can make life uncanny / "just wrong", or other art.

And what about clowns...

[Cyangmou]

yeah definition hickhack is shit. And yes, the uncanny valley definitely is a very specific term. However I dunno if there is a term for "feels odd" so it seemed that uncanny valley is the closest one.

Actually I think portraits are the easiest thing to point out several things which might be off.
However there are some other things I'd love to talk about later on, if we get an actual discussion going.

And of course, the effec tis strongest for what we see. Drawing hands and heads actually is so hard, because we see them a really long time every day. Cars are for example also really hard to draw and yeah, I think it's connected.

@Mr.F
No andylittles portrait might fall a lot into it, however I wanted to start the topic earlier.
I actually wanted to talk more about game styles than about "portraits" (that's why the cuteness factor is equally important for the topic), and portraits are a good way to find out why stuff don't works. I looked into them a lot lately, but I dunno how much knowledge and interest for tha topic is here.  but maybe by pointing out stuff in portraits and making things clear we can also try to draw parallels to game stuff later on.

You are right that portrait like a lot of others would be a good subject, however I don't even know if we may use it as subject for an art discussion.

@RAV
indeed, using the effect intentionally for a certain purpose is of course also a possibility. Horror, which I am a big fan of does it a lot and very effectively too.

[RAV]

It's so funny you mention cars. I was just about to say, that I've seen even objects like cars that felt uncanny the same way, so it's not just living things.