"There is no uncanny valley any more," declares French developer

[Nappi]

The movement of the flesh on the edge of the eye on the nose side looks weird somehow.

Yes, I'm picky, but they were the ones who claimed they have passed the valley. I would think it's little details like that make all the difference. I wonder why they didn't show the whole face....

[Wrestlevania]

Mmm, rubbery!

The movement of the flesh on the edge of the eye on the nose side looks weird somehow.

Yes, I'm picky, but they were the ones who claimed they have passed the valley. I would think it's little details like that make all the difference. I wonder why they didn't show the whole face....

No, no - that's the whole point of the uncanny valley. The closer to a human being something animatronic appears, the more anything unusual about its nature becomes accentuated. You're quite right (in my opinion) -- I'm not even remotely convinced, regardless of how good their animation system is.

[Thrik]

There's no way these guys have even hit the uncanny valley, let alone crossed it. I do, however, think it'll be one very sweetly presented game. I really hope noone's expecting true post-uncanny material though, because you're inevitably going to be disappointed. I genuinely believe it's impossible on this generation of consoles.

[vimes]

It doesn't really compare well to what the guys at Crytek

achieved

[Wrestlevania]

I'd say you're right.

The most striking aspects of the CryTEK human characters would appear to be the lighting and the texturing. As stills, they're photorealistic.

There are still some illusion-breaking artifacts when the character's facial features are animated (the "fleshy eyelids" slide for example). But that final comp shot is, well... wow.

[Thrik]

The Crysis models are OK, but they're really not that impressive in motion. Crysis' eye animation is clearly inferior to that of the clip posted earlier, and the mouths look downright poor.

I'd say the closest contender to the Heavy Rain (or whatever) clip is still Source, which sports superb facial animation in Episode Two that looks more convincing and lifelike than Crysis' in my opinion.

[brkl]

Too much Botox...

They say they've passed the valley, but if you look at the actor with the tracking stuff on, there's nowhere near enough precision to track subtle changes of expression.

[Thrik]

Of course they haven't. Anyone who believes they could have is mental.

I do think a bodyful of detail matching that seen in the eye video could be pretty superb in an actual game, though. You're typically not going to be close enough to spot the tiniest nuances of movement, but with it all happening at once I'm guessing the end result will be startlingly lifelike.

Just not uncanny valley.

[JoannaDark]

Where's Joanna, isn't she meant to be doing a thesis on this?

hello!

Jo is currently looking at hands and the uncanny effect. hands are important. Jo is also submitting to APGV in august in LA - wahoo!

its so weird. gamers have a completely different take on the valley than do scientists. we'd rarely use highly photoreal models, because were not using facial closeups at the minute. the discussion of photorealism in games is, in my opinion, getting a little too carried away with the notion. As in, i do some stuff on perceptual equivalence of real and virtual models, and there are some neuroscience papers that show that different neural areas are activated when looking at real and not-quite-real humans. Getting to the stage where we can incite the same neural subsystems using virtual and real STILLS is going to be a challenge, never mind movies, and extrapolating that to a maybe 30 hour game where the virtual is interacting with other virtuals and its environment is going to take some time. one of our colleagues in carnagie mellon thinks its to do with fear of death and that itll never be crossed. i think it will, but considering all the work it takes to accurately light one patch of skin....i think it'll be a while.

that's just me though.

[vimes]

Now, http://www.image-metrics.com/]THIS is awesome.

[Marek]

Impressive! Though I still pay most attention the (fairly hot) actress on the left. Halfway through I was like... hang on, I'm supposed to watch the 3D face. Then I saw a 3D woman who has Gollum living inside her.

[brkl]

That's better, and they do that without shiny blobs on the woman's face.

[Roderick]

Is that real-time? Or unedited input, or whatever? In that case, it's amazing.

Also, I am in agreeance (yes) with Jo about what she said just then.

[Thrik]

I wouldn't imagine it's real-time. That's probably the kind of implementation you'd expect to see in a CG movie.

I'm not really impressed by motion capture stuff so much. That's kind of what movies have been working on for years, and games are inevitably going to be significantly further behind on the raw polygon and shader side of things.

I'm guessing Quantic Dream are aiming to have real-time stuff that isn't completely motion captured. For example, their eye video looks like it could be the product of a player simply looking around like you might in any FPS. While you could probably match that with motion capture, it's nowhere near as impressive to me as doing it procedurally.

It's kind of like how Half-Life 2's facial system is all done procedurally, based on the sound files it's given. They could probably come up with some great stuff if it were all motion captured, but then they'd have to do that for every voice sample they use instead of just having one system can process any.

Granted, their first Heavy Rain video kind of clashes with this assumption seeing as that was motion captured up the arse.

[Wrestlevania]

http://cubo.cc/

[brkl]

You've had enough caffeine, Cubo.

[n0wak]

http://cubo.cc/

I was just going to post that. I think that's considerably different. It's mostly photo manipulation which is clearly not quite the same thing as CGI.

[Wrestlevania]

It's still trying to replicate a human appearance though -- the method's arguably irrelevant.

[vimes]

The Uncanny Valley being so discussed in the industry right now, I wonder why no major studio came out with an idea that actually exploits this limitation ... to make a point and fill the void while the solution is being figured out. From the top of my head, it could be used to highlight the un-humanly aspect of certain people by representing them with photo realistic visual while giving compelling characters a 'puppet' look that would convey better emotion, etc...

Thinking about it - and to my knowledge - no major limitation or malfunction of the visual part of the medium has ever been taken advantage of : wouldn't it be nice if designers could use streaming that takes too long and makes LOD very noticeable, the absence of depth of field/shadow or incoherent shadow or impossible visual composition... to actually build a underlying discourse or shape an atmosphere ?

[LOPcagney]

Yeah, the most important thing in convincing your audience something is real is in making it move like a real thing. Photo realistic paintings aren't nearly as evocative of a human being as well-puppeteered marionettes.

[Nachimir]

The Uncanny Valley being so discussed in the industry right now, I wonder why no major studio came out with an idea that actually exploits this limitation ... to make a point and fill the void while the solution is being figured out. From the top of my head, it could be used to highlight the un-humanly aspect of certain people by representing them with photo realistic visual while giving compelling characters a 'puppet' look that would convey better emotion, etc...

I think because reactions to it vary so much and people are adaptive.

There's little that's guaranteed to get the same reaction from everyone, and horror games already go for it in spades. Anything more subtle people tend to get accustomed to and just adapt it out, i.e. once I unconsciously know something isn't real, the negative reaction to uncanny visuals stops.

Thinking about it - and to my knowledge - no major limitation or malfunction of the visual part of the medium has ever been taken advantage of : wouldn't it be nice if designers could use streaming that takes too long and makes LOD very noticeable, the absence of depth of field/shadow or incoherent shadow or impossible visual composition... to actually build a underlying discourse or shape an atmosphere ?

That's an interesting idea; it of reminds me of Passage, which uses a lot of 8bit-like limitations in an expressive way (I recommend playing it before reading the creator's statement).

[vimes]

IThat's an interesting idea; it of reminds me of Passage, which uses a lot of 8bit-like limitations in an expressive way (I recommend playing it before reading the creator's statement).

Wow, I don't think I can thank you enough for making me discover this game!

This is spot on what I've been searching for a while; namely, 'gameplay' ideas which bear a symbolic (or as you, say, expressive) meaning that the player gets better as he plays ... it perfectly use the feedback loop, action-effect that only interactive medium display. And that's rare, right ?

Plus there's no tutorial - which is awesome- no loading screen, the guy stripped it down to the most essential thing -throwing away any Video game convention - and yet, it is subtle, interesting, but not elitist...Again, wow!

[Roderick]

One example of a game doing this is err, I don't know the name, but it's this game set in the second world war in Paris and you're playing some underground whatever. The interesting bit: Nazi occupied areas in the city are greyscale, and when you deal with it the colour returns. It's a little more blunt and in-your-face than what you were proposing, but it's still a good idea.

You could have a game about someone who's mind is unravelling, and is this happens textures fall away, bumpmapping vanishes, motion blur and anti-aliasing go, polygon count dwindles... you could really make a nice arty high concept with that. But it's a one-time thing, obviously. It's the surprise that makes it worth it, like the Eternal Darkness insanity effects. It'd get boring if lots of games would do that.

[vimes]

It's the surprise that makes it worth it, like the Eternal Darkness insanity effects. It'd get boring if lots of games would do that.

I don't know...

I totally get your point because I noticed it too in lots of games.

But, you see, I'm not completely OK with it because it sounds likes: 'Killing people from a first person perspective. Nice idea, but a one time thing : you'd get bored'.

Of course, there are two pathological cases (1) make 'use limitations' as your gameplay drive and bet on the small novelties (2) make one limitation a core gameplay feature and leave at that. In both case, you end up with a gimmick and it would work across several games.

However, I believe that if you consider limitations as a mirror to new technologies, then they become another tool, whose use must be challenged by how it is needed by the game : you don't need to make the gameplay to revolve around it, you just have to use it when the situation calls for it. Like split screen in movies or Quick Time Event.

[miffy495]

That's an interesting idea; it of reminds me of Passage, which uses a lot of 8bit-like limitations in an expressive way (I recommend playing it before reading the creator's statement).

Oh my god that was brilliant.

I was so shocked when I realized they were aging, and then when the girl died it caught me completely off guard and actually made me feel really sad.

So powerful for something so understated. Thank you for posting that one.