We're fascinated with the battle against uncanny valley—that point in graphics or robotics where the human likeness isn't quite reached, which in turn repulses the viewer. And while we've seen graphics card makers and robot designers tackling the problem before, for the first time a video game developer claims to have bridged the gap. Quantic Dream is a maker of highly cinematic games and houses one of the most advanced motion capture studios in the world. Here's what they've gone on record to say:
But today, I can officially announce that there is no uncanny valley any more, not in real-time.The last we've seen of Quantic Dream's upcoming PS3 game Heavy Rain was back at E3 06. Their non-interactive tech demo (screenshot above) was amazing for its time, and the word I received through the grapevine was that they had just weeks to get it done. But this is a BIG claim. [gamesindustry via kotaku]









Comments
i watched that demo...it's not real time...and it's not crossed "uncanny valley"
very impressive though.
I remember this demo. Quite impressive at the time. I would like to see what they have done today.
This demo's been around for a while. I'm sure he was referring to newer work. I would love to see it in a game ASAP.
@EMoShunz: Yeah, the dev actually references the E3 demo and how it fell short in that linked interview.
That is still creepy.
This would be amazing, if true.
This could also be big news for Sony if Quantic Dream makes a follow-up statement to the effect of "... but only on the PS3." We know that the PS3 is a powerhouse relative to the 360 (very hard to program for, but powerful nonetheless). It may be that for this generation of consoles "Non-Uncanny Games" will be reserved for the PS3. If that's the case, the PS3 would have a huge advantage in attracting exclusive titles with A-List actors - and the gamers and profits would follow.
that picture is still incredibly creepy. perhaps because it's a pixel woman, or perhaps because you don't really expect cute but dead eyed girls to be pointing guns at you. if it had been a pixelated Rambo or Bushwick Bill, then maybe i could believe it was real.
yeah, that picture hasn't crossed uncanny valley. and its got nothing to do with the fact that she's dead-eyed and pointing a gun at me.
@Brock: Exactly. People don't realize that the PS3 processor is very new and hard to create for because studios haven't had access to it. Sony has since opened more up so developers can get to work. I may be wrong but I don't see how this could benefit the 360. If it does, we'll be seeing which one really has the best graphics once and for all.
I still remember how Sony promised us a Toy Story level of graphic complexity and crispness, in real-time, for the PlayStation2. Forgetting for a moment that was 8 years ago, and neglecting entirely that Shadow of the Colossus and God of War are downright beautiful, please forgive my skepticism.
Why's Sarah Silverman so sad?
Is she sad because she's shooting people? that usually makes her happy.
@omg-ponies: lmao
Heavy rain...
Anyone remember "Chubby Rain" that excellent Bowfinger flick with Kit Ramsey and that famous final line "Gotcha suckas!"
Priceless.
the woman in the photo looks like a plastic mannequin with unfocused pupils. The haven't crossed anything.
I doubt this claim, the uncanny valley is deep, dark and subtle...
The human face and form is immensely complex, and it doesn't help that it's the single most intimately familiar / recognizeable object to other people. I don't doubt we could push enough polygons to render the shape correctly, but to make it come alive and get all intimate and innate cues to bring it too life is vastly beyond what game artists are capable of yet. Beowolf was still in the uncanny valley and the tools they used are leap years beyond what can be rendered in real time.
No human created human; animation, sculpture, robot, whatever has ever successfully crossed the uncanny valley.
I just started playing Indigo Prophecy (found an XBox copy for $10 at Fry's)...
It's pretty good. I'm definitely curious to see what this studio can do.
@Argyle: I'm sure glad you got that out. And your expertise in this area is...?
i'd prefer my games to look like beowulf rather than no country for old men. I like knowing i'm playing a vv good looking game, not forsaking reality
If they can make Beowulf in real-time...
... but seriously, it's all about the cinematics; not graphics.
@w00zzy:
we all look at a lot of faces. is that not enough? especially ones with blurry man hands in the foreground?
@Emoshunz
@White
@Mark 2000
Re-read the blurb or the article, that image is from E32006, and is NOT the output from the new tech they say they have developed.
1)Read
2)Post
So many people seem to skip step 1.
If they have passed said valley, it would only be for the video game visual realm. Any claim to more realistic robotics that actually function in real time still chills me and creeps me far far out.
*shivers
@w00zzy:
While harder to develop for there isn't some massive potential to be unlocked, hidden from developers. The very things that make it hard to develop for prevent it from being a powerhouse. That doesn't address the root of the problem either way, which has little to do with processing.
The Beowolf comparison bears repeating. Even a fully pre rendered 100% controllable animated film regularly entered into moments of uncanny valley, and was most successful when it avoided mimicking movements 100% and instead stuck with more animated movements. Uncanny valley isn't about polygons or lighting, most of it comes down to animation. Unfortunately most animation issues can't just be solved with more processing. While I am sure they have gotten close, and that many of their cut scenes could cross in and out of that boundary I sincerely doubt they have solved anything related to uncanny valley.
Yah this doesn't really make any sense. Uncanny valley theory doesn't really apply to things seen on the television, does it? Just in created things seen in real life - humanoid robots. I mean, there have been plenty of human-like and humanoid cartoons and animations, and none of them affect anyone in any particular way. Certainly aren't creepy. They're television.
But whatever. Mentioning the words "uncanny valley" gets you about a hundred thousand nerd-hits on the internet, so why not do it?
Uncanny valley applies to images on TV or robots in person. Remember when they were bragging how long the '100 Neos' fight in Matrix Reloaded took to render and how real it looked? I can only imagine the reaction when they first saw the finished sequence after days of rendering... Rubber Neo!
@juriko: I agree with you...most of it comes down to animation...AND the transition of one animation to the next. Indigo Prophecy did this pretty well...so I somewhat believe QD for making that bold claim of crossing uncanny valley...but I am not gonna drink the Kool-Aid until I see the game footage.
@farcast: But the unnerving quality that is supposed to be the hallmark of u.v. doesn't exist with the television, and never has.
@Pope John Peeps II: and a little check into the wiki reveals that the theory was first made by a roboticist, talking only about reacting to humanoid robots, but initially invented by Freud, who was talking only about automatons in fairy tales.
Applying this theory to 3-d- animation is bullshit, because the initial unsettling condition never really existed.
"Uncanny valley" has at least as much to do with the way objects move, as they do to their static visual appearance.
Even if you perfectly reproduce the appearance of somebody's face, then model the underlying musculature perfectly, to produce visually perfect movement...
That still gives you the look of a very wooden, uncomfortable actor. It lacks the TIMING of the movements of the different facial muscles, some of which are very subtle, that goes into the formation of an expression, and an emotional projection.
It's a heavy task to create a face that just makes a proper "idiot grin", and "exaggerated anger" of a well sculpted mannikin... yet they refer to stuff like that as "crossing the uncanny valley".
-
They haven't even started on creating facial MOVEMENT and TIMING that CONVINCINGLY projects real anger, sadness, melancholy, happiness, satisfaction, deceptive smile that's not obviously sinister, etc., etc.
How can they model it when most actors can't even break it down themselves? They just do it naturally.
Movies are full of BAD real-life actors... yet CGI has yet to map performances as good as unwatchably BAD junior high play actors, let alone a CGI DeNiro.
You guys are all arguing stupid ish!
Who cares if it has to do with robotics or video games. This was a PR move designed to get you to talk about their upcoming product. We all know that Sarah Silverman up there doesn't look real ...they didn't cross anything ...but they made you talk about it.
Pixar cartoonizes their characters to avoid the uncanny valley. They talk about it on the bonus content disk of "The Incredibles"
I think that exumed Orville Redenbacher ad is a great example of the hazards of a near miss in GCI:
www.youtube.com/watch?v=Fcn4p213Zg8
@Argyle: What he said. The human brain is too darned good at processing extremely subtle and subliminal visual cues.
I think what these guys are missing in their research is the "subliminal" (which is hard to figure out since it is, of course, subliminal :-). Without that level of not quite visual information, almost anyone will notice that there is something "wrong" and creepy. And gamers in particular will know it's a fake... but we'll play anyway... why do we really care?
Err.. CGI.
Anyway, watch the ad because it's amusingly bad in it's own right (uncanny valley or not)
Er, CGI. Anyway, you should watch the ad because it's amusingly bad in its own right.
The best example of that right now would be the film "Beowulf".
That was rendered off line with 10,000's of processors.
Then each frame was hand tweaked to look as good as possible.
Even with all that effort you can, for the most part, tell it is CG.
AUughghghahhghahhghghggh. *sputter*
No, no. I remember when people were screeching about the uncanny valley back when "Spirits Within" came out. To me, that still looked like a cartoon, a caricature.
This, however, is the beginning of the true Uncanny Valley. When developers try oh so hard to push realtime graphics toward realism. It's getting too damn close.
I say back off for a generation and stick with the caricatures, they're more lovable than actual human faces anyway. Aren't there graduate theses that say people can become more attached to emoticons than highly realistic faces?
@Pope John Peeps II:
gaming media adopted the term long ago when games first started trying to hit realism. Where the term started makes no difference, it is used when speaking about games to talk about a very specific problem with rendering realistic human figures. It doesn't refer as much to complete unease as a general sense of "wrong". This almost always comes down to animation, particularly facial animation, and flesh tones due to a lack of subsurface scattering. Other things tend not to have the same effect as long as they are consistent, but because of the way our brain interprets human figures when they start to look more human they come across as less accurate due to the above flaws and others.
It is the same reason artists can get way with poorly drawing animals or even a lot of scenery, but poorly drawn human figures jump out.
@pr3isdentspence: how about you read my post. i didn't say the video was the new tech, i said 'that' which was referring to the posted image.
1) read
2) post your 4$$h0l3 remarks.
@EMoShunz:
You said, "i watched that demo...it's not real time...and it's not crossed 'uncanny valley'
very impressive though."
Let's break that down:
"i watched that demo" = what it reads as
"it's not real time" is not relevant to discussion
and the important bit:
"and it's not crossed 'uncanny valley'
very impressive though." = arguing against the claim WHICH WAS MADE FOR THE NEW TECHNOLOGY not the old one.
Your sentence (and I use that word loosely, since the only punctuation you seem to know is the elipsis) had an implicit reference to the NEW technology. So, you said, "I watched that demo and it doesn't stand up to the claims they are making."
IF you were talking about the old tech and understood that the picture/demo and the article were two different things (which I'm not sure I believe you did), then don't get mad at me for your writing an ambiguous post.
Sincerely,
4$$h0l3
All this talk and no one has yet cried out for a make-your-own-porn application? I'm serious. Anybody that could market a "build characters that look like anyone you want and control the way they get freaky" simulator that reached Beowulf quality would have a license to print money.
I perceive a lack of imagination around here. :-)
@EMoShunz:
Holy crap dude, you've made 34 comments in 6 hours.
I now understand why you don't have the time to read the articles or write unambiguous comments.
Now, are you at work?
Love,
4$$hol3^2
@pr3isdentspence: is there a demo for the new tech? no. so how could i be referring to watching the new tech? i can't be.
the demo i was referring to is less than 2 years old, so it is still a fair comparison anyway, even if i was making a statement about the companies abilities it would still fit in context (along with all the other comments about beowulf).
i will continue to write ambiguous posts until gawker pays me to be a journalist, which i have no interest in doing since i am not a trained writer. apparently you are though. i come here for info and fun, and most people are respectful, but there are a few...(blank spot for ambiguity)
@pr3isdentspence: i'm sure you couldn't find any ponies that strider along this site making ambiguous posts for fun...from work if that matters to you. don't like it, find a job you can type and talk at the same time.
@EMoShunz:
Your logic is impecible:
"is there a demo for the new tech? no. so how could i be referring to watching the new tech? i can't be."
is the same as:
"What I said was factually impossible, therefore that can't be what I said." (And if you follow that to it's logical progression your numbered necklace will starting blinking until your circuits burn out.)
Now what I was saying is that I don't think you read enough of the article, or understood it well enough, to understand that there was an old tech and a new tech.
By the way, you've got a lot of nerve talking about respect when you called me a four-four-dollar-hol-three, C0:CK:$U:CK:3R.
@EMoShunz: "the demo i was referring to is less than 2 years old, so it is still a fair comparison"
Not really. In 2 years, they could have made significant strides in making the game more realistic (2 years is a LOT of time in the dev world). I agree with PR3ISDENTSPENCE that your remark was half-cocked at best.
@pr3isdentspence: wow, let me apologize, i really didn't think you would take it to heart that much, my opinion really isn't that important (if it was i wouldn't be commenting on gizmodo). that is not sarcasm, it appears i truly offended you, for that i am sorry. can't expect more than that when you make attacking statements though, can you? that said, your "is the same as" comment is not really the same, but i love the blinking necklace conundrum reference.