Developers to Cross ‘Uncanny Valley’ Within Two Years

At least, according to Obsidian’s David Kunkler

Posted by Staff
No, that's nothing like a real person!
No, that's nothing like a real person!
Obsidian Entertainment producer, David Kunkler, claims that super-realistic facial expressions in games that don’t suffer from the ‘uncanny valley’ problem will be with us in less than two years.

"Give us another year or two, and we'll be able to completely get across that uncanny valley," Kunkler told BBC World Service's Digital Planet programme.

"People are doing motion capture very well, getting the 'exactly right' facial expressions - eyes moving correctly, every little crease, wrinkle and nod - really coming across," Kunkler added.

The ‘uncanny valley’ effect refers to in-game facial animations that, as Kunkler describes it, “are too close to real, but not quite real.”

Opinion is divided on the issue within the development community. For example, in our recent interview with Bethesda’s VP of marketing, Pete Hines, SPOnG was told that, ”We've gotten that uncanny valley question before and I don't think it applies, only because we're not even remotely close to something being so realistic that by one aspect being slightly off it's disturbing. I still think we have a long way to go in terms of lip-synching, facial animations, expressions and all that stuff, but you know…. they're definitely getting better, but there's light-years to go where we are going to be approaching realism, where you have to worry about it being just that little bit off to be disturbing.”

Let us know what you think? Is ‘the end of the uncanny valley’ mere developer/PR hype or genuine near-future possibility?

† A phrase coined by Japanese roboticist Doctor Masahiro Mori. Stated simply, the idea is that if one were to plot emotional response against similarity to human appearance and movement, the curve is not a sure, steady upward trend. Instead, there is a peak shortly before one reaches a completely human “look” . . . but then a deep chasm (the ‘uncanny valley’) plunges below neutrality into a strongly negative response before rebounding to a second peak where resemblance to humanity is complete.
A more detailed explanation can be found at Arclight

Companies:

Comments

OptimusP 20 Feb 2007 15:41
1/4
I believe HL2 is the only game that doesn't suffer from a negative uncanny valley and the reason is simple: subtlety.
You can do all the motion capture you want, but if you don't blend those animations togheter well enough it will look like a robot going from one facial routine...go back to neutral face for no reason...run second facial routine.

Like that Tiger Woods facial demo EA showed on Sony's E3 2006 show. It did one facial animation and then froze and went to a neutral face.

It needs to flow fluid into each other and no amount of technology will fix that. Animation expercience does. Nintendo managed better (more life-like) facial expressions with TP then those in Oblivion by far. Just because Nintendo has much superior animators.
RiseFromYourGrave 20 Feb 2007 17:11
2/4
like the other dude said, games like Half Life 2 have done spiffing work in recreating the human 'spirit' in its character's animations. with brilliant animators, in 2 years we could have some sweet stuff, sure. but id say plastic skin and dead eyes are the thing that that give CGI humans away, and id be very suprised if they get that right soon, and be able to run the software in a game
more comments below our sponsor's message
actionmonkey 20 Feb 2007 21:45
3/4
Half life 2 was good but nowhere near realistic. The acting was pretty good and the facial animation the best yet but they didn't match the wildly gesticulating bodies which threw the whole thing off and made it look like a school play.

As far as I'm concerned only the work Weta has done on King Kong and Golem have come anywhere close to realism and they took a lot of people a large amount of time and also had the benefit of not being human. This isn't going to happen in two years.

OptimusP 21 Feb 2007 12:45
4/4
Weta used a lot of tricks to make everything look as real as possible, the biggest CG scenes in LOTR are always from a very zoomed out position so you can't really see the faces or expressions combined with very good motion capturering and using real world enviroments (all the environments the CG-models run on were crafted in real life as mini-envoriments)

In other instances they put the CG-models close to the real people/animals which tricks the human brain in saying all of the units as real. Like the Rohirim charge in movie three, all the horses you see being shot down are actually cg-horses, quite hard to spot (also because the horses are incredibly well done) but when you pay some attention to it you can see the differences. But because those CG-horses are running with real horses you're brain registers all the horses as real because that makes the most structural sense.

Technology won't close the uncanny valley, using sublte tricks and a buttload of animation expercience does it...if you're not really paying attention.
Posting of new comments is now locked for this page.