L.A. Noire's Gag Reel Bridges the Uncanny Valley

[Cognimancer]

L.A. Noire's Gag Reel Bridges the Uncanny Valley


You don't have to be a detective to find the humor in these motion-captured outtakes.

Back in 2011, Team Bondi's L.A. Noire turned heads with its dialogue-driven gameplay and, more notably, its cutting edge face-scanning technology. As impressive as the end result was, the process did revolve around human actors, and actors make mistakes. The interesting bit is that the motion capture software recorded all those slip-ups, and the in-game models mimicked them. Ladies and gentlemen, we're looking at the first fully-organic blooper reel in a video game.

Depth Analysis, the company that provided the game's facial animation technology, has released a gag reel on their site to show off "the level of performance and humanity captured by the MotionScan process." The video certainly delivers on that claim - watching a digital character sneeze is more than a little surreal.

In an ordinary live-action production, these bloopers wouldn't be much to laugh at - there's the typical assortment of botched lines and inappropriate chuckles. What's truly bizarre is how much of the life behind these goof-ups is captured by the animated characters. When a suspect screws up her line, the detectives share a laugh at her mistake. An opening remark gets flubbed, and the characters actually walk off the in-game set so they can do another take. The interactions between the character models may still fall within the uncanny valley, but we're at least on the uphill slope.

Source: Depth Analysis [http://depthanalysis.com/]

Permalink

 

[FrostyCoolSlug]

It's funny, when playing LA Noire, I really didn't think too much of the facial expressions, sure they looked pretty real and impressive, but it kinda felt, I dunno.. scripted?

Seeing the realness of unscripted events in the technology really does a much better job of demonstrating it's capabilities, and gives a new perspective on all that I've seen before, a fantastic job.

 

[Casual Shinji]

Damn, it feels weird hearing that soundbooth background laughter when the characters are standing outside.

 

[jollybarracuda]

I feel like they could have released something like this before the game came out, and the game would have gotten a hell of a lot more attention than it sadly did.

The tech behind this game is just so damn impressive. I hope we see a hell of a lot more of that kind of advancement in this next generation of consoles.

 

[Remus]

The eyebrow at 1:58 is hypnotic. Oh and it's going by the same tune that the girl in the next clip was singing. How often were they sitting around jamming to classic rock that it got stuck in the heads of 2 of the actors?

 

[Casual Shinji]

Seriously am I the only one who wasn't impressed with the facial expressions in this game? Yes, TECHNOLOGICALLY impressive but uncanny as hell.

The facial expressions are impressive, but that makes the plasticine looking CGI bodies stand out even more. It seriously looks like they're wearing rubber suits.

 

[Shinsei-J]

Those were hilarious.

Bridges the uncanny valley? No it doesn't. Quite the opposite. Friggin' plastic muppets came alive.

Now what would make a great game would be LA Noire with the all the main characters being muppets.
I would play that everyday before I go to sleep at night.

 

[BrotherRool]

I actually think it was pretty good at bridging the valley, most of the time they looked like real enough things that I didn't think of them as poor imitations of humans. Some of the extras looked a bit uncanny and every now and then you'd get a facial expression (particular one that interacts with the hair or edge of the body) that looks forced.

It makes me wish LA Noire was a better game. They had lots of good idea and lots of bad ideas in equal amounts. (seriously, as filk hulk critic pointed out, whose idea was it to show how the murder was done in the teaser? Isn#t part of the fun supposed to be figuring that out?)

 

[Metalrocks]

lol, that was funny. im hoping tomb raider will have the same.

 

[Treblaine]

I realise what this game was missing for me... it was too by the numbers, these moments of breaking character are more warm and relatable than all the melodrama of LA Confidential riffing.

You need a bit of humour, a bit of levity. People need to actually laugh. Or else they're robots.

I mean why couldn't the character have sneezed right there? Did that really break character? Did he seem less human because he sneezed at an inconvenient moment.

 

[BrotherRool]

First time I've actually paid attention to emotions on characters in videogames really.

whose idea was it to show how the murder was done in the teaser?

Wasn't it that the footage of the murder being done in the intro of the mission is never how it really turns out?

There were still things like the head being used to prop the car breaks etc. And they force you to solve a lot of crimes long after you know that you're banging up the wrong suspects.

 

[CarlsonAndPeeters]

Does anyone else not really believe that these clips are automatically generated? I mean, L.A. Noire came out almost two years ago. It wouldn't be hard at all to animate these scenes manually (combined with the facial capture data). Things like the characters walking off the screen and the tongues sticking out seem like they could be done in post. Still cool tech, but I feel the same way about it now that I felt when playing L.A. Noire; it looks cool, but its not really moving anything forward.

 

[Lilani]

Seriously am I the only one who wasn't impressed with the facial expressions in this game? Yes, TECHNOLOGICALLY impressive but uncanny as hell.

The facial expressions are impressive, but that makes the plasticine looking CGI bodies stand out even more. It seriously looks like they're wearing rubber suits.

Yeah, I think the bodies and textures are the biggest things holding them back. The lack of cloth movement is really noticeable. Also, and I don't know if they instructed the actors to stand that way or what, but sometimes they are standing there with their arms out to the side like they're trying to impersonate a penguin.

Also, I wonder at what frame rate the mocap software was recording them at. Because people can see about about 60 frames per second, most movies are recorded at 24 fps (48 at the highest), and when a game is running optimally it should get 60. If the mocap software is capturing less than 60 then that means it has to make up the missing frames with automated tweening, which can affect the trueness of the movement. Also, 60 fps is pretty brutal on details. Any movement or twinge that isn't quite right will be much more visible at 60 fps than 24.

I think this is a huge step in the right direction, and yes the technology is on an uphill climb, but I'm afraid until game engines are able to master cloth a bit better (or more devs find a way to work around it with the style) real-time cloth movement is going to be one of the biggest hurdles games have to jump to achieve photorealism.

In other news, damn those tongues. I think those impressed me the most. Just...DAMN.

 

[Casual Shinji]

Also, and I don't know if they instructed the actors to stand that way or what, but sometimes they are standing there with their arms out to the side like they're trying to impersonate a penguin.

You see this a lot in motion capture. I think actors are either instructed to do this so that the infrared cameras can detect the dots more easily, or because the dots restrict them from having their limbs close to their body. Or maybe both.

 

[Proverbial Jon]

I always thought that each actor was recorded separately and then the lines were edited together in the final game. But the way the other actors are actually reacting to the slip-ups obviously means they did the scenes together. I didn't realise that.

Plus, how are they "walking off scene" after each take? I thought the bodies were animated separately.