L.A Noire: The technology behind the performance

[Eren Ozkural]

Anybody else think that this is where filmmaking is heading?

CLICK HERE

 

[Richard Foster]

No, but it is a good fit for the gamers.

 

[Robert Frank]

I may be missing something but in this application it seems like reinventing the wheel.

The technology is impressive but I'm not sure I see the advantage to doing motion capture of the performance and then a separate facial capture for the dialogue, when you could just do this all as blue/green screen and composite later? Is it because then you have options for the characters performance in the game that only need to be created with data rather than large video files?


I can see using mocap for creating animation of other types of characters. I nearly purchased a Moven rig (that's a whole other story) to build a project around it but decided on traditional compositing, until the project fell apart.

But I'm still interested in uses for mocap.

 

[Bruce Allen]

Yes but face and mocap needs to be captured at the same time.

It's REALLY HARD for an actor to recreate an emotional scene when they were moving around... if they're sitting on a frikkin' chair pretending for the facial pass.

I'm more interested in having say 40 Scarlets recording a scene at 3K, 120fps. Some getting the wide scene, others focused on the face, getting it from various angles, etc.

Then you feed those streams into FoundryPfTrackUltraMatchVoxelCaptureThing. And you'd have a 3D reconstruction of the whole scene.

This would match up nicely if you had say an autostereoscopic display. Autostereo sucks right now because films only record 2 views. So you are restricted to a very narrow viewing angle - if you move your head to the left, things get messed up quickly. What we need is if you move your head to the left, your left eye should see a new "FAR LEFT" camera's view and the right eye sees the old LEFT camera's view. Such a system could enable this.

Main problem is capturing enough data and having a 3D scene reconstruction software good enough to deal with transparency, etc.

We need an order of magnitude more data and an order of magnitude better software to process such data.

10 years?

Bruce Allen
www.boacinema.com

 

[Kwan Khan]

It's very streamlined. Geometry, colour, plus body orientation and voice, and it's all recorded in one go.

 

[Ace]

Quite amazing. Go Team Bondi!

I have to say, I'm quite proud that Aussies are consistently leading the way in the advancements of storytelling.

 

[Bruce Allen]

It's very streamlined. Geometry, colour, plus body orientation and voice, and it's all recorded in one go.

Did watch the Youtube video? It shows them on a stage first doing all of the whole body movements, wearing costumes with markers - eg a Vicon-like setup. Then the actors go back and film the facial stuff in their stage, while sitting down.

Which is why the more frantic stuff seems a lot more fake than the stuff of actors talking calmly - because the heads were recorded separate to the physical actions.

This makes it really hard for the actor - because physical action is connected to emotion which is connected to performance. Obviously for technical reasons. But this is a huge problem. I think the Avatar way is better until they make the next tech leap.

Bruce Allen
www.boacinema.com

 

[Petr Dvorak]

synthetic

 

[hans de vries]

the technology behind it is interesting but I see no reason why 'filmmaking' should head this way. I couldn't care less about the end result.

 

[Roberto Lequeux]

Eventually, yes.

 

[Craig W. Bickerstaff]

Some filmmakers are already heading in this direction, Robert Zemekis being the main one perhaps Jim Cameron could move into it exclusively but I don't see ALL of films going in this direction it's just a different kind of filmmaking this is just another way to make a film like 3d animation, stop motion or hand drawn.

 

[Noah Kadner]

Well this is what they sorta did for Avatar too. Doesn't mean it's the 'future' of filmmaking just another tool in the tool kit. For every Avatar there's a Polar Express in terms of technology not quite there yet. Ironically sitting in a chair in a white room by yourself seems like about the least natural way to get a good acting performance. Humans play off other humans in real locations best. Just look at a certain recent prequel series for confirmation of this.

Noah

 

[Roberto Lequeux]

Think on location, and as many actors as you want at the same time.

 

[Craig W. Bickerstaff]

Thing is unlike avatar when they are sitting in those chairs they are getting more then just movement information it looks like they are also getting mesh detail and texturing which is why they put the actors through makeup before doing the bits.
For the record though I assume when you say "A certain prequel series" Humans did in fact play off other Humans in that film even when there was a digital character they had the character playing the voice stand in for the scenes often on real sets, so blame the bad acting on the actors.