Uncharted 3

Sorry for insisting, but what's with the discrepency between Drake's character's models...perhaps I am mistaken, but in some instances, the protagonist seems poly starved...
 
Avatar's sytem works with just make-up, no need for reflective markers. I think they can also track eyes to an extent, using image analysis. And yeah, 90% of the performance is from the facecam capture. Then again it's usually the last 5% that's the hardest and most work intensive to get right.

Yes, that's what I meant. After that they can improve on more specialized tech to minimize the human effort.
 
After that they can improve on more specialized tech to minimize the human effort.

It's unlikely, especially when you're not working with a 1:1 recreation of the actor.

And once again this is something that requires human intelligence and empathy, to be able to be able look behind the skin and understand what the performance tries to convey and also be able to recreate it on the CG version. Computers can't do that yet and it doesn't look like they'll be able to in the foreseeable future.
 
I think the discrepancy between the cutscene and realtime Drake model.

I'm not well versed but I observed that in some frames -I'm talking about gameplay with no reference to cut-scenes- Drake's face looks much worse (Clear loss of details and perhaps inferior geometry...don't know, hence the question) than in other phases. It's difficult to convey the point given how approximative my english is...
There is a scene where the hero struggles to lift an equipement (the player is required to free him by pressing the triangle button) when water starts flowing in, and you can see what I'm pointing at! :smile:
 
Totally forgot about Laa-Yosh's explanation. Do the Avatar folks wear markers on their face ? I read that they also carry a face camera over their head to capture their raw facial input, to supplement facial animation.

Please, just watch this: http://bcove.me/mdebwyce You can see they have some spots on their face to help facial detection, and somewhere in the movie you can also briefly see what the live interpretation looks like, which includes tracking the eyes.
 
Yes, that's the article I read. The Wired article mentioned that they encountered uncanny valley effect when using the automated facial generation system. The article didn't talk about how they solve this problem. Perhaps the team solved it by touching up the final model manually.
 
Huhh? Are you talking about Avatar? Modeling and animation are different aspects of CG character production...

The faces themselves are all artistic creations, created completely manually, usually starting with a clay maquette and refining it in the computer, then building the final models.
Animation is implemented with blend shapes, which is basically using hundreds of hand sculpted elemental expressions, created from the same base face model. Each blendshape defines a simple translation for every vertex of the model. These are combined into a relatively limited set of 80-150 controls which is what we call a character rig.
They do create a lot of photographs and scan the actors' faces, but these are only used as reference.


This character rig is driven by the facial capture data, but the animation won't be lifelike enough using the results of the computer analysis only. So every time an animator has to go through the result to clean it up and add the little touches that put the illusion of life into the CG character.

This is one of the main reasons IMHO why Avatar has worked so much better than completely digitizing based approaches like the Matrix sequels, a human touch still adds a lot to the results.
 
Huhh? Are you talking about Avatar? Modeling and animation are different aspects of CG character production...

Yes ! Still talking about Avatar. Now I know what to call the animation data. :p

This character rig is driven by the facial capture data, but the animation won't be lifelike enough using the results of the computer analysis only. So every time an animator has to go through the result to clean it up and add the little touches that put the illusion of life into the CG character.

This is one of the main reasons IMHO why Avatar has worked so much better than completely digitizing based approaches like the Matrix sequels, a human touch still adds a lot to the results.

Is there on-going R&D work done on converting facial capture data to animation rig directly with minimal/no human assistance ?
 
MotionScan is that R&D, but it's a dead end for almost everything. Movie directors will always want creative input and there's usually no need to 100% reproduce actors whom you can simply film with a camera. For digital doubles in stunt scenes / head replacements, there are cheaper methods; and even there it's important to have more control than just a "replay" button.

Matrix tried to use digitizing everything as well, I guess you can see for yourself how unconvincing it was. Movies are supposed to be larger than life anyway, we don't go there to see reality. Human assistance is a welcome feature for most, and not a hindrance to be replaced with computers.
 
AlStrong, don't get me wrong. I am just pointing out that game comparison and critic can be done objectively and may be informative. We had a few before. But what were posted above are not game comparison.
 
We had a few before. But what were posted above are not game comparison.

There's a game tech thread for sane discussion, but what was going on here was another example of how some particular game is superior to everything and thus nothing else is impressive - how does this add to discussion? It doesn't. All we end up with is a list of requirements and examples and then- Oh wait, this is the game thread. There's always some crazy need to downplay or diminish other's opinions on what they think looks good by naming some other game, and quite frankly it's beyond old for these forums. Grow the hell up.
 
Thank you Al for doing this. T_T Some of the user impressions are not technical enough to post in the tech section. A good game is more than math anyway. The same math can be implemented differently to suit multiple needs and profiles. I'd imagine we can still compare and critic games in good spirit ?

I'd actually paused the trailer to check out Lycan's comment about Nate's model discrepancy. The facial expression looks fine but does look a little rough.
 
There's a game tech thread for sane discussion, but what was going on here was another example of how some particular game is superior to everything and thus nothing else is impressive - how does this add to discussion? It doesn't. All we end up with is a list of requirements and examples and then- Oh wait, this is the game thread. There's always some crazy need to downplay or diminish other's opinions on what they think looks good by naming some other game, and quite frankly it's beyond old for these forums. Grow the hell up.

It's OK if you want to remove the Crysis stuff (which wasn't even a problem since it was simply an example of a specific graphical effect, not a "crysis 2 is better" type of comment), but removing the whole discussion? Come on, you're basically saying that only circle-jerking is allowed :rolleyes: You could have at least moved it into the game technology thread...
 
You can still post it in the Crysis thread if you like what's in the game. I suspect people will follow up if they have more to add. They would also be able to appreciate and see for themselves the differences, instead one 1-2 lines of out of context comments.
 
Back
Top