UPDATED: Ninja Theory: Uncharted 2 Animation Was "Overdone"

@Tameem

Naughty Dawg does the mo-cap and the voice work on stage at the same time.

http://www.msnbc.msn.com/id/33025731/ns/technology_and_science-games/

http://www.gametrailers.com/video/behind-the-uncharted-2/58093

@Tameem

First of all it's great to see you take the time to contribute to these forums as well as help set the records straight.

I can understand wanting to capture the audio along side the visuals when you can see the characters face, I think the dialogue between Trip and Monkey at the end of the demo showed this really well. However in Uncharted 2 for example we spent most of the time looking over Mr Drakes shoulder or watching things unfold in front of him. So I can understand capturing assets at different times as it didn't detract from the overall experience.

With enslaved can we expect to see more face time with the characters so that we can get the benefit of performance capture? As I'm assuming that performance capture is a technique that is used in film (as we tend to see more of the characters faces and interactions), is there much difference in cost between the different capture methods?

@Shane

Thanks for the links - I stand corrected. They do capture audio on set. However, it cannot be called performance capture if you are not capturing face which is crucial to the overal dramatic performace.

@Tameem

I understand and respect your point. While I find it interesting how different developers such as Ninja Theory and Naughty Dog(not Dawg, my mistake) go about performance capture I don't really care how each of you may define it as long as the intended result is achieved. Despite not fitting your definition, Naughty Dog does do a better job overall with it's performance capture, even their facial animation, than most other developers out there(my opinion, of course). And, no, I'm not comparing them with you and your team at Ninja Theory, that would be an incredible disservice to the work that each of you do. I truly respect what both you and Naughty Dog have done for character performance and storytelling in games and I will continue to support both studios.

Animation, Audio Capture, Performance... as a consumer did it matter to me when the game play was so fricken awesome, err no. although I did notice the vice acting as it was very good. all comes back to the same thing though, visuals vs game play.

and my spelling of `voice` wasnt so good.

Have a look at the facial reactions in the Enslaved demo and you'll see that they're as good as any sterling effort in recent games to date. It's up there with Uncharted and the likes of Yakuza, Siren: Blood Curse and others to say the least.

AAAAAHHHHHHAHAHAHAHAHAAHAHAHAHAAHAHAHAHAHA! ""[In comparison,] in Uncharted 2 [Naughty Dog] captured voice in a booth, then did motion capture to the voices..."

AHAHAH! HHHHAHAHAHAHA! I love it when people are this stupid and don't know what the f**k they are talking about!

Let me clarify. Uncharted 2: Among Thieves did its voice acting AND motion capture AT THE SAME TIME on a MOTION CAPTURE STAGE! The faciel animation was done separately! I want to kick this mother f**ker in the FACE for spreading such blasphemy!

Well the characters in Uncharted 2 looked very natural. The characters in the Enslaved look too stiff and wooden. So maybe Ninja Theory should have 'overdone itself' too. Even with Andy Serkis involved, the characters look like empty shells to me. Games don't have Avatar level fidelity, so any subtle movements need to be exaggerated or they become completely lost in the polygons.

Just read Tameem's comments. Feel slightly bad now. I still get the impression that something is lost somewhere in the motion capture though. Can't put my finger on it, but its a similar feeling I have with characters in Robert Zemeckis' CGI movies. Anyway, wooden acting or not, that's only part of the experience and it wouldn't sway me from enjoying it.

In Heavy Rain we recorded our individual facial capture and voice simultaneously in a booth, then went into the mocap studio and matched the body capture to the replay of the cut-together voices some weeks later. At this point I got to work with some of the actors I had scenes 'with', but often it was only their voices and a stand in (not wearing a mocap suit).

Incidentally, I was at Bafta, and saw Tameem talk. He praised Naughty Dog, and cited the different approaches, and went on to make a more general statement, and that's how it came across. At first sight I thought the "overdone" was just a classic headline hook, but as it was a real contextual misunderstanding, I'm glad that TA has had a chance to clear it up.

Thanks for the comments Pascal. Any news on what you're up to next?

Regards

Tim Smith

For me it's the Uncanny Valley of the character modelling and textures. When you get very close to having the characters appear human, even the slightest shortfall screams "freeeeeeeks". It used to be the movement and lips synching (Final Fantasy Spirits Within movie), but better full body facial capture has pretty much bagged that. Eye movement has been a problem too, but mocap refinements have conquered that too - Beowulf's magic nerve impulse measurement has got to be the most novel approach.

Texturing tends to be the problem these days. Skin texture remains an issue (FFSW again, and Polar Express) - it's very difficult to mimic the skin's translucency and light transmissive nature, especially as a changes from one centimetre to the next across the whole body. It's almost beaten, but it was still evident at times in Beowulf. Dead eyes are another problem, again probably due to the complex structure of eyes and the way that they are layers of translucent and opaque tissue, causing light to refract and reflect internally and affect how the pupil is lit.

The biggest performance fail for me in Uncharted 2 was the facial capture - the mouth, eyes and brows all moved, but it did seem a bit seem a bit heavy handed. As mentioned by other commentators, it lacked subtlety.

However, above all of this the biggest failing was the eyes - it looked like hard, glossy pebbles had been pushed into their faces. Even in dimly lit shots they all had this demonic specular glint going on - I was definitely screaming "freeeeks" at that.

Oh, and while we're talking Uncanny Valley; nobody should never think it's a good idea to make photo-realistic penguins, owls or any other animal act and move like a human. It's an abomination, and that's without the f**king mention of dancing and singing.

I'm in a couple of episodes of "Any Human Heart" for TV, and of course I'm continuing the building of Motives in Movement which is all about improving dramatic content in games. I was involved with Milo & Kate, too, so now you could say I'm 'available' al... NDA NDA NDA NDA NDA (play music....)

Before Tameem's talk I was about to post a blog looking at the effect of FACS on attribution of emotion. The whole debate about human expression fidelity and the uncanny valley intrigues me, so some of the posts I've seen on this forum have been quite interesting, and will inform the next draft! So, thanks!

Hi Pascal,
Excellent stuff. We've got some very knowledgeable posters in the Forum.

What's your blog address? And do you fancy doing an interview with us?

Cheers

Tim
tim@spong.com

Thanks for taking the time to share your thoughts here, Pascal.

My interest comes from the technical point of view. 15 years ago I had aspirations to work in CG, but the web grabbed me by the short and curlies and tore me off in a different direction. My intrigue in the subject hasn't waned, and as those that know me will attest I tend to over-analyse CG, often coming across as simply negative, when actually I'm just deconstructing what I see :-\

I think the fidelity of expression is vital in keeping away of Uncanny Valley. In the same way that we immediately pickup on a skin that doesn't quite have the right translucency, any "major" facial expression that's delivered without its subtle complements we'll perceive as false, wooden or emphasised. I've no doubt that in most cases it's not because that expression has actually been "overdone", but rather that without faithful reconstruction of complementary movement to provide harmony, the expression seems to shout out against the silence. We all know that a mouth that smiles is insincere if the eyes don't "smile" too - I think the same goes for every expression, it's just that they're not always so obvious that we can deconstruct the required components easily.

enslaved is a load of crap, the reviews for it are mediocre, the highest score it got is 9.1, graphically its not as good, gameplay is not that good and no good replay value, the only person who could make comments on UC2 is someone who has made a better product than and in ninja theorys case they have never made a game close to UC2's standards

enslaved is a load of crap, the reviews for it are mediocre, the highest score it got is 9.1, graphically its not as good, gameplay is not that good and no good replay value, the only person who could make comments on UC2 is someone who has made a better product than and in ninja theorys case they have never made a game close to UC2's standards

I haven't played the game, but when the hell did 9.1 become a mediocre score???

Jason,Thanks for the great tip on using the SEP with CNX2. I used the trial vseiron and it worked just great. Was able to go back and play with some old images. Will look forward to more of your podcasts and browsing your ebooks.Thank you again