Now if avatars could look like this!!

oh that's so cute ...

Sorry i couldn't resist ......

That's why they spray the actor with fluorescent paint. Right.

That's just an implementation detail.

As a frequent Wikipedia contributor, I know just how much respect is due ANY wiki page.

I like how nowhere in said wiki page does it refer to anything as 'Surface Capture'.

It does, however, state that more recent technologies (Motion Capture technologies) are ' able to generate accurate data by tracking surface features identified dynamically for each particular subject' without markers.

There also is no reference anywhere within Wikipedia about anything called 'Surface Capture'

Your wiki page has backfired.

(It also seems that what you're referring to is called Performance Capture on that page, though I could be incorrect. Either way, even that is still motion capture.)

Are u aware that a Home PC is just able to render one of of that Head (or face) and not 30 or more with their bodies as well? The SL grafic-engine has even more to render than one face, there re also the whole virtual world with particle-spots, water and much more..

So we've to wait some more time and dreaming from. I suppose at less 4 years.

All I see is humans...boring. And they're not all that interesting to look at to me, Vicki 4 can walk the walk and is publicly available. /me shrugs

For starters, it's really good, but I think the 'Emily' video is more impressive. Bradley, it's interesting that you say that, because I recently watched a television show about AI, and research has been done in the area of how realistic an artificial intelligence looks. Research has show that robots and androids can look human to a point, but, once a certain threshold of realism is hit, people start to get 'creeped out' by them. It's pretty interesting because you're saying the same thing about an avatar.

the technolgy can be (and has been) used to apply the motions observed to a character frame. so you could still be any character you like, just with more realistic expression and movement.

I'm not bothered by the photorealistic effects, but the uses they could be put to does concern me. but then I'm biased, I've always liked the idea of having a persocom

ETA:
interesting article from july/august IEEE Computer Graphics about Uncanny Valley effects (found on the MOVA site) also describes the contour capture system as a type of mocap (and the phosphorescent makeup serves as a randomized marker pattern with greater density than traditional methods)
http://www.mova.com/pdf/IEEE_Computer_Graphics_Uncanny_Valley.pdf

Train your dog to suck in his muzzle and point the camera at him.

I don't think I would care to wear special clothing or any special appliance just to play SL. Nevertheless, I do like any feature that adds a dash of realism to the SL experience. A better av mesh would be just fine and dandy.

@Void: what's a persocom?

All we need for Much Much better av's in SL is better avatar meshes as the textures do the
most of the work on the appearance anyway.

Currently no matter how good your textures are you can always make out blockiness in certain area of the avatar mesh. It is just a really primitive figure mesh. I have heard that the
current SL av mesh is from Poser 1 technology. Poser 1 was released in '95, meaning that the av tech we are using is 14 years old.

If we even just had meshes that managed to hide the blockiness like smooth hips and shoulders
along with some improved uv mapping on them this would allow skin makers many more options for better skins and and realism. Better mapping would allow for a more modular
approach to skin design allowing for easier applications of make up and tats not to mention
being able to switch faces without changing your entire skin.

Ongoing of course articulated hands feet and facial expression would be great. If they did
upgrade the av mesh by at least a few magnitudes then we could have those as well.

It would of course reset the clock in the skin and animator races but hey that is always the way with progress.

a persocom is basically a droid (a machine in the shape of a human) usually females, that do tasks for the owner

larger ones can have emotions and learning capabilities

Thanks, Rhaorth. Is that a fictional droid? Or is does it, for instance, apply to the fancy robots Japanese engineers seem fond of building?

Funnily enough, I read quite a lot of science fiction and my vocabulary is quite extensive but I never saw the reference until Void's post this morning.

Having been involved in producing animations using MOCAP studios in Dallas and London for the last couple of years, may I clarify a couple of things here.

MOCAP uses markers, and the actor is severely restricted by a huge swathe of cables attached to his/her body. See the images here for an idea, http://seneludens.utdallas.edu/facilities/

MOCAP, as Rock correctly defined, only captures motion.

Since MOCAP came along two new players have significantly moved the technology along.

1. Image Metrics
They use the term 'performance capture'. An example of how they do it is here:
http://www.image-metrics.com/services/games/full_service
They use no markers, no makeup and no special equipment. Image Metrics also produced the famous Emily video, see it here:
http://www.youtube.com/watch?v=UYgLFt5wfP4

2. MOVA
They use the term 'Contour Reality Capture'. Again they use no markers, but a special phosphorescent makeup is applied that enhances further the 3D capture depth. MOVA technology was behind the recent 2009 Oscar award for best special effects for Benjamin Button. How they do it is described here:
http://www.mova.com/flash/

These two players are so far out in front of the older MOCAP technology that to say they are the same thing, which I think Argent was trying to argue, is doing a great disservice to the technology, they are like chalk and cheese as far as I am concerned.

Unlike MOCAP the actors wear no markers, have no trailing cables which restrict their movement, and are free to act. The new technologies also capture the appearance of the actor, right down to every mole and freckle on their face (which MOCAP cannot do). The technology is not limited to current filming either. Older films can also be analysed by the software to produce Humphrey Bogart skins, Marilyn Munroe skins, etc, and have those skins applied to a modern day actor, to bring back the stars of yesteryear. MOCAP can not do that!

It is not surprising that these technologies are not mentioned on the MOCAP wiki page, they are so different they need a page of their own.

However, are these technologies likely to find their way into games, anytime soon? Well, both companies are pitching at Games Engine developers, and Image Metric technology has been used in several games already.

'What kind of supercomputer would be needed to see this technology though?' you may ask. Other than seeing very short sequences of facial animation you cannot expect to see the full power of this technology on any home rig, right now.

However, the current trends in PC and Internet development are moving towards a very different approach, where the software you use and the games you play are no longer held on your home PC, with all its dependencies, but are held on super-servers, which then use 'almost' lossless compression and minimal latency to deliver the resulting images over broadband to a home PC.

The advances we will see in the future will be more to do with broadband advances than PC advances. One company has already produced such streaming software so that you can play a high-end game on your PC, then continue the game while on the move on your mobile phone, using Flash, directly into a browser! See more info here:
http://www.multiverse.net/index.html


Hope this clarifies things a little.

thats not "facial recognition", but i digress.

i had a look at that shop, on avatar island... 2700L$ with a "no refunds whatsoever" policy, no thanks.

any old kMADD, soul, or laqroki skin looks better (maybe not that much like "you" than their automated results.

Boring. SL has enough of those magazine cover beauties already. How un-creative.

Sounds a lot like my wife . . .

Pep ( . . . except the bit about learning capabilities)

The only thing about our avys that really need changing are the fingers which always look clunky and immobile and some of the weird 'double jointedness' ....I wince every time my arm folds back when Im building.

It would be nice if hands didnt sink through other body parts , when youre sitting for example, and you hand suddenly fuses through your own knee or thigh..

I would put it a lot stronger than 'It would be nice if..'. I think it is ridiculous that with today's technologies we have to put up with hands going through other body parts, and al the various contortions and distortions that our avis go through sometimes after teleports and after anims. It does not happen in any of the other games I play, so I see no excuse why people here should have to put up with it.

I also play Crysis, where the avatars look amazingly real, and facial expressions are great (real hands too). Do they use the new technology being discussed here, or is that some other product?

I looked at the video and I've seen other similar ones. I think they are amazing.
'Jennifer' even looks a bit like me!

This probably goes further than we need in SL. Our avatars are pretty reasonable already, though a more detailed mesh and smoother joint articulation wouldn't go amiss. Non-phantom interface between flexis and avatar for hair and clothing would be good too.

I'm skinny, but it would be able to see fat avatars who are able to wobble nicely and not have hands disappearing into their love-handles!

Not just hands going through body parts. I was using that great anim from Bits and Bobs, Soul Sensations, and my g/f had long prim hair on, and when she arched her head backwards her prim hair penentrated my body to the other side and then proceeded to slice me up from waist to shoulders, Wolverine-style.

The things I put up with for SL, lol.

Rock

Meh... If you have to hire an actor/actress to pre-record every word and facial expression, why bother? That could NEVER be used with any effect for a 3D, real-time environment like SL, where any Resident can say anything, in any language, at any time.

It's a fine toy for making CGI movies and locked-sequence video games, where for a given scene, there are only a fixed number of response choices. The actor follows the script, and your Orc or Elven Princess comes to life in your CGI movie. Or the actor records thousands of sound bytes, and depending on what canned response choices the Player makes, the virtual girl does her thing. But it all takes tons of post-process work. None of that is real-time.

But for a real-time unrestricted environment? If you want it that "real", set up a webcam and Skype.

For me? Not on your life. I don't want a virtual clone of the real me. I want the nearly infinite possibilities of being furry or whatever in SL.

I don't see this kind of face capture as all that useful for realtime engines any time soon, least of all SL. Try to put a 100,000 polygon face in SL, and it's going to go, "Are you f'ing kidding me?" just before it chokes to death. It can barely handle the avatars we have now, and they're only 9000 polys for the whole body. It'll be a good long time before any realtime engine can make much use of technology like this.

But for CGI in movies it's usable right now. It sure beats the hell out of covering actors' faces with markers and making them wear awkward mocap suits.



Excellent post, Front. Very imformative. Thanks.

Fantasy is good for single player games, but I don't like the obsession with beauty in virtual worlds. I would much rather see avatars replaced with real people.

You'll take my avatar away over the corpse of my cold, dead account:

You'll take my avatar away over the corpse of my cold, dead account:

I meant the fantasy of being a beautiful human!.