Now if avatars could look like this!!

You mean like this guy I hang with?

I hope we can never smell in SL.

Chosen, you misunderstand my point. Systems that do motion capture to create CGI are wonderful, for environments that have the luxury of hundreds of hours of post-production time for every minute seen on screen. I'm sure Pixar and Industrial Light and Magic and various movie production studios are salivating over this new toy. And rightly so. It's a heck of a step forward from the first mocap suits with ping pong balls glued to the actor.

But it doesn't do real-time conversion of what the person behind the cameras is saying. Nor can it take input from text or audio only and animate the face that it generates.

I've seen and used DAZ|Studio and Mimic, where a much better avatar than SL could hope to use (or one just as crude as the one we use today) can be animated so its mouth and face moves based on phonemes in an audio track. That technology has been around for years, and runs on a reasonable-quality desktop PC. It uses an existing 3D avatar figure and an audio track, and the mouth and face animation is independant from how you pose the avatar for each frame. And it still takes many, many hours of post-production type work to render even a few seconds of animation.

I fail to see how this could be of any use at all to an avatar in a 3D real-time virtual world, who needs to be able to appear to say what the Player types, or what she says into a microphone via Voice Chat. The processing simply could not keep up.

Suppose you take an actress, and use this system to capture all the possible phonemes in all languages, for her face. They won't flow smoothly from one to another in the nearly random sequences necessary to communicate spontaneously. It won't alter her expression all that well if you change the slider settings on the avatar. It won't know what to do at all if the avatar is a furry with a prim head...

Is this promising? Of course it is? Are we likely to see any benefit from it in SL? I do not think so. At least not for decades, at best.

Thank you, Ceera. That's the point I was making before getting sidetracked over the definition of "motion capture". Whether you call the technology "motion capture", "surface capture", or "transforming degenerated quadrics using a particle filter" is beside the point... to be practical in virtual worlds it will have to work in real time, based on the actions of the user, and *keep working* whether the user is looking at the screen or not... or for that matter whether the user is typing, talking, or using a prerecorded gesture.

Now I'm rather curious what would happen if the user left the computer for a bit.... would their face melt off? XD

from my understanding they are fantasy (not a real physical thing in our world)

Nah, you'd just get a blank, fish-eyed stare. Rather like asking most Lindens for a straight answer on anything...

Ceera, let me clarify a few points. I am not clear from your last post if what you were referring to was traditional MOCAP or the newer Image Metrics/Mova offerings. If it was MOCAP you were referring to, then the comments are quite valid. However, if it was Image Metrics/Mova then it is not 100s of hours of post production now per minute, it is more like minutes per minute. Of course, depending on the customer, and the amount of further customization required, this can stretch into hours, but nowhere near the 100s of hours needed previously.

You are right though when you question if either Emily or Geni4 have any use in SL, they probably do not.

The main outputs of the software which are pertinent to SL are shapes (which can be converted to SL shapes quite easily), skins (also a small amount of conversion), and motion for animations. You really can make both your shape and skin look exactly like a particular favourite film star of yours (not nearly alike, but exactly alike). Of course, looking like a film star may not be to everyone's taste, but I am certain it will appeal to many in SL.

There are companies who are currently developing virtual worlds where voice-linked, and even text-linked emotes will be incorporated, in real time. Are you aware of the Crazy Talk software? You might like to try it out, http://www.reallusion.com/crazytalk/. The notion of facial animation in 3D real time is a lot closer than you think. I would say 12-18 months.

However, given the infrastructure of SL, I doubt if any of the most recent developments could be utilised. It would have to be completely rebuilt. One new Virtual World is about to launch that has zero lag, no time waiting for textures to 'rezz' and where the draw distance is a null concept, you can see all the way to the horizon, instantly. Even given these advantages, I do not see them being a direct competitor to SL, but if the SL infrastructure does not evolve, and radically, I do not see SL lasting decades at all, technology just will not wait for them.

Given the amazing and (to me) suprising software and hardware developments over the last 12 months, heavens knows what we can expect in the next 12-24 months, I have given up trying to predict on that score (but I do believe scene pre-caching is almost certain).

Sorry if my posts run on a bit, but animation technology is something very close to home with me.

Bah, we could do that in Colossal Cave in the '70s. It's all a matter of the bandwidth of your imagination.

There is something out there now that is affordable and could be applied to SL immediately:

http://www.ipisoft.com/

3D motion capture, no suit. Does body, not face yet but an express edition which will work with a single camera is only $250. Not sure what all the supported formats are besides BVH and SMD. 30 day free trial. Only real drawback is that either GTX285 or Radeon HD4980's are recommended.

Actually, they are, just not using the term "surface capture." They're under the heading "Related Technologies", with the subheading "Facial Motion Capture" - which may or may not be an accurate description.

In any event, it's amusing to watch people arguing semantics over evolving technology and its terminology. It reminds me of the days when people argued over "spam" versus "uce" or whether "spam" was strictly email or included unwanted communications in other technologies. In the absence of a trademark or a definition by some standards body, there's no such thing as an official definition. So just focus on the technological distinctions - which seem quite real. The terminology argument is a distraction.

Nobody took Steve Boursy and John Grubor seriously.

That is very cool technology! I think the demos were done with two cameras. Even so the price of the 2 cam version seems very reasonable given what it can do. I love this part:
It will be fun to see what comes out of all this stuff.

Avatar Island?

One thing - when I first got to SL, even the avys as they are NOW seemed kind of creepy to me.

If they made avatars that were actualy life-like, people would get used to it.
Thing is, would the human looking avys have all the nasty imperfections of a human being? Who would want that?

that is super cool. and hell yeah sign me up!

@ Rhaorth & Ephraim
Yes, persocoms are (very) fictional. They are found in the manga/mnime series Chobits. Essentially they are personal computers with bodies, and highly advanced sofware allowing them to mimic emotion. they come in a range of sizes from the full sized models, to the smaller 'laptops', to even tinier 'pda' types. They are often humanoid, but not always (the smaller ones seem to be more variable), and female models are what's most seen in the series (though there are references to male versions).

They have two interesting distinctions from most humanoid robots in popular culture. one is a recognition system that essentially ties each model to it's owner (it only recognizes specific people as authorized to give it instructions) and the other is that it's entirely removed from the bounds of Asimov's Laws of Robotics, by their originators choice.

it'd actually be nice if something like this were used to at least make more realistic facial expressions, even with the reduction in resolution needed for SL...

Japan's a funny place. I'm pretty sure that "persocom" can be any personal computer... I've seen the term applied to contemporary PDAs and cellphones as well. I think in Chobits the idea was just that these were what the personal computer had turned into.

It looks like crap. It looks like they paste a picture of the person on the avatar face.

I think that those realistic avatars would be cool but I would not have it look exactly like me. I already have one of me, I like my avatar to be different! LOL

I wouldn't mind more realistic overall avatars, but I don't want people running around looking like themselves. Too creepy. I like the caricature-ness of SL avatars.

I have seen something like this, but since I don't read kanji it's hard to get a sense of whether or not it was a matter of the translator taking liberties... TBH I'm not sure if the term started with them. I had thought it started with the Clamp writers (or was borrowed from another writer) and then became popularized. but considering the idea behind it, kinda makes sense, and my impression could be off by a mile.

Me too - it is awsome, but at this stage, I'm yet to be convinced.

Regards Zena Tiki

The thought of me sitting around in *my* body in a skin tight mocap suit...what will the Fedex guy think when I answer the door.

I dare not repost that picture, but I will ask if someone could kindly pass the eye bleach please?