By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Microsoft - Lee on Xbox Natal

NJ5 said:

I think he's being overly positive considering the failure of the E3 demo:

The human tracking algorithms that the teams have developed are well ahead of the state of the art in computer vision in this domain. The sophistication and performance of the algorithms rival or exceed anything that I’ve seen in academic research, never mind a consumer product. At times, working on this project has felt like a miniature “Manhattan project” with developers and researchers from around the world coming together to make this happen.


We all saw how easy it was to get their algorithms to mix up the arms and legs. It's like they aren't even taking advantage of the limitations of human anatomy in their tracking?

What I mean is... a well made model of the human skeleton would never consider the kind of movement we saw as a possibility. 

 

If you've ever tried rigging a 3D character for animation you'll know how easy it is for joints to get confused and things to get twisted horribly UNLESS the programming is put into place to compensate for that....for example, not allowing a limb to rotate to far in any direction, just as a real one would do.  The failure was with the avatar rigging, not with the tech.



Around the Network

That video was amazing when he switched to camera view with the head tracking. Can't wait to try Natal with my 9 foot projector screen!



I think it was pretty clear in my reply above that I was talking about algorithms i.e. software. Well, some algorithms are implemented in hardware, but that's probably not the case here.

Lee was talking about the software... so was I.



My Mario Kart Wii friend code: 2707-1866-0957

Anyone else find it funny that Johnny Lee is working for Microsoft? About 18 months ago he was the Nintendo poster boy. It's either funny or ironic, I can't decide.

I do find it hard to believe that MS is going to deliver a consumer technology that academic institutions don't have. I guess anything is possible.



Thanks for the input, Jeff.

 

 

NJ5 said:
I think it was pretty clear in my reply above that I was talking about algorithms i.e. software. Well, some algorithms are implemented in hardware, but that's probably not the case here.

Lee was talking about the software... so was I.

I'm pretty sure he's talking about the 3D tracking algorithms used inside Natal itself for tracking the person in front of them and being able to do things like predict movement, thats unrelated to how that translates to a rigged 3D avatar.   He's discussing Natal, not a specific application.  Or do you think every game is going to be using the exact same character rigging?



Around the Network
dbot said:
Anyone else find it funny that Johnny Lee is working for Microsoft? About 18 months ago he was the Nintendo poster boy. It's either funny or ironic, I can't decide.

I do find it hard to believe that MS is going to deliver a consumer technology that academic institutions don't have. I guess anything is possible.

As much as I hate using the word.....only fanboys would find this funny.  It's EXTREMELY common for people to work for companies that are in "competition", business is business.  The head of Media Molecule who made Little Big Planet used to work for Lionhead.  Amir M who was one of the driving forced behind HD DVD at Microsoft used to work for Sony.  It's just a job.



daroamer said:
NJ5 said:
I think it was pretty clear in my reply above that I was talking about algorithms i.e. software. Well, some algorithms are implemented in hardware, but that's probably not the case here.

Lee was talking about the software... so was I.

I'm pretty sure he's talking about the 3D tracking algorithms used inside Natal itself for tracking the person in front of them and being able to do things like predict movement, thats unrelated to how that translates to a rigged 3D avatar.   He's discussing Natal, not a specific application.  Or do you think every game is going to be using the exact same character rigging?

We clearly have a different view of how that demo worked. Read this interview, specifically the two quotes I put after:

http://www.eurogamer.net/articles/e3-post-natal-discussion-interview

So we have a custom chip that we put in the sensor itself. The chip we designed with Microsoft will be doing the majority of the processing for you, so as a game designer you can think about the sensor as a normal input device - something that's relatively free for you as a game designer.

Designers have 100 per cent of the resources of the console and this device is just another input device they can use. It's a fancy, cool, awesome device, but essentially you can just treat it from a free-to-platform perspective, because all of the magic - all of the processing - happens sensor-side.

Essentially we do a 3D body scan of you. We graph 48 joints in your body and then those 48 joints are tracked in real-time, at 30 frames per second. So several for your head, shoulders, elbows, hands, feet...

It seems logical to conclude that the avatar demo was simply querying for those points and rendering the avatar assuming those points are correct. The avatar demo should just be taking the data and rendering it on screen, not doing any fancy processing according to these quotes.

 



My Mario Kart Wii friend code: 2707-1866-0957

NJ5 said:
daroamer said:
NJ5 said:
I think it was pretty clear in my reply above that I was talking about algorithms i.e. software. Well, some algorithms are implemented in hardware, but that's probably not the case here.

Lee was talking about the software... so was I.

I'm pretty sure he's talking about the 3D tracking algorithms used inside Natal itself for tracking the person in front of them and being able to do things like predict movement, thats unrelated to how that translates to a rigged 3D avatar.   He's discussing Natal, not a specific application.  Or do you think every game is going to be using the exact same character rigging?

We clearly have a different view of how that demo worked. Read this interview, specifically the two quotes I put after:

http://www.eurogamer.net/articles/e3-post-natal-discussion-interview

So we have a custom chip that we put in the sensor itself. The chip we designed with Microsoft will be doing the majority of the processing for you, so as a game designer you can think about the sensor as a normal input device - something that's relatively free for you as a game designer.

Designers have 100 per cent of the resources of the console and this device is just another input device they can use. It's a fancy, cool, awesome device, but essentially you can just treat it from a free-to-platform perspective, because all of the magic - all of the processing - happens sensor-side.

Essentially we do a 3D body scan of you. We graph 48 joints in your body and then those 48 joints are tracked in real-time, at 30 frames per second. So several for your head, shoulders, elbows, hands, feet...

It seems logical to conclude that the avatar demo was simply querying for those points and rendering the avatar assuming those points are correct. The avatar demo should just be taking the data and rendering it on screen, not doing any fancy processing according to these quotes.

 

Yes EXCEPT that if you don't take into account joint restrictions there is nothing that isn't telling the avatar that his joint can't move 360 degrees and end up back at the same spot, only twisted.  It's not so much a question of is the joint in the right spot, it's how it arrived at that spot.

I'm not saying the system doesn't need work, and from all account it's still over a year from launch, but I am saying you can't say the system is a failure or won't work just because the avatar was poorly rigged.



@daroamer: I'm not saying it's a failure, I'm just saying I don't see how Lee's observation about being "way advanced" translates into anything tangible.



My Mario Kart Wii friend code: 2707-1866-0957

NJ5 said:
@daroamer: I'm not saying it's a failure, I'm just saying I don't see how Lee's observation about being "way advanced" translates into anything tangible.

Did you not read any of the reports of people actually using the device?  The reports have been unanimously positive, with a few nitpicks on things that still need work.  You don't think that's tangible?  All the reports of people using Natal to play Burnout have spoken about how immersive and natural the experience is.  Just because an avatar got twisted doesn't mean other things didn't work extremely well.