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

Forums - Microsoft Discussion - CNN: Kinect is impressive

Icyedge on 02 August 2010
Mr Puggsly said:
unknown_soul89 said:


A computer with 20 TB of ram can do things that a computer with 256 mbs can't, it doesn't mean that the ram is new tech, it's just more of or more refined old tech so you might want to check your logic considering the things that kinect can do that the eye toy can't (or rather can't do as well) is because of the 2ed camera 

Well here's the thing. The 256MB and 20TB of RAM actually do the same thing. They store data.

You could say the NES and Xbox 360 are the same because both play games. But clearly one has far more capabilities and potential than the other.

Hence, the Kinect has far more capabilities and potential than the Eyetoy. The Kinect will do things that a simple webcam was never designed to do.

 

Thats right, an easy example: playing in a dark area with eyetoy is hardware impossible. Whatever the software normal cam cant do that.



Around the Network
Icyedge on 02 August 2010
unknown_soul89 said:
Icyedge said:
unknown_soul89 said:
Icyedge said:
Mr Puggsly said:
unknown_soul89 said:
 



I've seen plenty of videos where the games work. And I've seen a couple videos where they had trouble on stage. With that said, it clearly has issues in certain enviornments. As did the Eyetoy.

The Kinect is new technology. It has capabilites that can't be duplicated on the Eyetoy. The Eyetoy is just a webcam.

Thats right, eyetoy body evaluation is only software. Kinect is new technology but use the same concept eyetoy already had. Which is allright.

How is a second camera new tech?

Do you know another device having Kinect hardware and software? Hence a new tech. 2 Cam by itself isnt but when put together to do that it is.

By that logic almost every update sony makes the ps3 is new tech in the same way kinect is new tech as in not really 

Yes sometime they bring new tech through software thats right. Every updates is a big exageration tough. Even more so that if the tech already exist somewhere else it cannot really be called a new tech.



youarebadatgames on 02 August 2010
Icyedge said:
Smidlee said:
 

What do you think this statement really means?

"What Natal does, is it evaluates effectively trillions of body configurations every frame. We've made it do that 30 times a second."

 X360 does not even have  a trillion bits.  Trillion is a really huge number. I've heard that  a trillion dollars in 100 dollar bills would stack in a pile that's 67 miles high.

It means that Natal select 1 body position (the one your in) out of trillions of different possible body configuration 30 time per second. Simply mean that it check your body configuration/position at 30 FPS.

In other words, while theres trillions of possible body position each frame, Natal will know yours. Its just that, but it was worded to look more awesome.

If not you were right in the sense that no super computer can evaluate trillions of body configuration 30 times per second. Tough, it can evaluate one out of those trillions possibilities.


The keyword here is "effectively", it probably does some sort of statistical inference per pixel to determine whether or not it is 1) part of a person, and 2) position in relation to the joints.  It doesn't mean that Natal is doing the work of 1 trillion body positions if the effective algorithm can come up with the same result with far less calculations.

Anybody can google scholar his papers.  One of his was getting sign language recognition to work on a 600mhz laptop with a webcam, a problem that was previously too computationally intensive.  It's not unbelievable that the tech has been advanced to a point that it can finally run on something as simple as an xbox.



Smidlee on 02 August 2010
Icyedge said:
Smidlee said:
 

What do you think this statement really means?

"What Natal does, is it evaluates effectively trillions of body configurations every frame. We've made it do that 30 times a second."

 X360 does not even have  a trillion bits.  Trillion is a really huge number. I've heard that  a trillion dollars in 100 dollar bills would stack in a pile that's 67 miles high.

It means that Natal select 1 body position (the one your in) out of trillions of different possible body configuration 30 time per second. Simply mean that it check your body configuration/position at 30 FPS.

In other words, while theres trillions of possible body position each frame, Natal will know yours. Its just that, but it was worded to look more awesome.

If not you were right in the sense that no super computer can evaluate trillions of body configuration 30 times per second. Tough, it can evaluate one out of those trillions possibilities.

Here some fun with numbers. For me to get my body into a trillion different positions within a 100 years I would have to be on average  in over 300 different positions per second.  

I agree the real information in his statement is Kinect  runs on 30 frame (inputs) per second.
 



Icyedge on 02 August 2010
Smidlee said:
Icyedge said:

Here some fun with numbers. For me to get my body into a trillion different positions within a 100 years I would have to be on average  in over 300 different positions per second. 


lol would be complicated.



Around the Network
welshbloke on 03 August 2010
Icyedge said:
unknown_soul89 said:
Icyedge said:
unknown_soul89 said:
Icyedge said:
Mr Puggsly said:
unknown_soul89 said:
 



I've seen plenty of videos where the games work. And I've seen a couple videos where they had trouble on stage. With that said, it clearly has issues in certain enviornments. As did the Eyetoy.

The Kinect is new technology. It has capabilites that can't be duplicated on the Eyetoy. The Eyetoy is just a webcam.

Thats right, eyetoy body evaluation is only software. Kinect is new technology but use the same concept eyetoy already had. Which is allright.

How is a second camera new tech?

Do you know another device having Kinect hardware and software? Hence a new tech. 2 Cam by itself isnt but when put together to do that it is.

By that logic almost every update sony makes the ps3 is new tech in the same way kinect is new tech as in not really 

Yes sometime they bring new tech through software thats right. Every updates is a big exageration tough. Even more so that if the tech already exist somewhere else it cannot really be called a new tech.

I am begginning to think anybody who says its just a PS Eye should start to see the ban hammer as this fud has been going on long enough.

The depth sensor which you guessed it the PS Eye never had is new tech. My favourite example of how the depth sensor works is the deflating ballon. A normal sensor can guess when a balloon is coming closer by calculating its size, the depth sensor can calculate this whilst the balloon is deflating the PS Eye would not be able to do that and would believe the balloon was stationary.



W.L.B.B. Member, Portsmouth Branch.

(Welsh(Folk) Living Beyond Borders)

Winner of the 2010 VGC Holiday sales prediction thread with an Average 1.6% accuracy rating. I am indeed awesome.

Kinect as seen by PS3 owners ...if you can pick at it   ...post it ... Did I mention the 360 was black and Shinny? Keeping Sigs obscure since 2007, Passed by the Sig police 5July10.
NJ5 on 03 August 2010
selnor said:
NJ5 said:
welshbloke said:
NJ5 said:

My calculator evaluates trillions of possible answers to 1 plus 1 every second. It knows that it is not 3, or 4, or 5, or ....

That's the only sense by which Kinect can evaluate anything trillions of times per frame.

I could take your point or I could believe that this chap knows his onions.

PhD at Edinburgh University, 1992:
Stable Segmentation of 2D Curves

Andrew Fitzgibbon is a senior researcher at Microsoft Research, Cambridge, UK. His research interests are in the intersection of computer vision and computer graphics, with excursions into neuroscience. Recent papers have been on the recovery of 3D geometry from 2D images, general-purpose camera calibration, human 3D perception, and the application of natural image statistics to problems of figure/ground separation and new-view synthesis.

He has twice received the IEEE's Marr Prize, the highest in computer vision; and software he wrote won an Engineering Emmy Award in 2002 for significant contributions to the creation of complex visual effects. In 2006 he was awarded the Roger Needham Award for his contributions to computer vision and machine learning.

He studied Mathematics and Computer Science at University College Cork and at Heriot-Watt University, and received his PhD from Edinburgh University in 1997, then spending 8 years at Oxford University's Department of Engineering Science before joining Microsoft in 2005.

 

I don't care if the guy is a Nobel Prize winning astronaut who has fucked Claudia Schiffer, that quote is wrong by the simple fact that the only mainstream hardware which can do a trillion of anything per second is a very fast GPU, and that's when counting very simple operations like a floating point calculation (and it's not even 30 trillion per second which would be required for doing it at 30 fps).

Listen to NJ5 a forum poster on the good old interwebz, or 

someone who knows what they are talking about.

Hmm. Love the internet.


You love the Internet and I love the appeal to authority fallacy without anything substantial to back it up or to point out where I'm wrong.



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