Microsoft: “No Reviews of Kinect!”

[Smidlee]

bobbert said:

Smidlee said:

bobbert said:

Smidlee said:

Well for one thing , kinect doesn't see things in 3-D no more than another other web cam. The claim Kinect "sees" in 3D is a little misleading as most people think 3D as like how they see. Kinect cam sees a completely flat image.  There are ways to try tracking something 3d on a flat image . For example Move uses the size of the ball on the image to determine where the move controller is in Z-axis. The more pixels the ball image uses the closer the ball is relative to the camera.  Kinect tracking the Z axis how the strength of IR light bouncing off an object.

 I can think of two major disadvantages of Kinect tracking you sitting vs standing up.

 1)  You are closer to surrounding objects including what you can sitting on. (standing puts distance between you and other objects)  While our brain can do this with ease even with one eye close yet it's a lot tougher for and  computer to do this.

 2) Your images itself takes up less pixels on the camera which means  Kinect is recieving  less information.

So while Kinect may work sitting down I doubt it will be able to track as well as standing up. This is good reason Microsoft wanted developers to develop games at least at launch standing up.

really? so any camera can see 3D as well as kinect's RGB camera near-IR emitter & IR sensor? Both cameras are working at all times. You might want to actually look at the kinect compenent list before acting like you know how that Z-axis tracking works. Your hypotheses might be more believable

I watch the video of how Kinect works and they make it clear this is  how it works. The problem with using RGB camera when it comes to depth  is most homes has multiply light sources makes is extremely difficult to track Z axis. Our brain can do this with ease with just one eye even though you don't see  true 3D with one eye. The brain uses lines, lighting, shapes, etc.  Even super computers have trouble take two images and make a true 3D image with any kind of speed  like our brain does.

 Kinect software is the tough part as it takes all those pixels and guesses the position of the player's body.

You obviously have no idea what you're talking about. All of the videos are very generic descriptions, and I have seen some that are so WAAAAAAAY off of what is happening that it's funny.

First, we have the primesensor-like unit:

And unless you know how to do fourier transforms, I can't even begin to describe to you how it works for detecting 3D. Also, more details about the chip:

Notice the optional RGB, that is because it is using a standard RGB camera like a standard webcam. Instead of just using depth, MS is using the RGB camera to get depth and color, taking advantage of that optional output. From what I read, there is discrepancy on whether they use IR light or light that is considered not IR, but outside the visible spectrum of 99.999% of humans.

Now, does that seem like your normal webcam? Well wait, there's more. There's a camera that detects IR, with a filter blocks out visible light (probably including the near-IR light emitted from the primesense setup). This is to determine whether the object in front of you is human or not, where different body parts are, etc.

Also, random lights in different parts of the room are not going to effect detection. The way you described it made me laugh so hard. The bright lighting has more to do with the noise of the detector. I'm sure it's a very high SNR detector, and that's why it's a low megapixel rating. There's a reason why you can get a point-and-shoot camera with a 16MP rating, yet it is difficult to find a professional camera with more than a 10MP rating. As the light detected by the detector goes up, the blips of noise become large hunks of noise. The problem with your lamp idea is that the lens will focus all of that light to a small portion of the detector and have a little effect on the noise of the rest of the image. It's when you flood the entire room with light that you will have an issue.

Thanks for taking the time to read this and clarifying your BS. Now please stop talking out of your ass. Also don't tell someone that they're wrong when you obviously have no freaking clue yourself.

They have stated that Kinect works in the dark and low light thus it does not track you with the RGB cam.  (Your brain can easily see depth with just one eye so it is possible to detect depth with RGB range.) It's just more simple if you track with only one color ... thus IR light.  Notice the RGB cam is right beside the IR cam; as close as possible so the images are same. RGB cam is to put your image (other images) on screen while IR cam does the tracking.

 This still doesn't show Kinect sees  3D no more that a RGB cam. Thus it misleading to claim Kinect sees 3D as there is difference between detecting depth (one eye)  which Move and TrackIR does with a different method and seeing 3D (two eyes).