Microsoft’s Kinect for Xbox 360: Quick Review

By Soumitra Mondal, Gaea News Network
Friday, November 5, 2010

You can now experience complete entertainment with instant fun as Microsoft has revolutionized gaming with the release of Kinect for Xbox 360. With Kinect, you don’t just play the game; you become a part of the game. Kinect’s motion sensing camera will conduct full-body tracking to put you in the center of the fun. Its amazing technology allows the sensor to recognize your body and your voice, and mirror your movements in the game, making you the controller.


Simply place this small bar that sits on a small stand beneath or on top of your television. Within that bar are a couple of cameras that track your every move. This device can also help you to chat with your friends, when you are not watching a movie. Here, the real excitement is not in video conferencing or watching movies, but the number of games you can play with it. Microsoft had already promised over 15 titles already for the launch day and that don’t only include a selection of games, but much more, ranging from racing games to fitness, to playing a Jedi knight in a Star Wars spin-off.

Microsoft predicts Kinect would sell 3 million units this holiday season. The stand alone pricing for Kinect for Xbox 360 is $149.99. This is not a sky high price tag if we consider the $500 million the company is spending on its massive marketing campaign. The device is now available in North America and if you are in the UK or Europe, you can have the fun a few days after. However, one thing is for certain, you will get to play it in your home before this year is out.

Filed under: Gadget

Tags: , , ,
November 6, 2010: 5:28 pm

You are all going to be very disappointed. Everything clever the ‘kinect’ does, it does with proprietary software on the xbox360. There is no depth camera, just an ordinary monocrome webcam reading a projected (fixed) infra-red pattern projected on the scene. The main purpose of this camera is to rapidly identify a silhouette of a human. Movement is largely identified by using statistical motion vector software (a technique used for years by previous console webcams). The monochrome camera allows the software in the xbox to ‘know’ which small part of the colour camera image to process when searching for motion vectors.The projected IR pattern allows crude identification of z-depth motion, but the resolution of the mono-camera should be a clue as to how limited this data is.The skeletal tracking is largely an AI system on the xbox360, and is a statistical assumptive algorithm, rather than any absolute measurement. Research into calculating limb position from simple body outline images pre-dates the xbox console by a very long time.Interestingly, several years before the kinect, MS gave massive promotion to the Codemaster’s game ‘You’re in the movies’. This game, using only the standard webcam, identified the silhouettes of the players in front of the camera with a high degree of accuracy, allowing ‘green-screening techniques to do background substitution, without the need for a green-screen. This shows that much of what the kinect does in hardware was already redundant using modern image processing algorithms.However, the kinect obviously makes such methods far more robust, at the cost of mechanical complexity (and a big spend by the consumer).Had MS been serious about general reading of z-depth, it would have deployed a 3 camera system, with the left and rightmost cameras being discrete and individually positionable, like speakers. These 2 would provide a ’stereo pair’ of images that would allow depth to be identified at each pixel position by using perspective variation algorithms that could be simply accelerated in hardware. However, setting up 3 discrete boxes would have been a pain for most consumers under 10, or over 20, and lost MS most of its intended audience.As an aside, many of us know that the companied MS allied with to do the ‘depth’ system was supposedly just about to launch a cheap z-depth camera for the PC at the time. However, I’m sure that camera was intended for objects way closer than 7-feet away, where the sharpness of the projected IR grid pattern would have returned much better information. Can the kinect work with objects placed much closer? Probably not if one wishes to use the colour camera as well. Then there is the focus issue of the optics, and the ’sharpness’ of the IR pattern.The bottom line is that the kinect is not like the wii-remotes and sony-moves giving us access to remarkably cheap and robust combinations of gyroscopes, accelerometers, and cameras. For visual processing, kinect is 90%+ a software system on the xbos360 side. Kinect is a bunch of simple hardware choices designed to assist the software. And what does that software drive? Largely a load of silly, imprecise casual games for people that can’t even bring themselves to take gaming seriously. That should inform you about the likely engineering choices, and their usefulness in other areas.Believe me, motion tracking studios won’t be replacing their multi camera setups, and body-suits, with anything like the tech in the kinect.PS like everyone here, I hope the kinect is turned into another ‘open’ USB device for everyone to exploit, as soon as possible. It is just that for visual image processing, it is already cheaper to buy multiple high quality USB cameras, and infra-red LEDs, and experiment with readily available open-source software. This just was not true of the tech in the wii-remote.

will not be displayed