Kung fu live has to be played with no movement in the background. So you gotta turn that fan off and kick your friends out of the room.
why no movement in the background? because its not tracking the player its just tracking pixel changes... it cant tell your fist from your dog running across the room.
markinthedark
Unless the developers were making up things, they still create bounding boxes. Saying "it's tracking pixel changes" isn't a bad thing. Kinect is only tracking pixels, but has the advantage of the color being the distance. Finding edges based on movement is much easier than finding them in a still image. So they both have much in common, as in "looking at lots of pixels and trying to figure out where the player and his body parts are". Even Kinect will struggle when your dog is running across the room if he does so very close to your legs (or worse, between them).
Of course 2D is much more limited and way easier to get confused, since everything is ending up on the same plane. But they still use bounding boxes and once you have such a box around every area that changed, you can find the "root" (in this case the face would be the obvious choice) and find boxes that are connected. The game apparently stops there, but from here on it would be simple to find the joints and do a "2D skeleton".
Kinect, can be played in a room with an entire party going on in the background because its actually tracking a 3d image of your body and not simply pixel changes.
In a party environment you constantly have people running in front of you and even Kinect doesn't have xray vision. People brushing past you or standing real close are pretty likely to confuse Kinect as well (heck, just crossing your arms or lifting them up close to your head has turned most tech demos into glitch fests). Don't forget that this "3d image of your body" is just a bunch of pixels as well and not magically filtering out everything else. It's just a lot easier to know which pixels belong to a body when you have depth.
Kung Fu Live is very limited, but what their software does is still a simplified version of what Kinect does while being limited to a 2D image. Not saying they could pull off Kinect if you gave them a depth camera, seeing how they seem to be small studio and RELIABLY detecting a players body/limbs is quite a task (which based on the tech demos not even a big player like MS was able to do 100% yet).
Log in to comment