I am planning a project using Kinect's depth sensoring but I do not have a Kinect yet.
Newest information from Microsoft's website says it can take over 4m, and Wikipedia page says over 6m. What I want is about 20m, is that OK for Kinect? If not what is the bottleneck, SDK or hardware?
Error will increase dramatically as you go farther out. At 20m, I don't think you'd get anywhere near the quality you need to run any kind of application reliably. Keep in mind that the resolution of the camera is only 640×480 pixels. Things farther away are smaller, and harder to discern. It's a hardware limitation. You will likely find it very difficult to use if you're expecting results at 20 meters.
See this blog post for more information about the limitations.
