You know the depthmap of the kinect. Points in space that represent what the kinect can see. Based on that is the skeleton and further processing.
The kinect is able to separate the user from the background, so all you see is the user's depthmap. The idea is as follows:
Combine the functionality of faast with the depthmap of the user. For instance a user moves his leg forwards and by doing that the whole point cloud moves forward. As the kinect sees only the part of the body being shown to it, it will create a half body out of points moving around. You can map on that the texture of the person using it but just the (dyed) points would create a more interesting atmosphere.
Of course, you must not be afraid of being seen the way you are.
This becomes very interesting when multiple people are moving around and see each other moving in 3D-Space - or better - see ghosts, as they are not walking but moving their leg forwards and are being cut in half - but if you stand behind another person, you see the silhuette from the back, the negative shape.
If the user makes certain movements, you could surround the ghost with a halo or make it change color. Or if the user uses voice commands.
You could do this in a huge virtual world, having virtual interaction. You yourself are always in first person perspective.
I am writing this cause I have a bohemian approach to life: I do things and think later. Unfortunately using this approach most of my works render themselves useless once finished. So I need suitable feedback. What is not suitable is feedback like "I'd use it" and once it's finished "Nah I wasn't so serious".
Just like this but without levels and color, just the points:
[media]
[/media]
Or picture this at 0:52 still walking around like the above:
[media]
[/media]
or just like this:
[media]http://vimeo.com/17552930[/media]
