Real World Kinect Point Cloud

Here is a download link for a .tox file which outputs a perfectly scaled Kinect Point Cloud. This means you can line it up perfectly in 3D modeling space as well.

It is programmed using GLSL and uses some numbers I found on the Microsoft Kinect forum where an engineer was discussing how they arrive at the real world skeleton point output (as opposed to the uv/image space coordinates).

www58.zippyshare.com/v/78450061/file.html

(file size was too large for forum)

Enjoy :slight_smile:

That’s great, I was wondering how to reproduce the depth with color sample from the sdk.
Where did you find the values you put in the transform2 to match RGB to depth?
Trial and error or are those hidden somewhere in the docs?

Thanks a lot

Also I don’t get the

if ( depth < 1 )
{
output = vec4(x,y,depth[0],0);
}
else
{
output = vec4(x,y,depth[0],1);
}

in the depth2pos shader.
It makes the points with a depth smaller than 1 disappear.

I guess I’ll dig the microsoft sample a bit more.