Kinect dot mesh using MS


I am attempting to get a dot mesh constructed from the 3d depth data sent by the kinect. I would like to be able to sample dots on a subset of the depth mesh to generate something much like Radiohead’s House of Cards video effect for a live gig.

My OpenNI kinect node shows no drivers on the driver pin, however I can see input coming in through the OpenNIViewer tool, this is using version 2 of the OpenNI drivers.

I have tried installing version in combination with vvvv beta 28.1 which I read somewhere was the last version to have these nodes working. This however showed a red OpenNi Kinect node.

I am thus stuck with MS Nodes, from which the depth is made available as a texture and not as XYZ data.

Any pointers on how to get XYZ data for depth readings using MS nodes, or how to get OpenNI to work?

According to this thread: new OpenNI compatible nodes will have to be written from scratch. I have no idea on what coding a vvvv Node is about but am an experienced .Net and Java developer and am willing to help for free to get this going.

That said, any help I can get (or give) in getting this to work is greatly appreciated.

as a .NET developer you should read this page: dynamic plugins

vvvv has an integrated C# code editor which lets you write plugins at runtime. hint: Alt+J to open the project explorer to handle the references.

but then on second thought, you really don’t want to get all the million pixels as points on the CPU and build a mesh, but rather directly move on to GPU and sample the depth texture as translation offset of a particle system: particlesgpu-shader-library