Kinect depth to Polygons


I projected the Kinect’s depth information to a polygon grid but my frame rate doesn’t go above 30fps. Using a mesh is faster but there isn’t much you can do with a mesh. The same goes for a GLSL shader: the new vertex points are not applied to the grid baring me from doing further effects.

Is there a more efficient way to do this?

How can I delete the faces connecting the user to the back wall of the grid?
kinect_polygons.toe (10.7 KB)

One thing you can do is stop exporting channels from info1 onto grid1. This export is causing grid1 to cook every frame which seems to take 3-5ms on my computer.

The other thing that it killing it here is that the Delete, Facet and Primitive SOPs are filtering a part of the SOP chain that is cooking every frame, so they also cook every frame as well. If you can delete earlier in your SOP chain where SOPs aren’t cooking, like right after grid1, then the Delete SOP won’t cook every frame as well. Generally SOPs are very hard on the CPU when animating dense geometry. You want keep them from cooking whenever possible. I don’t know what you are attempting with these SOPs however, so I don’t know if it is feasible to move them for your setup.

A GLSL shader should be much faster, though harder to setup.

Another tip, don’t use a Material SOP, generally much better to apply the Material at the Geometry COMP level on the Render page of parameters.

Thank you Ben. I’ll try to apply the SOP effect earlier to the chain.

What about the long polygons connecting the user to the grid (from the edges of the user stretching all the way to the back). What’s the best way to remove them?