generate 3d model/avatar with Kinect

Is there any way to capture point clouds and 3d skeleton from Kinect v2 and use it as an avatar inside Openframeworks? Specifically, for an art project, I want to capture point clouds from a performer, then try to move its hands myself.


welcome to the OF forum!

Sure you can. You have to do it in windows though because its the only platform where you can get the skeleton.
saving the point cloud is super straight forwards. you can either just save the depth and rgb images from the kinect and the skeleton data for that frame or preprocess the pointcloud. I thinkk that the examples for the kinect addons have the pointcloud done, then you just need to save it. ofMesh allows you to save in a basic format, but there are several other addons that allow you to save into more advanced formats. check for more addons.

I am courious, I am using your ofxNI2 addon in Linux (after compiled libfreenect), and I can get the skeleton. Is there a limitation I am not aware of?


WHat do you mean?

OpenNI’s skeleton tracking is super old and has a lot of problems that have been addressed in a much better way in the Kinect SDK for windows.

Both, but mostly the second, the least noisy as possible.


I can not really tell. I have not used it in Linux and I have not really used that addon for a lot of years. Although, if the library and addon run it should be quite decent.
As for stability do you mean as if the addon will crash or that the tracking and skeleton are stable (not noisy)?

I think I did a mess when responding to this thread, sorry if it gets barely understandable

It is not a problem at all, neither a mess.