Is there any ideas to make a rough depth-image like from kinect, by 21 hand skeleton points?

I'm now working on some hand gesture recognition stuff, although, image processing really hit me.

I wanna to create an depth image matrix just by the given 21 hand skeleton points(3 dimensions), makes it looks like a depth data from Kinect, as shown below.

After checking the internet, I found some useful matlab library like Libhand, Syngrasp. But Libhand doesn't work in Windows and Syngrasp just do plot work.

So, I may wonder is there any excellent,brilliant ideas or thinking to create such image matrix according to given points?(I think Gaussian distribution could be good to make that gradual stuff, in the centre of point)

Of course, I would appreciate it, if some simple code work or pseudocode could be provided.

Thanks a lot.