I am building a visualization system to work with joint data from the Kinect. Initially I did everything by patching together jitter objects etc, but that quickly became rather cumbersome. I figured that I could do much more interesting and complex drawing operations using scripted drawing commands, in my case [jit.gl.lua] (I chose Lua because it's easy to learn, looks clean and simple, and is purportedly faster than Javascript). I also intend to create in interface for quickly creating graphical elements tied to arbitrary joint data (e.g. in [texedit] one could enter "poly lefthand leftelbow leftshoulder" and create a polygon using the position of the entered joints). The problem is, there's a lot of data that is being passed into jit.gl.lua (3 floats x 14 joints, every ~30ms) and it crashes after a minute or so. So my questions are: 1) How does one use a scripting/procedural language in Max to perform drawing instructions based on incoming data (such as from sensor devices)? Is there a better way to access those values than inputting them into the 'language object' itself (and storing that in an internal list to be referenced when draw() is called)? (e.g. referencing an external matrix of those values?) I know it's not good to cross that 'barrier' between Max and Language-X too much. 2) Am I significantly better off doing this a different way entirely? Creating an external in C? Sending the Kinect data into Processing etc.? Sticking with straight patching of jitter objects?

Any pointers to resources on this kind of interactive graphics would be very helpful. My searching hasn't turned up anything specific enough for me to find useful. Thanks!