Hey,
I'd like to ask around, if anyone has any practical advice on reducing the latency when drawing objects on the screen that require sensor data (e.g. skeleton, pointcloud, ...).
Since our "recommended/supported" development platform is Java+Processing, relevant advice would be most appreciated, however, any constructive feedback is most welcome!
On the software-side, one possible thing to consider is the level of coupling between the code to draw/display something and the code to calculate/process the sensor data: i.e. code for movement calculation can be executed separately (in a thread), enabling us to fire events independently of Processing's run-loop. (Beware that this can reveal potential issues with thread-(un)safe code. See, how to make the relevant methods/properties synchronized, if you start getting runtime errors [thanks, Sergej].)