Performance v0.17

[maxOh_]

On the list some people already mentioned, that the performance was
not as good. Less then a month ago, i started to develop this project,
so my main focus was only on implementing the functions of openNI +
NITE and that it more or less runs. I didn't spend any time in
optimizing till now. But performance is quite important, so i
understand your concerns.
Like i mentioned before, in the long run i would like to run the hole
lib in an own thread, so the drawing thread is independed and can run
as fast as possible. But this takes more time and i would like to
complete the features first.
For version 0.17, i just clean up the code and optimized the hole
thing, but the lib still runs in the same thread as processing.
Everything should run faster. For example,the time in the draw-
function of SimpleOpenNI_NITE_CircleCtrl(coreI7, 2.2GHz):
v0.16 took ~95ms
v0.17 took ~35ms
It seams that OpenNI uses more or less always the same time(i didn't
check it for every combination). At the end i had ~30ms per OpenNI
update, without the java wrapper code. If you need higher framerates
you have to wait till i implemented the threading stuff, or you could
do the threading by yourself, or maybe not so clean, but you could
just call context.update() not every frame, maybe just every second or
third, depending on your fps.

[scambione]

Unbelievable work here! you can clearly see the difference, no more
lag and the overall performance are better!

this is awesome, thank you very much :)

[Miguel]

Hi Max,

Thanks for the hard work you've put into this. I would like to let you
know that I've tried my code with 0.17 and I got a speed improvement
of almost 400%. To be exactly, my sketch is running at 30fps.

Cheers,
Miguel Peres