i use gloves to make my hands ticker :-) at frames that makes problem i see that the depth data is weak. there few points for my hand.

i agree with you about to know the limits of this system and try to do with that limits.

Oh, I get it now. In that case, that may be a good idea. :)

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

http://www.youtube.com/watch?v=XM9Pvfq1KhE

after i watch this i started to think that what i do is nothing :-)

@Greenlaw. I see that you tilt the calibration board from side to side, I was wondering if that helps because I try to have it static and had three failed calibrations before finally having a good one the fourth time I reshoot my calibration video.

at the beginning i tried to calibrate like you. then i noticed that rotation board like 60 degree calibration cause better result. both kinects should catch variations of angles to set camera position precisely. i swap board like a window open/close.

i am not sure if i am right :-)

Yes, you need to rotate the board between the two cameras. The software uses the differences from frame to frame to accurately locate the cameras. The variations should average out as greater precision in positioning for iPi's virtual cameras. The important thing is to not over-rotate the boards so that it can't be tracked; this could lead to calibration errors.

You can figure out what the safe tilting range is by looking at the depth data as you rotate. Try to avoid having the board go too 'yellow' as this means the surface has become untrackable.

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

Thanks for the tips @Greenlaw I still need to get myself another Kinect camera (still pricey in Australia) but one question, the angle that you place the 2nd Kinect camera what is the better angle 45 degrees or 90 ? Because when you think about it well as far as I have seen when its 90 degrees it seems to track more of your body.

@Mrbones In your videos you are using 46 degrees and as far as you shown us it seems to track great, has there been any problems?

_________________
Cra0kalo
Offical Machinima.com Director
http://youtube.com/user/cra0kalo

Lead of Operation Reaper
http://projectreaper.com

I think 90 degrees is the current official optimal setting for the beta. Unfortunately for me, I can't space my sensors that far apart in my living room. That said, the narrow range I have now seems to work reasonably well (Sister's spinning 'chainsaw dance' for example.)

Technically, you can go full 180 and get maximum coverage now--Mr. Bones has demonstrated this a few times in his videos and it seems to work well. But even at 90 degrees, the cameras can get nearly full coverage. The recent demo video posted by iPi Soft uses 90 degrees to capture a spinning performer and the result seems remarkably accurate.

I haven't tried 90 or full 180 yet, but I'm imagining that 180 may leave more room for error since there is no allowance for 'overlap' in the volumes to average out the data for accuracy, and it also introduces the possibility of slight open 'seams' on both sides of the volume. (Less than 180 leaves a gap on one side only.) Keep in mind that I don't know this for certain, it's just an observational assumption.

Maybe the iPi guys will enlighten us on this topic? :)

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!