garageCube

Is there anything I can do to minimize the delay?

Used Intel Core i7, 2.2 GHz, Thunderbolt, MAD_KinectMasker 1.0 -> Quartz Composer 4 -> Modul8 2.6.4 -> MAD Mapper 1.1

I'm trying to do a follow spot type effect for a live performance.

http://www.youtube.com/watch?v=y8CpinnSNDU

This was my first test and I will likely do something along these lines with more of a large blur. Mostly because the tracking delay is very pronounced when I go tighter.

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

Hi VJkg,

don´t you think that the delay is ok ?

is this for a club ?

everybody that dances is moving, so i don´t think they will notice the delay.

if it´s for a performance, i don´t know? to me it seams like nearly no delay. 2 or 3 frames ....

have you asked ilan?

pele2010

(i like the dance! )

pele2010 wrote:Hi VJkg,

if it´s for a performance, i don´t know? to me it seams like nearly no delay. 2 or 3 frames ....

Live performance and I want to track the 2 lead singers. I actually used to do it with the mouse but I could only track one at a time.

pele2010 wrote:have you asked ilan?

No.

pele2010 wrote:pele2010

(i like the dance! )

That dancing is one of the reasons I do visuals. If I'm in the booth I don't have to try and pretend I can dance. I dance with Modul8!

hey i know...

keep smiling ....



rock on and ask ilan.

here his 1. entry on modul.us

http://www.modul8.us/?m=200701

pele2010

hi vjkg

If you want to track 2 persons, don´t you need two kinects ?

with one kinect you get only one mask? right?

pele2010

hey vjkg,

the other specialist is miguel espada.

this is his webpage.
http://www.miguelespada.es/?cat=41

this is one
http://www.miguelespada.es/?p=538
this is two
http://www.miguelespada.es/?p=579

but maybe you contact him.

he seems to be very nice.

i hope this helps.

pele2010

pele2010 wrote:hi vjkg

If you want to track 2 persons, don´t you need two kinects ?

with one kinect you get only one mask? right?

pele2010

I'm not tracking. I am masking any object that comes into the zone I set. So far the way I am doing it takes up two layers in Modul8 but it was the way I got it to work. Also my wife is pregnant and not wanting to come running every time I do some silly Vj thing so I have not tested 2 objects yet. Although maybe I'll try tonight with me and the high chair I was using for calibration.

I did not test but remembered all the tweaking I had to do to get and objects out of the frame. With my technique any object in my set zone becomes the transparent. I then put another layer behind the transparency. This layer generates texture and color and can be tweaked at will.

OK, so I must state how much I love and hate Quartz Composer. I love the power and options. I hate the way it never quite works for me. I used it for the above demo and using Synapse for skeleton tracking. I was able to get my latency issues greatly reduced using this method. My issue is that Modul8 just shows my qtz as a white image. I worked on this one issue for 2 hours. With much help from other tutorials I got the tracking implemented in under 30 minutes.

Is this one of those unsafe patch issues? Did I not turn some obscure thing on in the end of the chain? Can I send the OSC data from the Kinect -> Synapse -> Quartz Composer directly to Modul8 for tracking a layer?

With my original idea, even with the latency, setting the min and max range made everything in the zone I picked get light. With the new one a person must stay in frame at all times and if the leave for to long they must recalibrate (see image below). This might not be the best thing for a performer on stage.




I am working my dilemma from all ends testing what works for me but does anybody see something I do not in my setup?

So here is the final product from my Kinect masking follow spot tests. I was alone and had done no site survey. I had 2 hours before show time. I love being able to create on demand.

http://www.youtube.com/watch?v=xDZ9Q4CjtoI