Here is what I've been thinking: I have access to a completely white background.
If one of my actors is wearing a black, red or blue costume ( to make it "pop" from the background) and the other one is wearing a complete white costume, with mask and gloves, could the kinect or pseye "ignore" this white actor and do the capture one by one?
To my mind, the white costume over the white BG will create a "camouflage" to the camera.
Then, I'd capture the movement of the main actor (black) in both instances of the choreographed movement.

Here is an photo of my background and the costume.

For Kinect, the actor's color means nothing, because it is sensing the depth data. For PS Eye, this may replace the secondary actor with background, but won't add any additional information for the primary actor. So the accuracy won't differ from the case when the secondary actor wears any other color different enough from the primary. And the latter may be preferred when trying to capture the actors one by one from a single shot.


This is the most questionable part. Because you barely can do exact copy of the movement in the second instance, and that totally ruins the synchronization between the characters you are trying to achive with multiple persons capture.

I know we are roughly about one month of the Multiple Person Tracking release, but in case there is any other delay, I'm trying to create an alternative to the process.
Here is what I've been thinking:
In my animation, my character don't actually touch each other. Only their swords do. So I'm planning the following:
Record the action with ipi recorder 1.8 and save the file as an .avi.
Open the videos in a video editing software and individually removing the actors from the view.
For instance: I'll record the videos with 2 actors. Both of them doing their respective T poses. Then in the video editing software, I'll erase one of the actors, (whether by blocking them from view with a square box or by rotoscoping them out of view.). Then, I'd process the video with just one actor in Ipi DMC. And repeat the same operation for the second actor. I know it is a very tedious workflow. But, could it work?


If it doesn't work, can I be 100% sure that the Multiple Person feature will be released in May?

I do not think manual removing of one actor from the video will give any improvement on the tracking of another actor, provided that actors' clothes are different.
Have you ever tried to process recordings with two actors one by one without any preprocessing? See at this example:
http://www.youtube.com/watch?v=dS9VVGAP_Us


Sorry, no 100% guarantee :( Believe me, we too are stressed with this situation of constant deferment of this feature, and are doing all to make it ready sooner than later.

I had seen this movie from Manymotion before, but what worries me is one character occluding the other one. I mean, in a full circle setup, at least 2 cameras will be behind each actor. Wouldn't it be an issue if the character is "covering" the other from the view? After all, both of them should be visible in all cameras all the time, right?

Thanks for your honesty. I understand the difficulties you must be facing.

Of course, it's an issue, but maybe not so critical. Anyway, removing the pixels of the second actor from a video will not magically bring back the pixels of the first (obscured) actor needed for tracking.

Hahaha obviously it won't bring the pixels back. My only concern was the program not recognizing each individual actor. But the contrasting colors might solve it. Thanks!