You have 10. 10 Kinect v2s. 10. You have 10 pcs and one Kinect per pc. I've dreamed of a 16 PS3 eye setup which I know would require more than a couple of laptops to pull off, but Kinect v2s I have 2 myself. It's awesome for one actor and I don't see myself needing more than 2-3, but 10? I need to back-read some of the posts here about this. I'm curious. Just don't have that kind of $ for a 10 Kinect setup

...

They seem to want to try to cover a larger area, and more actors, which 2-3 Kinects set up won't really do, because of limited depth of field.

In order to spread out the coverage and still be able to use it as a single scene later, I was just suggesting the prior method, but I agree it would cost a bit, but seems they have the sensors already.

If they can get the right modem connections and have enough computers that will each handle the Kinect v2, it would be cool to see it achieved and have less requirements than multiple PS eyes.

Definitely not really feasible for the general user of Kinects though.

...

As an artist who normally prefers using multiple Kinect sensors, I have to agree with Snapz and others here. If you actually have the space and hardware to support 10 Kinect 2.0 sensors, you will be far better off using multiple PS3 Eye cameras instead.

I only recommend multi-Kinect for its convenience and when you have a 'small' space to work in--like a small studio with one performer. I like the multi-Kinect system because I can set it up and take it down in a couple of minutes on a whim, and it doesn't require special clothing or lighting. Since I typically record in the living room of our house, being able to set it up and break it down again quickly is a necessity. This process was far more complicated back when I was using PS3 Eye cameras in the same space. (This was mainly because I also had to set up two large soft boxes to get even lighting in the space, and wear special clothing. Also, once setup, the capture space was actually smaller than what I currently get with Kinect in this same room.)

However, for more advanced capture scenarios (i.e., multiple actors, limited interaction, fast motions, and large capture areas,) Kinect 2.0 is not going to cut it. If you have the space and appropriate lighting, you definitely want to go with a multi-PS Eye setup. With good lighting, you should be able to record better quality motions too.

Good luck with whichever system you decide on.

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

Watch a one minute excerpt on Vimeo now!

We want to separate the actor group, at the same time to record, so we need big scene in order to capture animation, not for the role of a single service，sametime is the point，so we need to record multi Kinect 2.0 at the same times ，thanks everyone who love this POST and ipi studio motion capture software，thanks，Maybe some plugins can help but still hope to provide the official functions rather than borrow other plugins，IPI studio motion capture software is very powerful，thanks

...

I understand, that is why I suggested the above set up, using 2 iPi Recorders on different IP Gateways, and pressing record on both at the same time, will be like 1 bigger scene, more separated performers, but they could still probably cross into the other recordings view.

Processing the footage would be different, but I am pretty sure it would work fine once done right and spliced back together in post.

Gotta think out of the box if you want to use multiple Kinect sensors above 4, because I doubt the software abilities are going to change anytime soon to meet your needs.

Not sure it is feasible for you, but an idea.

...

Hope the official will not set limit and the limitation of the software’actors, adding THE ACTOR according to the demand, so the software will be more better and The gap between ipi studio 3.0 and Optical capture or magnetic capture will be reduced，thanks，I hope ，maybe everybody hope so，some advice

...

Have you even tried tracking more than 3 actors as individual single actor projects, with your current set up? You can record the depth data from more than 3 people in the scene, that's not the issue, but that doesn't mean they will all be able to be tracked optimally, and you will see where the occlusion issues will come in, this has nothing to do with being able to select more actors at once in the tracking process, it has to do with limited space and occlusion.

At the tracking fps you get from trying to track 3 actors at once, it may be faster to just track them as single projects, (3 or more people recorded in 1 scene, but only track them one at a time from a new video each time), since they need to be exported as single files anyway and you may be able to fit one or two more performers in scene (maybe), but you will never be able to increase the maximum capture volume using Kinect sensors, no matter if iPi allowed 10 sensors to be used at once on one recorder, or allowed more actors to be tracked at once.

You can also try recording more people in one recording, then use the 3 actor tracking on just the first 3 in one project, then re-track the video again, selecting the other people in-scene, but you will still not get the results you think you will.

The limitations aren't really caused by the developer, the Kinect sensors have a limited capture volume and FOV, basically a maximum of 15 x 15 ft. of recordable area data, (and that's only when all pointing in a semi-linear direction, basically how you set them up), because you have to be a min of 5 ft. from the sensors to capture full standing data, maybe closer if in a seated position closer to the sensor, but based on the config you have shown, you still can not go further back or to the sides than the FOV will allow, or no data will get acquired (yellow areas) and the performer will disappear into that area also, so no data received by the sensors, means untrackable.

I am sure you could attempt spreading the sensors further apart and see if they still track successfully, and can be walked between, but I don't think you can get 3 or 4 sensors calibrated correctly that way.

iPi, for good reason, has concluded that 3 actor max. (for their selection amount at one time) is a respectable limit to avoid occlusion issues and the limitations of the Kinect sensors.

Since they want their program to run for users with as little error as possible, they are pretty well bound to this actor maximum tracking at one time, but that doesn't mean you can't do a work-around that may work.

If you switch to PS Eyes, you would have more freedom of camera placement, where Kinect v2 are also bound to a level plane set up, PS Eyes are not, but they aren't as easy to use or calibrate when many are used and also track at a much slower fps in studio, even with a very powerful machine.

Only iPiSoft can answer your request to your suggestions, but I think they have already saying, "We do not plan any research in this area for the near time."

Good Luck in your endeavors, hope it all works out well, no matter the final choice of system chosen.

...