before I go crazy spending a small fortune renting another professional package, I'd like to know if anyone has had good experience recording no less than 2 actors using only 2 kinect V2 sensors. I've done some distributed recording of myself which works wonderfully between 2 laptops with one Kinect V2 per laptop. I know that there are going to be moments that I may want to have interaction and I only got that working with 1 laptop with 8 PS3 cameras for about 5 seconds. I'm afraid to spend so much on something I can't try out so I'm wondering if anyone has had good experiences with 2 Kinect V2s for 2 actors at the same time.

I have a short narrative in mind that is definitely going to require interaction between actors physically (as opposed to mentally, lol). What is the best setup. Do I really need to go back to PS3 multi camera setup? I wish I had the space, but not sure that I'm going to have it and the Kinect V2 does a great job for one actor at least

...

Dual Kinects still require a minimum space in between them to even get a practical capture volume for one performer, whether 90 degree, or 180 degree set up.

If you barely have the room for one Kinect, dual won't make that any better.

So 2 performers in the same scene, cut that space in half again.

Really, depending on the actions, your performers never really have to touch, they can act out, or react, even one at time in scene, one performer off camera, just doing motions for the person on camera to react to.

The files are exported as separate files anyway and using Kinects you are gonna have more clean up to do anyway, so if just limited 2 person interactions, you are going to need 2 Kinects for better results, but for capture volume, you will need to improvise.

Hope that makes sense.

...

I get what you're saying but I'm not exactly sure I agree that 2 characters who engage each other physically can always be done separately. That said, in the ipisoft gallery there's a guy that did a video game and used ipisoft to act out (he says) 100s of assassination acts and death acts and then stitches them together. In a live performance of both at the same time, I think you're going to get a different chemistry. Two dancers for example. I do think that I'm going to want two actors at once because without the real-time interaction, cleanup will be nightmarishLy difficult.

Dual Kinect is the only way I'll do anything. I've tried with one and there's gonna be lost information for tracking if I don't see front and back side of my actor...and they will be turning around. Here's a sample I tested in my company hallway: https://youtu.be/lWnIzdR_9RA

I agree with Snapz. If you're going to record multiple actors physically interacting, you'll need more space and very likely more than two Kinect sensors...and if you have a big enough space, you may as well use PS3 Eye cameras with at least six cameras provided you can meet the other criteria for it. Unlike with single actor recording, you need a lot of extra coverage when simultaneously recording multiple actors to avoid occlusion issues.

Some tips: For our 'Brudders' productions, I don't bother with multi-actor recording. I usually perform all the characters myself and I perform to a looping audio track to keep 'multi-character' performances in sync during recording.

It may also help to do a short countdown on the audio track and perform a hand clap to sync the performance--although, TBH, I haven't bothered to go that far for a while.

Usually, just performing all the characters to the same audio track is enough. You can line the motions up in your animation program by slipping the animation. That's basically how it was done for Happy Box and for the music video where still working on.

If there's a little physical contact involved (like in Happy Box) you'll probably need to do some pose corrections in animation. Depending on your animation program and rigging setup, this should be fairly easy to do.

However, if there's a lot of physical contact, like in a grappling fight scene, multi-actor recording might then be worth pursuing. But even with a successful recording of such a scene, you're probably going to be facing a lot of post clean up work, especially where the arms and hands are concerned.

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

Watch a one minute excerpt on Vimeo now!

...

Yes, couples dances would be an exception to separate performing due to the feet placement to be accurate, or any motions like grappling wrestling moves as G stated, I haven't seen anyone do a video showing that, but I am sure it's possible.

I absolutely guarantee for many fighting actions, or many other motions, the actors never have to make contact and with some improvisation once put together and adjusted in a scene it will look fine, but there are exceptions to every possibility.

Good luck, it would be nice if you do attempt it and it works out for you, post the results if so :)

...

if I could use the PS3 eye cameras with enough room, I'd do that every time but unfortunately space has been an issue for me for some time. I actually have 12 PS3 cameras but I don't have a pro package anymore and didn't have a strong enough laptop to support more than 6 with stability. Now, I have 2 laptops and may one day soon get a 3rd for distributed recording. What I can't do is invest in the pro package until I know for sure what my film is...and I do have an idea...but need to storyboard it, Vo record in a sound room, and have me and my 2nd actor who will play out different parts with me. It's been a long year of learning and loving ipisoft for what it can do while I learned what the limits are with hardware. Presently, distributed recording between 2 v2s is the best and easiest setup I have done. I'll surely post and share when I have more to tell, but for now I need to really flesh out my narrative. Also I'm getting married in a few months so there's that to do too:)

I understand what you mean. I have six PS3 Eyes and four Kinect v2 sensors, but two to three Kinect v2 sensors is what I usually use.

The reasons are that it's more convenient to record in our living room with Kinect but 1.) the space is only big enough for up to 3 Kinect sensors and that gets a bit cramped, 2.) my current networked computers are only capable of supporting three v2 sensors at top speed, and 3.) if I want to use PS3 eye cameras, getting even lighting becomes an issue in this space.

Actually, I can get proper lighting using multiple soft boxes but that becomes a hassle to set up every time I need to record. It would make more sense to go back to our garage for that kind of setup but then I'm back to lugging a bunch of computer gear back and forth, and don't really want to go back to that.

I guess I've grown too comfortable with my current five minute setups using dual Kinect. :)

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

Watch a one minute excerpt on Vimeo now!

the hassle of SPACE and the right conditions is more difficult than I thought it would be. given my space, light, and the need for quick setups, breakdowns, and good cameras, the KinectV2 has been my top choice. more data in the point cloud than V1 so less cleanup. PS3 simply does the best job for fast motion, but so many conditions play a hand in having space, right clothes, right light, and I don't see myself using soft boxes unless I have access to them....which I could, but then we come back to the space....and that hallway is about the best that I can do @ work. Away, and around home, I have access to a photo studio, but I can't touch the lights:)