Good find, Snapz.

For those who just want to see the results, here's a direct link to a YouTube video for the sample file. It looks fantastic!:

https://www.youtube.com/watch?time_cont ... KPNE0BTMjE

I noticed a couple of small glitches but I'm assuming this is mostly raw, unedited data. In any case, I think this is very minor stuff to clean up.

The chief advantage with higher res these cameras offer is that they will allow you to record in a much larger space. If you don't have a huge space with controlled lighting, it's probably not worth the investment unless you don't already own compatible RGB cameras (like the PS3 Eye cameras.) The other advantage is higher framerates to choose from, which gives you more accurate captures. But as Snapz already noted, 120 fps requires a ton more bandwidth and disk space, so be prepared to beef up your hardware and network capabilities. (For many common mocap captures with these cameras, I imagine you'll probably fall back to lower res and 60 fps most of the time. But it's great to have the option to go higher when you need it.)

I have a single GoPro in our studio but I can't justify multiple GoPro cameras for mocap...they're just too expensive. However, the sjcam cameras, also recommended by iPi Soft, seem much more affordable, so I'm thinking about it. Very interested in hearing from users who have tried the sjcam cameras with iPi Mocap Studio.

That said, I'm still tight on performance space and will continue to use my multi-Kinect 2 sensors. :p

Later in the fall, when it cools down here in LA, I may finally get around to cleaning out our garage/stage, and maybe I'll start recording with RGB cameras again.

BTW, I imagine mixing different camera brands/models isn't recommended, (i.e., mixing GoPro with SJCam cameras. While mixing cameras is probably possible, it's likely to degrade the final data quality because, even if the resolution and framerate is the identical, the captured colors, luminance and or exposure may be different. Ultimately, you'll be defeating the reason for going with a higher quality device. Whichever camera you choose, it's probably best to stick with the same make/model. That's just my opinion though...curious to hear official info from the devs about this.

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

Watch a one minute excerpt on Vimeo now!

...

Seems it has been out a few days on YouTube, so I figured I would post it here instead for those not alerted to it.

Agreed, if going to go this route, would be best to use all same make and models of cams, better chance of all videos looking closer to each other.

It is definitely a lot more complicated to work with, I did run through the entire process myself with their files, but I think one, the cal-cam2.mp4 file is bad, it wouldn't convert as it should for me without throwing an error at the end that it failed, so I did some finagling and edited the mp4 file a bit shorter and re-saved it as the same, then it converted ok to proceeded as they describe it and it worked, so a plus there :)

I am not sure, but the space they used doesn't seem much more than PS Eyes will cover and still track, they didn't state the actual L & W of the area used, but the resolution is better for sure, even at 720p/60 fps and smoother than PS Eye recordings, but I really think a smaller capture volume would work out better, maybe just 4 x 4, or 5 x 5 m, or even 4 x 5 m, just judging by tracking the video, I don't know how they had the camera focal point set either, but it was definitely wide view, maybe 150 degree, or more.

There may be several drawbacks, but they are the first to ever even attempt it this way, so kudos for that! It might give those desiring this ability something to work with.

Another would be there is no way to use Move controllers, I know that is why there is some glitchiness in the head and arms on their video even after whatever clean up they did, as it was just really a basic test video for them too I am sure.

I did track the video in v4 Studio, but to me the environment was clashing with the hands and the lighting was a bit poor, way to dark on the one side causing the tracking to throw a lot more errors than it should, but I worked it out with camera switching and Actor color changes to get through it reasonably enough, but I didn't go crazy working with it, was just testing the processing myself with what I had to work with, but no where near as smooth as I can get with a lot less work using PS Eyes.

I would like to see a 720p @ 120 fps though once they get around to making one, I don't even think you would need the 1080p at all.

The file size after conversion was 170 MB each, all 4 totaling 680 MB, I know much lower than the over 1 to 2G recordings I run with 6-PS Eyes anyway for 1+ minute recordings with using JPG compression, but I can do 30 secs @ 500+ MB

Download just the videos, cal file and project and run through the tracking yourself and see, they set it up for easy access that way too, you don't need to do all the self calibrating and such, I just did to see how it all worked.

Thanks for setting up a test iPi, hope you can put out a couple more in a better environment and higher frame rate.

...

...

Update: calib-cam2.mp4 file download is working now.

I am not sure why it threw the conversion error when it was downloaded the first times, could have been the download was corrupt, or maybe the upload was corrupt, but I did download it twice from the link supplied the first time, to see if the download got corrupted and got same result, maybe iPi re-uploaded the file, or it was something else, never heard the cause, or know why.

I re-downloaded it today, looked like the same video, using the recent Recorder v4 update and it converted properly from the same folder I have the rest stored in, so all seems a go for full testing of the entire process by users.

Be sure to use all the suggested frame marks they list, or you may probably not get things to line up well enough for test tracking.

...

Yes, the file was corrupted at the server and we have re-uploaded it. Thanks for reporting @Snapz ;)

...

No problem, I just did a quick work around edit and it worked, but glad it is fixed now for other users.

BTW, Good work on the descriptions for the full processes with all the links now active.

It seems like a lot, but once users get to go through it a few times on their own, with there own set ups, referring to these docs will help, at least until they get their own process that works best for them.

I also see you included the capture volume used @ 4 x 4 m, it looked a little more oblong in the videos though, maybe due to the wider view.

It shows the camera FOV set at 75 degrees by default, should this be changed to whatever FOV the cameras are set to, I didn't know what FOV iPi used in those videos, it just says wide, so I left it at 75, but I am guessing this could have been incorrect, if the FOV was say 150 degrees, or is this determined automatically during the Optics calibration? I didn't change it and the recording looked appropriate once in Studio though for tracking.

The pixel shear/pixel aspect options I am guessing would be best to be left unchecked for most action cam models.

All steps worked as described though, it is important to do it step by step as described, not something you can just do in any order, especially while recording, or you may just get one big loss of time.

I will probably try this process using PS Eyes, without the Optics procedure, just to see how it functions.

...