Motion controller gloves - sample
Page 2 of 2

Author:  Snapz [ Mon May 29, 2017 3:33 am ]
Post subject:  Re: Motion controller gloves - sample


Nice, look a bit more breathable also, should work well.

Style of board shouldn't matter, as long as it calibrates correctly and connects without issue, there are different models, but some are after market knock offs that don't have the SONY logo on the cover, I am not sure if Sony changed the board design in all these years, but possible, the right glove is the board design I have in all 4 of mine.


Author:  Greenlaw [ Mon May 29, 2017 7:19 am ]
Post subject:  Re: Motion controller gloves - sample

FWIW, I believe you're meant to leave the Head Tracking option disabled if you intend to use the Move Controller data for the head bone. Otherwise, you might be applying double-rotations of head motion data on the head bone?

TBH, I'm not completely sure about that--the Move Controller data application is possibly additive but Mocap Studio might also just replace the data completely without adding rotation. Sorry, I haven't actually checked this but perhaps somebody from iPi Soft can confirm?

My understanding is that Mocap Studio's Head Tracking option really shouldn't affect the spine--there is a different spine bending option for that. If the native Head Tracking option does affect the spine then I think this might be a bug.

I'll be working in Mocap Studio today and will check the above, but hopefully, somebody from iPi Soft can comment on this too. (Thanks in advance guys.)

Author:  Snapz [ Mon May 29, 2017 7:27 am ]
Post subject:  Re: Motion controller gloves - sample


Well, I haven't tried it recently, but I know for a fact using a head Move without head tracking on used to cause poor spine bending, at least with recording with PS Eyes.

Adams video doesn't really show much spine bending, and I don't know what spine bend option was used on it, but looks to me like only the very top spine link was moving when he bent his head down, so maybe stiff spine was set.

Yes, the Move data will over ride any head bone tracking data anyway, but it doesn't correct any spine rotations or bending issues, it only applies data for the head bone, but for some reason used to cause the mid spine to lung forward on bending and I think it still does.

It would be good if you could test this with all the updates that have occurred since I tested it, used both ways and report, even though you would be using Kinect v2, it would be nice to know for those users, as I have never tested it with either version of Kinects.

Also when test this, import a character and see the effects on it also.

Someone from iPi is welcome to clarify this as well.

Here I found the post where I discovered the issue and ran the files he posted myself to correct the spine issue:


Author:  vmaslov [ Mon May 29, 2017 10:10 am ]
Post subject:  Re: Motion controller gloves - sample

Switching the Head tracking option determines whether Neck and Head bones are fixed during tracking. Naturally, this has some influence on rotations of the spine bones. Because a character's pose is optimized as a whole. I guess this influence is more noticeable on motions having high flexion of spine, which are common in Snapz' dancing projects.

Motion data from controller is applied only to the Head bone, overwriting its rotation - whether it has been previously tracked or not. It does not affect other bones in any way.

Author:  Greenlaw [ Mon May 29, 2017 11:33 am ]
Post subject:  Re: Motion controller gloves - sample

Ah, thanks for the clarification. It all makes more sense now. :)

My confusion came from not realizing that native head tracking also affected the neck bone. I assumed it didn't do that because applying Move Controller head tracking data only affected the head bone, not the neck.

Author:  Snapz [ Mon May 29, 2017 12:34 pm ]
Post subject:  Re: Motion controller gloves - sample


I have never used a head Move with the PS Eyes set up since I started using iPi, I just use the integral head tracking only without many issues at all that can't easily be quickly manually fixed, but when Mike posted the spine deformation issue in the forum link above using PS Eyes and the head Move and head tracking unchecked the spine was affected badly, (issue shown in his posted image for a common crouch motion), and the only way to make it work right was to turn head tracking on while tracking and then apply the head Move data after, (shown in my posted images), although with PS Eyes, I didn't really need to apply the head Move data after, I just did for my own knowledge, but it didn't make any difference in that small sample anyway, more complex head motions it may have, but I can only state my experiences without ever using a head Move and PS Eyes.

A head Move almost has to be used with Kinects, (or a lot of manual positioning needs to be applied in post), and since I haven't tried it with them maybe Greenlaw can try and see if it does make any difference in the spine with/without head tracking checked while tracking and post his findings and maybe a screen shot or 2 of each way doing a forward bending motion, or even a simple crouch move as Mike did?

iPi has done a lot to make the tracking much better since Mike posted that issue, and no one else has reported such an issue that bad, so may be much better now with box unchecked and head Move used for either camera system, I will test it when I record again and see for myself with PS Eyes, or I still have 1-Kinect 360 I can test, but if anyone is having spine issues and using a head Move they can try turning head tracking on while tracking to see if it corrects it, but it may add some processing time to the overall tracking, especially with PS Eyes, that just depends on their individual machine, settings chosen and components as to what processing fps they will achieve.

I just tested the same dance with head tracking on while tracking and it off while tracking on a PS Eye recording, here are 2 screen shots of the difference on the exact same frame, using the exact same tracking options set, (except head tracking of course), although no head Move was used, the tracking will still be the same initially and the tracking was much more poor overall without head tracking ON.

Head Tracking Off is top image:

Head Tracking Unchecked.JPG
Head Tracking Unchecked.JPG [ 38.39 KiB | Viewed 7853 times ]

Head Tracking On.JPG
Head Tracking On.JPG [ 39.18 KiB | Viewed 7853 times ]

Granted after head Move data applied and refine process run, it may adjust this better, but I can't test that until I make another recording with a head Move later.

Author:  adamcloud [ Tue May 30, 2017 9:59 am ]
Post subject:  Re: Motion controller gloves - sample

In my experience using Move controllers a lot, one check box or not (since I don't remember which is best) the Moves are your head tracking. If you use them whether it's PsEye or Kinect, they're going to make a world of difference getting head tracking to be what it actually is far better than relying on camera depth or color to do the job for you. I've rarely had any problems with either camera set up if the controllers are calibrated and are pointed at any camera during recording for orienting to camera later during processing in Studio.

These screenshots are awesome! Love the little ring you made for the performance! Not sure I'm answering questions here but I currently own 13 (thirteen) Move controllers and I've tried just about everything I can imagine except successful multi-actor recording. Working on that.

Author:  Snapz [ Tue May 30, 2017 10:28 am ]
Post subject:  Re: Motion controller gloves - sample


I was basically referring to the spine deformations caused without using the head tracking and PS Eyes recording tracking, and my personal tests still showed the issue, so really that is all I can go on, if it doesn't cause bad deformations when tracking in the spine for anyone else, I don't know how, but if not that's great.

Correct, the head Moves controls only the head bone once the data is applied, I am just showing that the tracking with head tracking box unchecked when tracking PS Eyes footage, the tracking is awful, as it tries to keep the head and neck bone in a straight line with the body and that's not good for high motion tracking, so even if using a head Move, I suggest turning on the head tracking while processing, or at least try it and see the difference in the tracking quality.

It seems Kinects don't track as bad with head tracking off, as they don't track the head motion very well even with it on, but I still saw where in certain areas the hips and lower spine would lunge forward, or deform oddly in the tracking and to me that just isn't proper, or would need to be fixed later, because running the Refine tracking doesn't correct it once it has tracked that poorly.

Whatever works best for each individual user is what they will do, but for me and PS Eye recordings, I will stick to no head Move, use a dark head wrap if light haired performer, and use the built in head tracking, as you can see on any of my posted animations, the head is tracked very well with very little need to actually manually correct it, but there is always those certain spots that need it, which is easy to fix.

The circular carpet works well so the performer knows how far they can place their footing, but the arms, or any motions like kicks that are higher can go outside of that farther, that is just a foot/body contact to floor limit area, and also a bit softer on the feet and body than the stone and concrete under it.


Author:  Greenlaw [ Tue May 30, 2017 10:48 am ]
Post subject:  Re: Motion controller gloves - sample

Snapz wrote:
...I was basically referring to the spine deformations caused without using the head tracking and PS Eyes recording tracking, and my personal tests still showed the issue, so really that is all I can go on, if it doesn't cause bad deformations when tracking in the spine for anyone else, I don't know how, but if not that's great.

I haven't had problems as extreme as that lately but, yeah, I have seen odd spine issues before.

This first weird spine bending showed up for me a few years ago when they switched the spine system to use the flexible method. I immediately go weird bends when the actor was sitting or otherwise hunched or bent over. In my case, the spine bent to the side very unnaturally (I'm sure there's an old post in this forum that shows it.) Not long after reporting that, we got the option to use the original rigid or flexible spine method. I tend to stick with the rigid option.

I didn't realize using native head tracking also helped the spine issue. It really hasn't been a big issue for me but, then again, I don't typically record dancing like you do. I'll have to keep that tip in mind--I think I've had Head Tracking disabled for some time now. D'oh! :)

(I meant to record tests for this yesterday but my home studio network started misbehaving. Ready to try again tonight though.)

BTW, just wanted to confirm body height does have a lot to do with how well spine tracking works.

Before we got adjustable iPi skeletons iPi Mocap Studio, I sometimes had to scale the skeleton slightly up or down during tracking depending on what the actor was doing. In my case, my torso was slightly longer than the iPi skeleton's, so the spine could get compressed if I didn't scale the body taller than normal. Unfortunately, that could cause tracking issues for the arms and legs if I wasn't closely supervising the process because the elbow and knee joints would be further out of alignment.

Of course, nowadays we can adjust limb proportions of the skeleton to better fit different real-world body shapes, and it's been less of a problem for me.

Anyway, good thread...it's been very educational.

Page 2 of 2 All times are UTC - 8 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group