Change font size
It is currently Tue Dec 18, 2018 5:01 pm


Post a new topicPost a reply Page 1 of 1   [ 8 posts ]
Author Message
PostPosted: Tue Dec 04, 2018 11:44 am 

Joined: Tue Dec 04, 2018 11:33 am
Posts: 2
With your latest update you turned off the head/neck-rotation-distribution, when motion controller data is applied to the head.
Quote:
U N D E R T H E H O O D:
§ Based on users feedback, we turned off rotation distribution between head and neck joints when data from motion controller is applied to head. (This distribution was introduced in version 4.0.0.223.)


In my opinion the mocap-results without head/neck rotation distribution are looking far less realistic than before. Actually the difference seems to be so big to me, that I switched to the old version 4.0.3.226 again.
Is it possible to add a little option to reactive or turn on/off the head/neck-rotation, when adding motion controller data to the head? So people could decide, which distribution-method they would prefer.

Thank you for considering.


Top
 Profile  
 
PostPosted: Tue Dec 04, 2018 11:34 pm 
iPi Soft

Joined: Thu Apr 09, 2009 6:44 am
Posts: 170
Hello, thank you for your feedback.
A couple of questions to clarify.
Do you mean that animation looks more natural after motion transfer when motion data is applied to both head and neck?
Or do you think that iPi built-in model pose looks better aligned with video in this case?
Which model are you transferring motions to?

_________________
iPi Soft


Top
 Profile  
 
PostPosted: Wed Dec 05, 2018 11:13 pm 

Joined: Thu Sep 04, 2014 9:47 am
Posts: 879
Location: Florida USA
...

There wasn't much of a change to the head/neck after the head Move data is applied from the last version, can you post a snapshot of the issue you are referring to, to show the result change better?

What tracking settings are being used and which system is being used, Kinect, or PS Eyes?

...


Top
 Profile  
 
PostPosted: Fri Dec 07, 2018 10:05 am 

Joined: Thu Sep 04, 2014 9:47 am
Posts: 879
Location: Florida USA
...

We think we have deduced the issue and reported it to iPi and they are working on it to hopefully make it work more beneficial for all users, to be released in the next update.

...


Top
 Profile  
 
PostPosted: Sun Dec 09, 2018 10:32 am 

Joined: Tue Dec 04, 2018 11:33 am
Posts: 2
Hi, thanks for your replies.

I made a little comparison-video with the different versions:
https://youtu.be/wRFgY683KYM

I use two kinnect 2.0-Sensors (in opposite corners) and PS Move for the motion controls. The used versions and tracking-methods are mentioned in the video.

Sure, the difference is not too big between the new and old version, but in my opinion it is still really noticable (For example: 0:02, when the model moves his head to the bottom right). Esspecially when you try to make realistic looking animation imo it is a disadvantage, if the neck bone does not receive any tracking-data at all. Except if you use head-tracking, which never provides acceptable results for me, which is why I turn it off in most cases.

I hope this a help to you. Your software is really great besides this little change and I am a thankful customer for already two years. :-)


Top
 Profile  
 
PostPosted: Sun Dec 09, 2018 12:10 pm 

Joined: Thu Sep 04, 2014 9:47 am
Posts: 879
Location: Florida USA
...

Thanks for the video, we are using a 6-PS Eyes set up, and we have no way to compare the 2 tracking systems.

We can tell that there is a big difference between the Kinects and the PS Eyes tracking from that video and we can see the results after applying the head data is much different too.

We reverted back to 4.226 and ran tests with the head tracking off (shown in named image below), We had to apply the head data, or there would be no head turn at all.

Then ran tracking with the head on (shown in named image below), We didn't apply the head data from the Move (as named on the image).

Then we did apply the head Move data (named on image), and it can be seen that with head tracking on, which we always use by preference, the head Move data causes the neck to snap back up straight and destroying the neck rotations to follow the natural motion on the video.

So from this test we can see there is a difference in the performance after the head Move data is applied when video is tracked with head on and we were unaware of this difference and we believe iPi was also, and hope it can be remedied to work with head tracking on, or off, to not affect the neck in this manner, but the neck to perform with distribution like it does in 4.226.

You are correct that the distributed head and neck rotations look better, so the issue was coming from tracking with the head tracking ON, and then the head Move data applied resetting the neck to have no rotations at all afterwards.

We reverted back to several older versions and this effect was occurring even in those versions when tracking was done with the head on.

We hope the next update will work better for all users and with all tracking setting options chosen.

Limit is 3 files, so we tried to show the scenario with head off tracking result and with head on tracking result after the Move data was applied.

You can clearly see the differences to the video in each image captured on the same frame in each.

...


Attachments:
.226 after head data applied with head tracking on.JPG
.226 after head data applied with head tracking on.JPG [ 13.75 KiB | Viewed 73 times ]
.226 before head data applied with head tracking on.JPG
.226 before head data applied with head tracking on.JPG [ 13.61 KiB | Viewed 73 times ]
.226 after head data applied with head tracking off.JPG
.226 after head data applied with head tracking off.JPG [ 14.7 KiB | Viewed 73 times ]
Top
 Profile  
 
PostPosted: Mon Dec 10, 2018 3:45 am 
iPi Soft

Joined: Thu Apr 09, 2009 6:44 am
Posts: 170
Ok, thanks everybody for your feedback.
We'll add such option to the next version in a month or so.
Meanwhile the only workaround is to use the version that works best for you.

_________________
iPi Soft


Top
 Profile  
 
PostPosted: Mon Dec 10, 2018 10:55 am 

Joined: Thu Sep 04, 2014 9:47 am
Posts: 879
Location: Florida USA
...

Thanks Pavel, Now that we can see it was a head OFF vs. head ON tracking issue that was affecting the head Move data application, and not the head/neck rotation distribution, we can adjust accordingly until the next update is released.

Hopefully it can be updated to work the same when using head ON tracking, but if not, we can revert to using head OFF tracking, as long as we can also use shoulders by video and very flex spine options without issue while tracking, we get the best results for our purposes this way, and so far in the last version we tried it was working fine that way with head tracking OFF, though it would be nice to be able to use head ON tracking as an option without issue as well.

...


Top
 Profile  
 
Display posts from previous:  Sort by  
Post a new topicPost a reply Page 1 of 1   [ 8 posts ]


Who is online

Users browsing this forum: No registered users and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron


Powered by phpBB® Forum Software © phpBB Group
610nm Style by Daniel St. Jules of Gamexe.net