Change font size
It is currently Mon Oct 15, 2018 1:03 pm


Post a new topicPost a reply Page 1 of 1   [ 7 posts ]
Author Message
PostPosted: Mon May 08, 2017 7:39 pm 

Joined: Mon May 08, 2017 7:09 pm
Posts: 3
My setup is as follows:
2 Kinect v2
2 Windows 10 machines
up to 3 psmoves at a time (Only 2 in these demos)

I am having difficulty getting decent hand tracking with the playstation moves and the kinects. The moves are just being held in my hands. But for some reason the left hand seems to be having difficulty being tracked correctly and the right hand seems to be start bent backwards some times. I have 2 recordings each move being in either hand. I have tried multiple ways to load the animations.

I always start with tracking.
I have tried Attaching the controllers before this tracking or after it doesn't seem to make a difference and Don't expect it to. I then find a Frame where I have intentionally pointed (with both controllers) at the main camera. "Match with Camera"
I have ran apply motion data starting with either hand and I always run for both, beginning at the same frame that I ran the "Match with Camera".

So I did a simple test, I put what I have labeled as controller 2 in the right hand and controller 1 in the left hand and ran a simple slash animation for bother hands. I ran this animation twice in two different ipi recording sessions. After this has been ran, the left hand at the first frame of my t-pose is pointing directly to the left (as to continue the arms length) the right hand is pointing straight ahead (As we want the result to be). Now when I then attempt to motion capture the switched hand version of these controllers I get the exact same result. instead of a consistent result of the right hand controller being incorrect (to signify a ps move issue) and the left hand then being correct.

To simplify:
Controller #2 in right hand
Controller #1 in left hand
Right hand is correct
left hand is pointing with the length of the arm to be incorrect

Controller #1 in right hand
Controller #2 in left hand
Right hand is correct
left hand is pointing with the length of the arm to be incorrect

So I tried this with almost every possibility I could and oddly enough only the first set worked as expected this was #1 in right hand and #2 in left hand. but the other combinations failed.
#1 right #2 left success X2
#2 right #1 left fail X2
#3 right #1 left fail x2
#1 right #3 left fail X2

All of the failing ones start the T pose with the right controller facing forward and the left controller facing to the left.

Any ideas or help at all would be greatly appreciated.


Top
 Profile  
 
PostPosted: Wed May 10, 2017 5:03 am 

Joined: Thu Sep 04, 2014 9:47 am
Posts: 838
Location: Florida USA
...

When you add and set up the Moves are you making sure to calibrate each? You didn't state anything that you did this step... just taking it you did.
This shouldn't be needed all the time, not even needed on close and re-open of the recorder usually, but good to re-do it once in a while, but you can't do this after a recording, it has to be done first before recording anything.

You say #1 Right / #2 Left worked as expected, is this config working consistently? If so, what's the issue?

You should put a mark on each controller so it is easily known which is set as Right/Left and always set them that way in recorder set up and view port position.

Issues can also easily occur if the dongle(s) are plugged directly into the machine ports, a short 5m, or longer USB repeater cable with dongle plugged into it will help with electrostatic interference by getting it away from the machine more and possibly closer to the capture area if needed.

If you say you have a working set up with a specific config, then that should show the Moves are working correctly, so you can rule out a bad controller.

Move data isn't perfect, but should be pretty close enough to easily deal with in post cleaning of any minor errors, but if way off consistently, there is something more wrong.

Always run Move data apply as the last process, especially after making manual adjustments within iPi Studio, or running any jitter, or refine phases and definitely before final export.

...


Top
 Profile  
 
PostPosted: Wed May 10, 2017 5:12 am 
iPi Soft

Joined: Wed Jan 11, 2012 6:12 am
Posts: 2029
Location: Moscow, Russia
Hi

DeltaDaemon wrote:
I have tried Attaching the controllers before this tracking or after it doesn't seem to make a difference and Don't expect it to. I then find a Frame where I have intentionally pointed (with both controllers) at the main camera. "Match with Camera"
I have ran apply motion data starting with either hand and I always run for both, beginning at the same frame that I ran the "Match with Camera".

Do you also adjust the rotation of the hand bone manually to match the video before applying the data? "Match with Camera" step is needed to align a controller with the scene. Making correct hand rotation in current frame aligns the controller with the hand.

I am not sure that I've understood correctly what happens on your end in different cases.
Maybe you could share one of your projects (along with iPiVideo file)? Or at least make a screencast of your activity within Mocap Studio. Thus I would have better understanding and could give you more specific advice.


Top
 Profile  
 
PostPosted: Wed May 10, 2017 8:29 pm 

Joined: Mon May 08, 2017 7:09 pm
Posts: 3
So I was not always rotating the hand. So Rotating the hand seems to help the end result. And the entire process gives me results that look alright in MoCap Studio. But Exporting them as an FBX and running the animations in Unity just doesn't give me the same results the swords just seem to be off depending on which hand which controller is in.

Here are some files that I created. The one with the first controller in the right hand seems to come out the best altough not perfect and the hand is rotated to a lot lower then a usual hand in order to get the swords to align correct. The other recording with #1 in the left hand just has the swords crossing.

Here are some files. There is 4 files 2 motion data and 2 mocap data. the name of the files should help you attach the correct motion controller.

https://drive.google.com/file/d/0BxgaqnrVcc1RdHpRWGNhZGtfUTA/view


Top
 Profile  
 
PostPosted: Thu May 11, 2017 9:50 am 

Joined: Mon Aug 03, 2009 1:34 pm
Posts: 2268
Location: Los Angeles
The guys already mentioned the first two things that came to my mind: Calibration and Rotation Adjustment.

The one other thing I can think of is the Move controllers' distance from the Bluetooth device and possible interference in-between.

If your capture space is far from the computer/Bluetooth device where your computer is, you might get different or erratic results with the Move controllers at the capture location. This may happen because of a weaker BT signal or interference from nearby electronic appliances (i.e., large TV screens, computers, etc.,)

To avoid or at least minimize the issue, I like to run a USB cable to the capture space with the Bluetooth dongle attached there. Then I calibrate the Move controllers at the capture space. In my case, Move controller accuracy improved significantly after I started doing this.

_________________
Greenlaw
Artist/Partner - Little Green Dog | Demo Reel (2017) | Demo Reel (2015) | Demo Reel (2013)

Image
Watch a one minute excerpt on Vimeo now!


Top
 Profile  
 
PostPosted: Thu May 11, 2017 5:19 pm 

Joined: Thu Sep 04, 2014 9:47 am
Posts: 838
Location: Florida USA
...

I tested the first Motion file, it was a bit tricky, more than I am used to with PS Eyes anyway, but I was able to get it very close after a few attempts and using the sword as the controller to get the aim better when pointed at the camera 1.

I sent a video to your email with a character overlaid on the video to show you.

I also mentioned some other issues I saw with the iPi skeleton set up.

...


Top
 Profile  
 
PostPosted: Fri May 26, 2017 1:39 pm 

Joined: Mon May 08, 2017 7:09 pm
Posts: 3
Thanks guys for the help and advice. We were able to get some good animations from the advice given. Seems like the biggest thing was moving the bluetooth dongle closer to the capturing area. and then setting the character model in the studio to more reflect the actor. So we have some reasonable accurate animations now that are holding up a lot better in unity.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post a new topicPost a reply Page 1 of 1   [ 7 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