ipisoft.com
http://forum.ipisoft.com/

Things I think could be optimized in V3
http://forum.ipisoft.com/viewtopic.php?f=4&t=9245
Page 2 of 2

Author:  vmaslov [ Tue Jan 06, 2015 3:34 pm ]
Post subject:  Re: Things I think could be optimized in V3

Mocap Studio does not specifically track eyes for head rotations. So may be you simply set more appropriate height for the model which improves tracking overall, not only for the head. The fact that model's height you set does not match the actual height of an actor may be caused by the improper scale of the scene. Have you set the scale after calibration so that camera heights are close to their actual heights?

And you SHOULD refit the pose any time you do some adjustments to the model. Because a pose determined with the wrong actor's model is not optimal. Also, you SHOULD refit the pose after manually editing the pose when doing cleanup. Basically, you SHOULD refit the pose any time you are going to start the tracking.
(Maybe even do this automatically when the tracking is started - a nice idea to think about)

We had not either implemented any specific behavior for the camera attached to a particular USB port. I can't tell whether the behavior you see is a sort of coincidence or some occasional "feature".

Anyway, you may share your project (along with the video) to better illustrate your problems.

Author:  Snapz [ Tue Jan 06, 2015 7:37 pm ]
Post subject:  Re: Things I think could be optimized in V3

Thx for the reply...

Yes, I have tracked with all different configurations that could be thought of and results are not as should be, in my opinion.

Cam scaling is set as close as what I am given, which isn't correct, but to me .3m off of actual cam 1 height as in V3 is alot in my opinion and its throwing tracking off, even though I can refit actor to video fine, the dancers I use are women and shorter in height than me so on refit the scaling is a bit off, so I adjusted it manually and didnt refit, with better tracking.As soon as you hit track forward the actor rescales itself a bit more anyway, so I can never get true accurate fit of these dancers.

I get better refit at my height of 5' 11" to the actual video, but not with their heights. Maybe later, if individual bone scaling is implemented, this may be better.

The fine tune floor updates helps, but there is still something happening.

I have a message into support tech to get files transfered to him for review.

I recorded and tracked a video in V2 and exported it without these issues, so something is going on with V3.

All the calibrations and cam heights were set after calibration yes, but V3 calibration gives improper scaling to actual cam heights, where V2 calibration does not do this, V2 is almost dead on to actual cam heights, V3 calibrates wrong (i.e. cam 1 is set at 1.52m, V3 calibrates it at 1.82m) V2 doesn't do this, its is very close to correct everytime.

I also opened same recording in V3 and tracked and exported it and it has very poor results.

Get with the tech to discuss what I send to him.

I will post if any insight to these issues can be determined.

Thx

Author:  vmaslov [ Wed Jan 07, 2015 1:57 pm ]
Post subject:  Re: Things I think could be optimized in V3

Snapz wrote:
First thing, is to remove the annoying working wheel animation from middle of the screen, who thought that was optimal placement?

Done in v3.0.3.172.

Author:  Snapz [ Wed Jan 07, 2015 7:39 pm ]
Post subject:  Re: Things I think could be optimized in V3

Thx, great news! and also good fix on the actors scaling now to match floor height, tested and working much more within reason now... Good Job!

Author:  Snapz [ Fri Jan 09, 2015 7:26 pm ]
Post subject:  Re: Things I think could be optimized in V3

Update is working great, tracking accuracy is the best its been for me!

I don't know if this update was for resolving my issues, but it seems to have addressed them.

The actor scaling from floor grid was a needed improvement and seems it has helped in the ankle positioning to floor and heel penetration issue.

I discussed the camera calibration with someone and he explained the new calibration algorythm a bit, being determined differently from V2, so what I was doing for scaling

was correct, it was working good, I just thought the initial scale was supposed to be closer to actual camera heights.

After scaling cam heights, I was always well within reason at .02 - .04cm per cam.

Thanks for either promptly addressing the reported issue above, or if it was already in progess from the team noticing it, either way, Thanks for implementing it.

Page 2 of 2 All times are UTC - 8 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group
http://www.phpbb.com/