Page 1 of 1

Motive 1.5 versus 1.0.2 RigidBody settings

Posted: Wed Sep 04, 2013 10:34 am
by martinv
Hi,

In Motive 1.0.2, in the Rigid Body Settings, it seems there is much more settings than in Motive 1.5

For instance, in Motive 1.0.2, in the Advanced tab here are the settings
Solver Settings => Enabled
Solver Settings =>Dynamic Override
Solver Settings =>Static override
Solver Settings =>Force exhaustive
Solver Settings =>Acquisition delay

I do not find those settings in Motive 1.5.

What is the reason to no longer have those settings ?
(Or maybe they are located elsewhere ?)

Also in Motive 1.0.2, there was a Smoothing Settings for Translation and Orientation.

In Motive 1.5, I only see a setting called Smoothing in the RigidBody Settings->General Properties

Is it the same functionality ?



regards

Martin

Re: Motive 1.5 versus 1.0.2 RigidBody settings

Posted: Thu Sep 05, 2013 4:03 pm
by beckdo
Hey Martin,

We've spent considerable resources on improving the Rigid Body solver since Motive v1.0.2. As a result there have been numerous changes under the hood that have affected what is visible in the UI.

In regards to your questions about settings removed:

Solver Setting==>Enabled: This setting still exists. It's now controlled by the checkbox next to the rigid body name in the assets panel.

Solver Settings==Dynamic & Static Override: All UI settings for dynamic & static constraint are now removed and handled differently by the solver. Rather than try to define a dynamic setting that says "move no more than ___ distance in ___ time", the solver takes an improved approach. It will observe potential rigid body solutions and ultimately choose the correct solution based on a statistical confidence that it measures internally. As a result, UI settings are no longer needed--it's all automatic.

Solver Settings==>Force Exhaustive & Acquisition Delay: Due to performance improvements in the solving algorithm, this has been improved. The solver always solves exhaustively and there is always no acquisition delay.

Smoothing has been revamped and simplified. Orientation and positional smoothing have been combined into a single slider. The smoothing algorithm used in Motive v1.5 is Double Exponential Smoothing.

Let us know your experiences with rigid body tracking in Motive v1.5. We're interested in hearing your feedback.

Thanks,
Doug

Re: Motive 1.5 versus 1.0.2 RigidBody settings

Posted: Thu Sep 26, 2013 9:30 am
by doktorvr
Orientation and positional smoothing have been combined into a single slider.
This is a strange decision. Positional and rotational data are subject to different degrees of perceptible jitter and thus require different degrees of smoothing. In my experience I need to smooth rotation more than position.

Please provide two individual sliders in the final release.

This decision, plus your decision to "deprecate" the choice of right/left hand coordinate system only serve to restrict Motive in unecessary ways that your customers have enjoyed in previous versions. I'm quite disappointed in this direction and hope that these decisions change with the final release of 1.5.

Re: Motive 1.5 versus 1.0.2 RigidBody settings

Posted: Fri Sep 27, 2013 10:28 am
by NaturalPoint-Dustin
Hello Doktorvr,

Thank you for your feedback and concern. I invite you to to test the rigid body smoothing in the betas. If you find that you do indeed need separate rotation and translation smoothin, we will consider this. The setting have not just collapsed together. We have also implemented Double Exponential Smoothing that should give better smoothing results.

We had different handedness in Motive depending on how you got your data out, what type of data it was, where you were in the app, etc. This discontinuity created confusion at it was encountered by customers. So to make Motive more consistent and to make integration more straightforward and reliable, we standardized the handedness throughout the app. We also added conversion classes and code examples for those using the streams and exported files to convert to left-handed coordinates if they need them. We welcome any feedback on how we can generally improve workflows and make Motive easier to work with.


Best Regards,

Re: Motive 1.5 versus 1.0.2 RigidBody settings

Posted: Sat Sep 28, 2013 1:37 am
by doktorvr
Thanks Dustin.

We can live with not being able to set the handedness of the coordinate system.

We would however really appreciate splitting the smoothing into two separate controls for position and rotation. We use Motive for live head tracking data for a head mounted display. When wearing the hmd, jitter in the rotation data is far more perceptible than that in position data. Also as smoothing increases latency we need to find "sweet spots" and these differ in position and rotation.

Re: Motive 1.5 versus 1.0.2 RigidBody settings

Posted: Sat Sep 28, 2013 1:37 am
by doktorvr
Thanks Dustin.

We can live with not being able to set the handedness of the coordinate system.

We would however really appreciate splitting the smoothing into two separate controls for position and rotation. We use Motive for live head tracking data for a head mounted display. When wearing the hmd, jitter in the rotation data is far more perceptible than that in position data. Also as smoothing increases latency we need to find "sweet spots" and these differ in position and rotation.