Scaled Model Orientation

[Jesal Parekh]

Hello,

I used Open Sim Gait2392_Simbody model to scale to data captured using MAC system. The model markers and segments scaled as desired but the orientation was skewed. The scaled model was displaced by about 90 degrees counter-clockwise to the original model. Any suggestions on how this can be corrected? I looked through all the scaling and set up files but failed to locate any orientation/coordinate axis code to change.

Thanks!

Jesal Parekh

[Daniel Leib]

Hi Jesal,

How is your lab coordinate system set up? I have the same issue when using a different set of software; it doesn't affect the data outcomes unless the actual x,y,z coordinate order is important, as the rest of the kinematic and kinetic data is rotated to the same positions. If it's possible, try recollecting your data so that it matches the pose of the original model, or maybe calibrating your space with a rotated reference frame.

The x,y,z orientation of the trc file containing your "marker data for static pose" is what will cause the scaled model to be rotated, as it will fit the model to the pose in the data set.

If it's important to your purpose that you have the scaled model in the same orientation as the original model, you can uncheck "adjust model markers" in the scaling too. This will scale the geometry without moving the model to a new location/orientation.

If I'm not mistaken, the orientation of segments is built in to the .vtp files, so you'd have to change a global coordinate scheme within OpenSim, not in the .osim file. Not sure there's a way to do that offhand through the GUI. You could also go in and rebuild each vtp file in your desired orientation if you really need a permanent change as well.

Have a good one,

Dan

[Ajay Seth]

Hi Jesal:

OpenSim assumes X to be fore(+ve)-aft, Y to vertical(+ve up) and Z to be medio-lateral. This assumption is necessary because of how ground reactions forces are applied to the model and some other issues.

The easiest solution is NOT to recollect your data but to simply rotate your data (markers and force-plate) to match the OpenSim convention.

Cheers

[J.J. Mordang]

Hi,


I've a problem with my model orientation as well. I made a model based on the One-Leg model and made a marker set for it. When I change coordinates of the segments and the markers will move properly with it. However when I want to scale it with measurement data that's already been converted to the axis of OpenSim (validated with the preview motion data option). The result is very strange. The segments of the model are flipped 90 degrees (the coordinates of the joints are changed), with the virtual markers in the normal state upon it. And the experimental markers are not visualised. When I change the coordinates of the scaled model back to the initial position. The virtual markers are all a bit scrambled and the experimental markers can still not be seen.

I hope it is a little bit clear what I mean.
I tried several things but it still doesn't work. Can anyone help me with this?


Thanks in advance,

Jan-Jurre

[Ayman Habib]

Jan Jurre,

Changing coordinates moves the model markers. The trc file used in IK (and to adjust-model-markers while scaling) has the global positions of the experimental markers. Can you load this trc file using the previewer and over lay the motion with the model that has markers? I suspect that's where there's a disconnect.

-Ayman

[J.J. Mordang]

Yes that's where it was, thanks for you help!

[J.J. Mordang]

Hi,


It's going all wrong again. The experimental markers and virtual markers now overlap, sort of (al little difference because the foot is at an angle and not pointed along the x-axis). But the virtual marker located at the foot float around the shin after scaling. And again the experimental markers are nowhere to be seen.

Can someone help me with this problem?

Kind regards,

Jan-Jurre

[Ayman Habib]

Hi Jan-Jurre,

First I'd do scaling without marker placement (uncheck the box for adjust model markers). Now redo the exercise of previewing the trc file, and overlaying it on the scaled model. That should tell you how much markers need to be moved by the (adjust model markers).

Hope this helps,
-Ayman

[J.J. Mordang]

Hi Ayman,


When I overlay the model markers with the experimental markers, it looks alright. The markers don't overlap quite right because of the position of the foot during the experiment itself, the foot is pointed outwards in contrast to the model foot which is pointed along the X-axis. I noticed that the pelvis of the model was fixed to the ground. Maybe this is a problem? How can a joint be made between the pelvis and the ground that allows translation along the 3 axis?

Thank you very much for your help,

Jan-Jurre

[J.J. Mordang]

Hi,


A joint was made between the ground and the pelvis. This solved the problem for a little bit. Still the virtual markers float al bit above the feet and are not located at their original position on the feet itself. Also during inverse kinematics the patella comes of the knee which looks a bit strange.
Does someone know what I'm doing wrong?


Kind regards.

Jan-Jurre