simtk.org

Hello,

I have a full body model with a limited muscle set around the elbow and forearm. I have been trying to port the muscle set and geometry from the MoBL_ARMS model onto the model to include more shoulder muscles, but have had some problems.

My original model was just a ball and socket shoulder joint, so I have added the clavicle and scapula bodies, along with the necessary joints, constraints, and the muscles. The problem is that the "torso_offsets" used do not work as expected in my model. Using the offsets from the MoBL_ARMS model, all of the bodies and muscles are offset from a different position. I also noticed that the "torso_offset" used to establish the shoulder joint in my original model is (0.003155,0.3715,0.17) vs. the sternoclavicular joint offset in MoBL_ARMS which is (0.006325,0.00693,0.025465). The Y-value is hugely different which I think is part of the issue.



Opening the two models in OpenSim, my model appears in the visualizer with the pelvis halfway through the ground plane, where as the MoBL_ARMS model appears mostly blow it, with the suprasternal notch just appearing above the ground plane (see attached)

My thoughts:

• The MoBL_ARMS model references the torso as the ground plane, where as in my model the pelvis is referenced as the ground plane. Could that be throwing off the offsets?

• Looking at the geometry files, I found that MoBL_ARMS uses the "throax.vtp" file, where as my model uses the "hat_ribs.vtp" file. When switching the MoBL_ARMS model to the hat_ribs.vtp file a similar thing happens where the offsets are no longer correct. Maybe the different toros geometries require different offsets?

When I change from the thorax.vtp to the hat_ribs.vtp the same offset issues arise (see attached).

Any help is greatly appreciated, I think I just need to figure out what the "torso_offset" needs to be so I can apply it to the joints and muscles. I attached a couple extra screenshots for reference.

Thanks,

Joe Marsh

Hi Joseph,

I'm not sure where the MoBL_ARMS model stands in terms of migration to version 4.0. I would start there using their provided geometry/mesh files, get that behaving correctly in 4.0 before adding your own changes/modifications. Generally the mesh files are used only for visualization so you may see a set of bones misplaced if you use meshes from one model in another model (since origin/transform could be baked into the meshes) but muscles are unaffected. If you're using custom meshes, make sure they are in a Geometry folder under your model or are placed on the Geometry search path in the application preference.

Hope this helps,
-Ayman

Hi Ayman,

Thanks for the quick reply!

I wasn't using any custom meshes other than the ones that came with the file.

I had ported opened the MoBL_ARMS model in 4.0 with no problem, I couldn't pick up on any differences from v3.3. Looking through the geometry folder for 4.0 there is no thorax.vtp file there by default, it seems to have been replaced by the hat_ribs.vtp file. I added the thorax.vtp file into the Geometry folder and it works as I described above, with the different origin/ offset.

I suspect that the origin/ transform between the two meshes is the issue - do you know a way to find that difference between the two meshes?

Thanks,
Joe Marsh

Hi Joe,

What I mean by custom meshes is meshes that are not included in the official OpenSim distribution, so you're actually using custom meshes if the model comes with its own.

Indeed OpenSim 4.0 geometry does not include thorax.vtp, neither did version 3.3. If you got the file somewhere else it may have made some assumptions built into it as to how it is attached to the underlying body that do not hold for MoBL_ARMS. You can fix that by changing the transform of the mesh in the user interface (this maybe easier done in version 3.3 user interface).

Hope this helps,
-Ayman

Hi Ayman,

Thanks for the reply - I think you got me pointed in the right direction.

I couldn't find the actual origins in OpenSim 3.3, but I did find an opensource software (called ParaView) that allowed my to find the origin differences between the two thorax mesh models.

From there I just applied those translations to the muscles and joints and it all worked out!

Thanks again,

Joe