Model Scaling Process with Vicon data

[Tord Natlandsmyr]

Hi, OpenSim community!

I am trying to use OpenSim 4.4 to calculate the joint center and angles from motion capture data, and have gotten unexpected results when scaling our base model using measurement based scaling. We are currently getting a weighted squared error of 14cm. I have previously used Vicon to calculate this, and I am now trying OpenSim, which seems to have a different scaling process. Vicon uses anthropomorphic measurements to scale a model to calculate the joint centers, but OpenSim seems to only be using the experimental marker data. Marker locations are correct and have been fixed where needed according to participant data. As most of the guides available online are outdated, tips would be appreciated.

Q1: Can one use anthropomorphic measurements in OpenSim to scale a model?
Q2: Does OpenSim only use the scale factors to scale a model?
Q3: Is there a good way to evaluate your scaling settings without using addbiomechanics.org?


Base Model: Rajagopal et.al. 2016 (https://nmbl.stanford.edu/wp-content/up ... al2016.pdf)
Marker set: 33-marker Vicon Plug-in Gait (error ~2-3cm) (excluding head and hand markers) (https://docs.vicon.com/download/attachm ... 000&api=v2)
Subject in questions: 74kg, 186cm
Scale Settings: See attachment

(addbiomechanics.org is sadly not an option as it is outside the EU)

[Thomas Uchida]

Q1: Can one use anthropomorphic measurements in OpenSim to scale a model?

You can provide "manual scales" to scale a body segment by specified amounts in each dimension. Please see the "Using Manual Scale Factors" section on the "How to Use the Scale Tool" page in the documentation: https://simtk-confluence.stanford.edu:8 ... Scale+Tool.

Q2: Does OpenSim only use the scale factors to scale a model?

Each body segment is scaled by a scale factor in each dimension. The scale factors are either computed based on distances between marker pairs (OpenSim calls these "measurements") or provided by the user ("manual scales").

Q3: Is there a good way to evaluate your scaling settings without using addbiomechanics.org?

Some rough guidelines can be found in the "Evaluating your Results" section on the "Getting Started with Scaling" page in the documentation: https://simtk-confluence.stanford.edu:8 ... th+Scaling.