Scaling

[Pia Stefanek]

Hi,

I performed scaling with the MoBL-ARM model. I have the data of 8 persons from experiments.
I proceeded in the following way:
1. I set markers in the MoBL-ARM model as they were set during the experiments.
2. I performed scaling thereby using measurement based scaling. I reduced the maximal error and the RMS by changing weights on the markers and coordinates. I got results of in the following range: RMS = 0,013 - 0.038 and maximal error = 0,018 - 0,059. I saved the resulting models.
3. I opened the scaled models and set them in the scaling pose. I viewed the experimental marker data and adjusted the virtual markers with the experimental ones. I saved the model again.
4. Then I performed scaling again and got very low RMS and maximal error results.

Are these proceedings correct?

Thank you
Pia

[jimmy d]

2. I performed scaling thereby using measurement based scaling. I reduced the maximal error and the RMS by changing weights on the markers and coordinates. I got results of in the following range: RMS = 0,013 - 0.038 and maximal error = 0,018 - 0,059. I saved the resulting models.

The model is posed by performing a single frame IK solution. The reported errors are from IK and are derived from the markers you used to to pose the model. It is not obvious if the sum of RMS errors is useful or not because there are some markers that matter more than others (joint centers should have very little error while markers on soft tissue are expected to have plenty of error.

4. Then I performed scaling again and got very low RMS and maximal error results.

This is expected because the markers got moved in step 2 and mean anything. This only indicates that step 2 was performed-- not if the scaling or registration is any good.