Scaling - Low Error but Wrong Posture

[Josée Mallah]

Hello,
I am using the publicly available human walking dataset of C. Fukuchi et al. with the gait2392 model.
I want to scale the model, so I am using the same scale setup file provided with the 2392 model, basically only changing the marker set (I added some markers and removed some to match the experimental marker set).
I tried to move my virtual markers to match the experimental ones/rerun the scaling tool using the optimised marker positions obtained in a previous run, etc. I ended up with a not-too-bad error, 1.5 cm squared error, 2.5 RMS and max error 6.6cm - however, the scaled model's torso is bending forward, which does not look right.
Any idea what I am doing wrong/how to fix this?
Thank you

[hans kainz]

Hi,
Do you have experimental markers on the torso of your participant?
If not, the torso will move in the same way as the pelvis segment, which might look strange and explain your observation.

[Josée Mallah]

Thank you Hans, yeah I don't have any upper-body markers, and it was almost my first attempt at scaling. Eventually I ended up removing the torso viewtopic.php?f=91&t=17849&sid=129db9d5 ... f7cbfcd01a, any further insights are welcome!