we are using the xyz scaling option and CalculateSegmentDimensionsFromBVH.any to scale a model to our own set of collected Xsens data, but obtain unrealistically short trunk, so as shown below:
Looking at the file CalculateSegmentDimensionsFromBVH.any indicates that indeed the trunk height is determined from (approx.) L5 (=Xsens Chest2 node) to C1 (=Xsens Head node) so as calculated in the AnyBody Trunk ModelParameters.any.
There is also a factor (currently 1.08) that pre-multiplies the trunk height, which we could arbitrarily increase, but is there a rationale behind it? Why does segment scaling from Xsens data not work as expected in our case? Should we rather use additional points/trials from Xsens to properly scale the model?
When looking at the image i can see that the BVH stick figure and Skeleton fits well in hip and Sternum area. So as i see it the skeleton does its best to follow the stick-figure from xsens.
You can adjust the factor but it will not solve the problem as i see it, since the models already corresponds well, though not very realistic, but thus seems to be caused by the data.
Hi Søren,
Thanks for the reply. When double-checking the model in the Xsens software, we indeed noticed the same body dis-proportions, so we concluded that perhaps the input body dimensions were not measured accurately enough, causing a bad scaling. We are currently investigating if re-measuring the body dimensions fixes this problem.
Best regards,
Marco
