This is similar to the optional 'shift' of coordinates in Photoscan. From what I could find, this is done to increase the numerical stability of the computations.
Thinking in a proper description of methods (for a near-future scientific paper), I wonder if someone could provide a simple explanation about the 'numerical stability', or point me to a reference.
Checking the log for a dataset processed with ODM using GCPs, I noticed this (only 4 first lines of GCPS shown):
This is similar to the optional 'shift' of coordinates in Photoscan. From what I could find, this is done to increase the numerical stability of the computations.
Thinking in a proper description of methods (for a near-future scientific paper), I wonder if someone could provide a simple explanation about the 'numerical stability', or point me to a reference.
thanks