It’s better not to read the coordinates of a point from Google Earth and use them as Ground Control Point (GCP) in a structure from motion processing.
Well, you can do it but then it is no longer a topographic survey.
GCPs are points that can be seen in photos taken during a photogrammetric acquisition and are used to orient, scale and georeference the 3D model created by the software.
Their position must be measured with precise topographical instrumentation.
Even admitting that you are sure that you found a clearly visible point in Google Earth that matches what’s in your photos, you have to know that the accuracy of its position (its coordinates) is about few meters.
A topographic instrument allows you to have precision of a few centimeters.
If you choose a Total Station it could even be better: few millimiters.
If you use 10 GCP and you have taken all their coordinates from Google Earth you have a positioning error, of a few meters, on each point.
That error is amplified for all the other points that are taken in account in the sfm processing.
It grows (it gets worse!).
It doesn’t stay the same.
It will end to a generally distributed precision even greater than 5 m.
Google Earth is a very good software but it is not a topographic tool.
Do not use it as if it were!