Recently, I have been processing few of the datasets with GCP. Sometimes, I get high GCP RMSE value and sometimes it is very low, as good as, 2-8 cm in x,y and z.
One thing, I have observed is that when processing larger datasets, say like 150-300 Hectares, the RMSE value is little high. What could be the reason?
I usually mark around 4-5 photos with respect to each GCP. I hope that is sufficient enough to calibrate.
What are the factors that affect GCP RMSE values? How can I improve them?
What version(s) are you observing this with? We pushed a tweak that might greatly reduce this error for Build 63 and above.
In more general terms, 5 images should be totally sufficient. If you aren’t being super-precise when tagging the pixel locations on each image, that will create and compound error. Using natural features or other objects where determining an exact position to tag is difficult, you’ll likely have more error. Low-quality GPS position data when recording the location of a GCP location is also a potential source of error.