Problem with Alignment

[Felix Mueller]

Hi,

we are running a Matrice 210 RTK v2 (with base station). So the geotags with respect to the base station are very accurate.

During the field mission I had to reposition and reconnect the RTK base station. This has resulted in a data set consisting of two different abosult accuracies.
I have 200 pictures based on position A of the RTK base station and I have 200 pictures based on position B of the RTK base station.
So all pictures of data set A have a very good releative accuracy, so do the pictures of data set B in terms of relative accuracy (relative to the base station) - but not in terms of absolut accuracy.  Because of the fact that the base station itself is operating on a single receiver solution the absolute accuracy of the base station is within a normal gps accuracy (meters) resulting that data set A and B have different absolut accuracies.

This somehow yields to alignment problems. When I import the photos with geotags (lat, lon, h + orientation angles) and do the alignment I have nice results (<2cm) up to the moment where i changed the base stations position. From there on I have erros of about 1.5m in the camera locations. Thats all without any markes, just using the geotags of the pictures.

Off course I have markers distributed all over the mission area, but how can I combine the the alignment process with the markers? I want it to be adjusted all together. So when I check all the photos with theier geotags and I check all the markes and try to align I get a message that the vertical datum is out of range. This probably happens because the photo tags are in WGS84 and the marker coordinates are in ETRS89/UTM32 - GCG2016. But I have set the coordinates in the reference pane, I even have the geoidtiff. But it just dont work out.

What would be the procedure here?

Thank your for your help.
Best Regrards,
Felix

[Felix Mueller]

I think I kinda found the solution, not 100% sure.

The problem ist that the imported photos have accuracies entries, which are in the range of <2cm. This is because the camaera postions are quiete accurate with respect to the mobile base station. So if you fly a mission with a fixed position of the mobile base station you are perfectly fine. But as you change the postion of that mobile base station, the second data set will get an offset to the first data set. However all the data sets based on base station postion 1 AND base station position 2 will get the same accuracy, which is not true - there are not equal relativ to each other, only consistent within each other.
So I deleted the accuracies entries and let agisoft calculate the alignment with a realistic a priori accuracies (meters). I also found a solution to thaht vertical datum issue. My GCPs had the format 32.400.800,123, so I deleted the 32 to get EPSG 25832 cosistent coordinates - et voila mesage disappeared and my gcp have an average error of 8mm.

Best regards,
Felix

[dsmapping]

How did you delete the accuracy entries?