GCP, check points accuracy don't match the model.

[upland@mcsnet.ca]

I"ve  been having a few issues with my accuracy on check points not matching the ground elevations.  My check points once I have them all aligned to my GCP's they show an accuracy of 3cm or so.  However when I hover my mouse over the DTM under the check point, the difference in elevation between the model and my actual surveyed GCP is about 50cm.  I"m very confused as to why this difference.  What is the software referencing to when it returns an accuracy of 3cm?  Please advise.
I've attached some screen shots of Agisoft to show the issue.

[gto234]

The XY values on the second screenshot does not mach the XY of the GCP2.
So maybe you are looking in different spot.

[Alexey Pasumansky]

Hello upland@mcsnet.ca,

The 3D coordinates of the markers are estimated according to the their projections on the individual projections on the source photos. Whereas the altitude of the corresponding XY point on the reconstructed DEM may be different, due to, for example, the quality of the dense cloud reconstruction.

Which quality have you been using for the dense cloud generation and whether the DEM has been generated after the camera alignment optimization?

[upland@mcsnet.ca]

Hi Alexey,  thanks for the reply.  The dense point cloud was done at the medium level as any higher than that I find unnecessary for surface models. I did discover since my post something I’ve never had before with coordinate systems.  Recently I’ve been using the Nad 83 CSRS instead of just Nad 83 and I realized that it was adding a shift to my coordinates when generating the DEM without knowing and that is how my elevations got shifted.? Im not sure why this is happening but for now I will go back to using only the Nad 83.  I’m finding the CSRS is causing me some grief in some instances.  Im a bit confused how Agisoft handles these different coordinate systems as it seams to have alot of areas where you can change the coordinate system.  This concerns me as I think at times I’m shifting the data by accident.  My raw photo coordinates are lat and long and geoid HT2 height, then i convert them in Agisoft after importing them to UTM Nad 83 CSRS Z12.  My ground controls are UTM Nad 83 CSRS z12 as well with Geoid HT2.  I think I will go back to just Nad83.  What are your thoughts?