Need explanation,Error in pics and gcp depending of what i choose for optimizing

[Mr.Vain]

Hi forum!

I have a big problem, its because i dont really understand whats happpening så it would be great if someone can axplain it to me. We are using a matrice 300 with P1 camera and when computing projects we get different results. We always use gcp:s so we know that the data should be correct but we get wrong data.

First i align the photos and place the gcps:s. When i do "optimize" i do this with all pics unchecked and just gcp:s checked. This usually gives me an error of 0.005 or soemthing, really good. But if i look at the errors of the pics they could get errors up to 1m and when checking the cloud to the gcp.s we get big differences and really poor accuracy. What does really happen when i uncheck all pics and just have the gcp:s checked? Why do i get that  big errors on the pics?

If i align and then do the optimize with both pics and gcp:s checked i get more bad accuracy on the gcp:s but much better error on the pics. AND the cloud is getting much more correct. Is this the way to compute, with both gcp:s and pics checked when optimizing?

I must also say that i had experiences when using both pics and gcp:s checked and had really bad result becasuse of big error on pics and when unchecking them helped. Can someone please explain to me how this works and what workflow that should be used, i cant understand it.

I will attach 2 screenshots here also with optimized checked and not checked.

Thanks!

[Paulo]

Hello Mr.Vain,

definitely I would say there is an accuracy offset between your GCPs and camera positions. From the accuracy, it seems the camera positions were RTK fixed (cm level accuracy). Where they determined from same base as GCPs? And maybe you can can show the error tab in reference pane to see how errors are distributed...

[Mr.Vain]

Hello Mr.Vain,

definitely I would say there is an accuracy offset between your GCPs and camera positions. From the accuracy, it seems the camera positions were RTK fixed (cm level accuracy). Where they determined from same base as GCPs? And maybe you can can show the error tab in reference pane to see how errors are distributed...

Thanks for your answer. Here is a screendump with error showing and the project is optimized without using the pics.

[Paulo]

Sorry this is not the tab, the second tab from source tab gives errors for each camera, marker in East, North and Altitude....

[Mr.Vain]

Sorry this is not the tab, the second tab from source tab gives errors for each camera, marker in East, North and Altitude....

Aha, here is a new screenshot

[Paulo]

hello,

from what I can see, there is a systematic error of around 15 cm in North and 30 cm in Altitude on your camera positions relative to GCPs, while the east position respect the expected accuracy of RTK.

I would suggest to recheck youir ground control survey. Was it fixed? what base did you use? and base entered coordinates are they correct? Also check datum of both surveys...Camera RTK are WGS84, is ground survey in same datum?....

[olihar]

Have you projected the drone images into the correct GEOid?