Hi. I ve been using reality capture for a brief time and i am exited with the programs capabilities. I use an RTK drone to create models and orthophotos. I am based in greece.
When i create a model, with the drones rtk measurements and then i check my orthos with my gcp’s measured with an rtk gps, but not used in the solution, i find significant differences.
when i use only the gcp’s coordinates to create the model all is ok. But u cannot always do that.
So i would like to have any suggestions in what am i doing wrong?
Also i have to mention that in order to have accuracy in the models coordinates, GPS data need to be transformed from wgs84 to greek grid (EGSA87) and then to Hellenic positioning system (Htrs07). coord sys. So we have 3 different datums with a main problem to solve, that there is a difference between orthometric heights and ellipsoid heights due to the use of different reference geoids. In order to solve this problem there is an unofficial tool the hepos transformation tool provided by the company that runs the greek cadaster.
let’s say that i find a solution for everything, i then need to change the mrk. file of the drones measurements, import it to the program and redo the process.
can i do that and how?
Hello Lampros,
how big differences are there? In which axis?
RealityCapture uses only ellipsoidal heights, so this could be the first part of your problem.
The GCPs are in Greek coordinate system?
Also, what are your alignment settings? Are you considering the RTK precision in the computation? And for such data it is good to know the camera’s calibration parameters. That could help to input them directly into computation.
It is not possible to import MRK file into RealityCapture directly, but you can use the part of it as a flight log. Just create a TXT file with the image names, coordinates and if you want, you can also use precisions (Change the coordinates of the photos - #11 by geoanture86).
First of all thank you a lot you for your answer. i will explore the solution you attached.
The differences vary between 0.4m and 1.5m at the examples i have worked with, and maybe they depend on how big the area is and the difference between nadir and zenith points of the area.
Yes the gcp s are in greek grid collected with a reliable gps.
From your answer i see the first mistake that i make, you see i had the impression that when i insert the photos and files the program automatically recognizes the parameters of the drones camera i use. Specifically i use a dji mavic 3 with rtk and i thought it was an automated procedure. ( I do not own a drone and a company lends me one when i need it, and did not get that info)
To do the first alignment i use the camera positions as captured by the drone with the rtk and i usually assume the are about 0.03 m accurate, witch probably is not true, but the default parameter is 0.05.
Then i insert the gcps, assign them, ignore the rtk positions of the photos and run an alignment based on the gcps only. And then everything falls to place. One time i even closed the rtk while collecting and only used the gcps and had good results.
If u have any more tips and procedure guidelines please feel free to add them.
Hi Lampros, the best option for you is using the GCPs, I suppose.