Hello,
I'm trying to process a photo dataset captured by a Wingtra One fixed wing drone with a MAP61 camera. It is a 61 MPx camera with a focal length of 17mm (93° field of view). The subject is a farm area surrounded by woods, but I'm not even trying the woods parts by now (actually I've tried but the results were so terrible that I´m concentrating in fixing that easier area first).
The first image I've attached shows the full area covered by the drone flights (that was generated using the same dataset in Wingtra Cloud. As you can see it worked quite well). The area I´m trying to process now is the center area, where you see "human activities" (buildings, crops, etc). It´s a 200 ha area.
More important info: the flight was done using terrain following (although Wingtra follows the terrain quite roughly), at 150 meteres high. The photos were post processed using PPK, so their coordinates are quite accurate. The superposition was 80% lateral and 85% frontal.
The problem lies at the very beginning: the photo alignment is not working as it should, generating a sparse cloud very deformed, with an "inverted saddle" shape. What bugs me mostly, is that Metashape is shifting the photos position quite a lot (and their angle) as you can see by the errors in the last screenshot. I´ve tried a lot of differente options in Align Photos dialog box: increase the Key Point Limit up to 150,000, the Tie point limit set to 0 (removing it´s limit), tried the Guided image matching with Key point per Mpx up to 7,000. Also, I´ve tried to clean the tie points and running the Optimize cameras several times without avail.
Also, I´ve tried to check (and uncheck) the camera angles checkbox to see if it changes something. It didn´t.
I know that a possible solution was to rerun the flight at higher altitude, but I´d like to know if there is a way to achieve better results with the present dataset.
Thanks in advance,
Bruno