We're collecting photos with our DJI Mini 2 for the purpose of creating ortho-photos and/or 3D-models. We're flying at 100m and overlap typically 80/75% to get a final resolution of approximately 2-3cm/pixel. Multiple GCP;s are spread around the project area.
When importing photos in blocks to C-C, photos are shown as if they where taken horizontal, not vertically as they actually are. It's when looking at the project photo block in the 3D-view we realise all cameras are pointing forward and not downwards. Seems like there is something missing in the photo metadata but they are imported straight from the drone SD-card including position data from the drone GPS, flight control and camera.
Anyone experienced this and know whether it affects the final aerotriangulation or if there is a way to fix this?
I've noticed it's impossible to point out our GCP;s on the photos before running a first A-T without the GCP;S. Then, after that's been run, I may add the survey data and get a suggestion from C-C on what pictures to point the GCP;s out. There are between 2000 and 3000 photos in the project, so not possible to search manually for potential matches between GCP;s and photos showing them.
Sorry for my bad grammar and spelling!
/Anders Theodorsson, Ludvika Municipal GIS-data office
Does it have latest firmware?Try to import in some other software to compare.
I've now tried importing some 200-300 photos from my older DJI Mavic 2 Pro and a few from an DJI Mavic Mini 1. They all appear differently than the ones from our Mavic Mini 2. These pictures now show their correct pose and angles straight away without any adjustments.
So obviously there must be something wrong or missing in the metadata from the Mavic Mini 2. I noticed the third value in the row of figures describing the photo pose is null - zero - for all pictures coming from the Mini 2. While it has varying values on the photos from the other two drones.
Would there be a way to add that missing value before bringing the photos into C-C? It's not constant, so I guess that would be difficult.
Try to read metadata using Exiftool and compare with Mavic 2. There were issues with those DJI Air UAVs that their GPS coordinates were rounded to a 1m. Maybe something similar here. DJI is trying make consumer drones not suitable for pro use. Rotations are not so critical so you could delete rotations if they are wrong so it doesn't interfere with aerotriangulation.
Sorry for a late respons! I noticed that all cameras turned out correct after a first Aerotriangulation. So regardless of the reason for the in-correct pose at the first stage, it all came out well anyway. I wasn't able to check the metadata but in my later projects importing photos from the same drone, the poses were correct from start - strange! Must be some,"glitch" while reading all the photos (several thousands).
Let's consider the problem with bad photo poses being solved and thanks a lot for your support!
There was glitch from Mavic Air which flow over water in low altitude and it put 6000m as value in GPS height in photos because ground sensor got confused.