Clarification on Gimbal/Camera/Drone yaw pitch and roll - what does Agisoft want

[TimCull]

When importing EXIF information (or reading it from the image metadata) from UAV images - what does Agisoft want with regards to pitch, yaw and roll? Is it after the Y/P/R of the gimbal, camera or drone?

I would've expected it to be the camera (which should be equal to the gimbal I'd expect) but given that most EXIF datasets contain several styles - can't be sure which one Agisoft is after...

[SAV]

Hi TimCull,

Commonly reference frames used by photogrammetry software packages (such as PhotoScan) and drones differ from each other. Moreover, when gimbals are used it can get quite messy/complicated to calculate the correct Y/P/R values (because gimbal Y/P/R and UAV Y/P/R both need to be considered when computing camera Y/P/R).

Here is the good news. There is (generally) no need to import Y/P/R values into PhotoScan (has been discussed in this forum a few times before). Camera orientation (Y/P/R) will be estimated by Photoscan based on key points/tie points in the imagery. Importing location information (LAT/LONG/ALT) from image EXIF is all you need to do.

Regards,
SAV

[GPC]

If I have high-accuracy positioning on the camera locations (via on-board PPK GNSS) is it ok to do a low-quality alignment? Is it ok to assume the XYZ positions of the images are very accurate so the rotations do not require high accuracy alignment?

[SAV]

Hi GPC,

If you have highly accurate camera locations (= 'air control points') then you probably get more accurate results compared to using ground control points, as shown in a publication by James et al 2017.

Survey precision under direct georeferencing could be 2–3 times better than from GCP-control.

Regarding the low-quality alignment. This setting will change at what resolution the aerial images are used (HIGH = 100% = all pixels) to detect and match features (key points/tie points). Overall, at higher quality settings results in more key points that are detected and a more robust tie point cloud that is used to align each camera. Note that the number of key points also depends on the resolution of your images and on the scene itself (e.g., lack of texture in images = less key points).

Regards,
SAV

[TimCull]

Hi SAV,

Thanks for clearing that up! So I guess it Agisoft doesn't use the Y/P/R in the reference pre-selection for alignment either?

Thanks,

Tim

[n@sk]

Hi GPC,

If you have highly accurate camera locations (= 'air control points') then you probably get more accurate results compared to using ground control points, as shown in a publication by James et al 2017.

Survey precision under direct georeferencing could be 2–3 times better than from GCP-control.

Regards,
SAV

This is a misinterpretation of that statement. Precision is not accuracy.
Chances are that you cannot get "highly accurate camera locations" and the higher you fly the less accurate the ground coordinates will be.

regards