Alignment of a mix of Images with RTK data and no GPS data at all

[4DS Niklas]

Hello everyone,

i have a giant set of 4000 images of a building of which i plan to generate a detailed pointcloud and a lot of orthomosaics of the buildings facades.
Because of vegetation the quality of the pictures location data varies a lot, with about 80% having precise coordinates and the rest varying between normal gps quality and no position at all.

In the past i had the problem in a similar case, where the program matched my images into two groups when i had source preselection activated.
I tried to get these to align together by manually creating markers, but it did not work at the time.
Any suggestions on how i would best proceed in this case? I also have around 80 points around the building measured with a total station, so i dont neccessarily need the images RTK data, but i assume using them would cut the time needed for alignment by a lot. Since its so many images, i wanted to try to get some input first, since its hard to expirement with this much data.
Thanks a lot in advance!

Niklas

[Alexey Pasumansky]

Hello Niklas,

Usually for the oblique shooting scenario (for example, when the drone is flying around the large building with the camera pointed to the facade) it is recommended to disable Reference preselection option and use only Generic preselection. You can also uncheck all the coordinates of the cameras in the Reference pane before starting the Align Photos operation.

But note, that the matching+alignment in such conditions could take a while for 4000 images, like a few hours on high-end machine.

Having marker input prior to the alignment wouldn't speed up the processing.