I am using DJI Phantom 4 RTK drone and WebOBM to capturing topography around bridge structures. While the flight flown with mission planning typically come out well, typically, the area under the bridges is not caputrued due to the high-altitude flights. I have tried unsuccessfully to take low-altitude flights with manually taken photos of regions under the bridge from each side. Whether I process the manual photos independently or in combination with the high-altitude photos, the manual photos are not used by WebODM or end up with bizarre results. Can anyone tell me how to best combine this data what settings would work best? I have tried many combinations including the Default, Building, 3D Model as well as forcing the use of the camera gps coordinates without success.
For starters, it would be helpful to just get the manual photos to process properly independently from the flight plan photos.
A link to a sample set of the manual photos is here: Dropbox - Manual Photos Sheldon Cnyn
The most important thing is to make sure you have plenty of common features in overlap areas for the flight plan and manual images. Also aim not to have too great a variation in effective GSD. IE very close up images from under the bridge might not work very well with drone images from 60m AGL, even with plenty of overlap.
I’m not sure there is enough overlap and images from different angles to get a decent model- all the images appear to be taken along a single plane, plus there’s a lot of sky.
Processing now, but this isn’t ideal:
2023-03-11 16:02:46,662 INFO: Reconstruction 0: 27 images, 38018 points
2023-03-11 16:02:46,662 INFO: Reconstruction 1: 15 images, 21810 points
The result was a disjointed mess, nothing much recognisable.
min-num-features: 12000, optimize-disk-space: true, pc-quality: ultra, skip-orthophoto: true, sky-removal: true, use-3dmesh: true
|Average GSD:|0.05 cm|
You really need images from varying heights and different angles to get a decent reconstruction of complex shapes like the bridge and surroundings.