You might also have some more coverage of the railings if you tweak --pc-filter and --pc-sample, but this may also increase noise in the Point Cloud (as well as possibly massively increase processing time).
It’s my opinion that for the railings to be appropriately modelled you will need more points, so the averaged GSD must be set in te order of 0.2 - 0.3 cm.
You can also reduce the processing area to limit the resulting file sizes.
Probably is best to flight it as if you are making a 3d model for a building, so adding 2 or 3 orbits around the railings would make a big difference.
I hope this link could help
Hi Israelbar, yes, more points definitely needed, although orbits from relatively low heights AGL are not possible due to the large number of trees in this case. I did manually fly, to avoid trees, a couple of lower passes in the main area of interest with the camera at 70deg (In the link you gave, he seems to have the camera angles backwards, calling nadir images 0 deg, rather than 90 etc).
I’ll try to grab a few more images manually of the main eroded area when the weather clears, assuming they don’t start repairing it before then.
I’ve since done some 80 deg imaging with 90 forward/80 side overlap+ some 60 deg angled shots of our chook shed, which has worked fairly well, although I still need a few more images to cover the walls under the overhanging roof.
I dont know your camera specs but in order to keep GSD on lower values, you must manually fly at heights or distances not greater than 15 or 20 metres, and maybe closer. If not possible, maybe pre-processing and upscaling the images can help having more points on those structures.
Nice job so far.
Its a M2P, 20m AGL will give just under 5mm GSD.
Here are the camera positions I used, top level 50m AGL, then 45m, and manual smaller area mostly about 30m, and a few at 45m AGL.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.