I've searched the forum for any recommendation regarding best camera angle when collecting photos for both orthophotos and 3D-models with a drone. In other forums I may find recommendations spanning from minus 30 to minus 80 degrees from horizon but they all suggest multiple flights. I am trying to limit the flight to one cross pattern (grid) with one camera angle at one height. Then combining that with a set of nadir photos from same height with great overlap.
In Bentleys drone capturing guide it is said a recommended overlap between same direction flights is 70% and to acchive that flying back and forth I need to set the overlap between single flight lines to 85% creating a big number of lines and photos. So adding extra flights with different camera angle is not really an option when flying large areas.
Has anyone tried different angles iflying over a small urban area consisting of 3 to 10 storey buildings mixed with parks and open areas, a typicall small town area? I would be happy get any guidance but I understand it might not be possible to acchive top results in the 3D-model using only one set of oblique photos.
/Anders Theodorsson - Ludvika Municipal GIS-office
Double grid with camera angle between 60-70 gives the best result and no need for Nadir at all. You can generate orthophoto from 3D process then but orthophotos from 2.5D process from Nadir only could look slightly better(trees, building corners) also 2.5D is easier to retouch. Unfortunately Contextcapture doesn't produce best looking orthophotos even in 2.5D at least for me.
Also Bentley now recommends overlapping circles missions(some planning apps has such) as it gives better calibration but I would recommend stick with good pre-calibration anyway and overlapping circles may not be easiest to get consistent coverage.
Thank you Oto, quick response - as always! I'll stick with the grid pattern, it will be too time consuming running circles over the whole town. Maybe try that on some important objects where higher resolution might be wanted.
Also I'll try creating the orthophoto from the 3D-model as suggested and compare with one made from nadir photos or a mix of oblique and nadir. I might even run a set of photos in Agisoft to compare the orthophotos.
Do you - or anyone else have a recommendation regarding overlap between photos in the grid? I think a final overlap of 70% in one direction is quite much. It reults in the need of 85% overlap when running back and forth! A 15 minutes mission with our drone suddenly rise to 55-60 minutes going from only nadir in lawn mower pattern to oblique in grid pattern.
Anders Theodorsson said:quick response - as always!
Better some answer as no answer as often questions are left without any answer.
Anders Theodorsson said:A 15 minutes mission with our drone suddenly rise to 55-60 minutes going from only nadir in lawn mower pattern to oblique in grid pattern.
That unfortunately is expected. You could lower overlap but risk that some photos may be lost but in your case it may not be problem if possible to fly again as normally it is not easy possible to go back to the site. Your drone is not well suited for covering larger areas you would need to use fixed wing drone or with high resolution camera so can fly higher and cover larger area. When flying at 100m may work with less overlap as trees doesn't mess so much with aeortriangulation at least if not windy. When flying lower like 50m and missions take too long time the moving shadows also start to cause issue with AT.