By looking at the sample query dataset I was under the impression that by using Nadir photos I would be able to general and D.Point Cloud that shows the elevations of buildings / elements. But I was mistaken. All I am getting are roofs. Am I wrong is this? do I need to make adjustment to how high or low I am flying?
Too bad we can’t post images on these forums.
Hi there!, The question is, what do you want to obtain from your photos?.
Because the point of taking nadir photos is, from my point of view, to generate an orthomosaic, so the fact that you didn’t have the elevantions of the buildings or objets and you are getting the roofs instead, is actually something good i guess, as you will have a better result when processing the ortho.
If your intention is to have a D.Point Cloud/3dMesh where the elevations of the building are showed, i would suggest you to mix nadir photos, oblicual photos, and even terrestial photos.
Hope i helped you.
Thanks, yes that is figured needed to be done. I think I was a little confused based on the data set I was looking at (the quarry).
But you are right !! Already been down that road and the results are impressive.
I just noticed that the higher you fly in relation to buildings the mode of their front / sides you will capture… do you agree?
Hi all,
I agree with David, that nadir images are mainly taken in order to generate orthomosaics.
In general, your model will have altitudes of any point that is visible on the input images. If, for example, the wall of a building is visible on the input images (at least 2 images), then this point will be reconstructed in 3D (you will take X,Y,Z coordinates). If the wall of a building is not visible, then it is impossible to be reconstructed in 3D.
So, if your images have only the roof, you should expect the altitude for the roof. If you want the altitude of the walls, you should fly in oblique mode (tilt your camera).
Flying higher you will capture more the facades. You could do the same flying lower, but increasing the image rate.
I hope it helps,
Marc