Hello again,
some idea on this?
I looked again on my data and I get the feeling, that the height field projection takes place along the average plane of the photo distribution, which is not obligatory perpendicular to the z-axis (and will never be strictly perpendicular to the z-axis), specially when one is mapping landscapes with a lot of relief and the photo height is adapted to the topography. I do not understand why it should be like this, since the height field mesh generation is mainly used for mapping applications, where the idea is to project the final 2D map exactly along the z-axis.
The reason why I did not always see this is due to the fact of having mapped mainly flat landscapes.
But I hope I'm wrong on this one and someone can explain me what I'm doing wrong.
Best regards
Patribus
PS: A solution would be to use arbitrary mesh generation, but using (ultra) high dense point clouds this is almost impossible with my computational power for larger projects.