Maybe this ist of interest for other users:
Thanks @bemis I could read this article:
http://onlinelibrary.wiley.com/doi/10.1002/esp.3609/fullAs I often have problems with the bowl-effect when creating DEMs, I tried it out.
First I planed and made an "normal" UAV-flight with 68 nadir fotos.
The result (DEM) was quite poor, with a big bowl-effect.
Then I took 10 more fotos and tilt the camera for 30°.
I alligned the nadir and oblique fotos.
The resulting DEM (78 fotos) did not show the bowl effect.
In the attachments you can see the flight plans, the dense clouds, the DEMs and profiles.
I did not use any GCPs.
It also works wit other tilt-angles of the camera.