In our nearest project, we took roughly 700 photos to generate a riverside area. Including: ~160 vertical photos and ~550 oblique photos. At the first time we started this project, there were 2 error occured in aerotriangulation submitting process :
- All of Vertical photos cannot be used for reconstruction. (image no.1)
- At 3D view : appeared another model of that area (same with normal ones but have different scale) and it located in the air .(image no.2; 3)
However, after that we created another project, used the same photos and it was ok.
Althougth this area is generated completely, but I still want to know the reason why those error occured (guarantee for next big areas)
Thank you.
did you solve the problem?
can you tell the right way?
Hi Felix,
We have tried to use Multi-pass in Estimation policies. However it was fail,maybe it required more Ram memory than the normal process, I will try it later !
Here is our latest 3D model of a new urban area and it has a small error : a coner of villa block has wrong elevation, lower than earth surface. We took large number of nadir photos (height: 60 meters), and fewer oblique photos (height: 100 meters), I wonder if the difference of camera altutide was the reason of error .
Do you have any idea ?
Thanks.
Hi Nguyen,
Its not clear why this happened. You may just have been unlucky (or hit a glitch) with the first try, and lucky with the second try (if you used the exact same version of ContextCapture on each attempt).
Some part of the algorithm uses random selection of keypoints during the aerotriangulation step. When the dataset is “easy”, it doesn’t affect the results, but when the viewing angles and resolution from the obliques and nadir are quite different, it can be effective.
We would recommend that you use the “mutli-pass” option for the “Optical properties estimation mode” in the AT settings.
This will help the software during the AT step doing a better match for difficult datasets.
Answer Verified By: Nguyen Vu