I'm having a difficult time understanding why CC is rejecting so many of my photos. When I run AT 140 out 438 photos can not be used for reconstruction.
When I view the photo information on CC there are blocks of photos that seem to have no attached positional data.
I used Intel Mission Control to georeference all the photos before importing to CC. Mission Control has an option for exporting the data to a .csv file. When I view the .csv file I see data for ALL the photos. Why is CC rejecting so many photos? This problem is happening with several jobs.
Thanks!
Mark H
Your photos doesn't have pose metadata so when you import positions from CSV files it only updates photo pose but it doesn't update metadata(EXIF) so it is normal that after aerotriangulation poses are adjusted and pose is lost if photo is rejected. Maybe it was possible to import pose as metadata but don't have CC available now to check.
Rejection could be related to:
From these screenshots not possible to get idea what kind of mission it was and what is the overlap between pictures.
Oto,
Thank you for your reply. Let me give you some more info. The area in the project is heavily wooded. Leaves are off the trees so we hoped to see the ground well enough to understand what is going on. I flew the Intel Falcon 8 with a 35mm Sony camera. Camera angle was set at 30 degrees. Photo overlap was 80% front and 70% on sides. Flight elevation was 250ft. Ground control was set in 5 places on the perimeter of the project area and tied out with survey grade gps equipment.
Initially when I exported the flight data and photos from Intel Mission Control I thought all metadata was included with the photos. The screen shots i posted were taken before I discovered that Mission Control is not exporting metadata completely. I got the EXIF info updated and started a new project in CC. Ultimately I got the same result. Out of 243 pictures aerotriangulation (before picking GCPs) rejects 150 to 170 photos.
Attached is a screenshot from Intel Mission Control showing the flight path. Note that there are 2 flights shown. When I realized that my coverage of the north end of the project was week, I flew a second flight in that area.
We are beginning to think that we are asking CC to do the impossible. Are we? We are still learning about this software and drone surveying in general. I realize now that I might be better off flying at a much higher altitude as trees make the photos very complex. The job is over an hour away from the office so we are motivated to make this data work. Any advice is much appreciated.
Thank you!
Hard to say without looking at your dataset in details but it definitely looks like something that should be possible with CC.
Could you share the aero-triangulation settings you used?
For forested areas you need maximum overlap also wind in trees causes issues with tie points. Vertical views could improve results but elevation accuracy will be lower.