Anyone else having problems with control points looking like this?

Hello John

I was able to have a look at your dataset approx 715 Nadir and 223 oblique images captured with dji fc6310r camera, you provided to our support. After processing the images I also agree with Sylvain that the photo quality and areas of Nadir only imagery are causing the distortion you see. 

 here you can see the actual image on left and the model at the same zoom extent on right

here is the same model where the are of the GCP used both Nadir and oblique 

The mis-alignment that Sylvain referenced may be due to the following..

1. lack of a camera calibration - Be sure to perform a camera calibration on each camera lens combo you have. The FC6310r can export photos in 2 modes , distorted and undistorted, each mode needs a different sensor size input when running the calibration AT on the photoset, I believe this may be due to some sort of cropping of the images dependent on how they are saved from the camera

2. be sure to save your calibration to the camera database and use the appropriate camera on phot import. -NOTE: we will post the camera calibration document for your reference 

3. be sure to import your photos as Ellipsoid - the geotags on the photos appear to reference ellipsoid - this will make finding your GCP which reference NAVD88 much easier

4. concerning Nadir / obliques images- these appear to be flown at the same height, when using the same camera lens combination it is advised when safely possible to fly obliques auto missions at a lower altitude in order to more closely match resolution of the image between nadir and oblique images, this will help with alignment... your obliques angles look good, but be sure not to exceed 45 degrees (stow camera point up "0"degrees)

Hello John, 

Here I can see you are displaying images of the 3D mesh. Such distortion can be reconstructed because of the quality of the input photo, the capture process (nadir only will generally show greater imprecision along the Z axis), and camera calibration. 

This said, I am wondering if there is not some photo missalignement as well, as I can see an orange stain that shouln't be there ? (Having a few screenshot of the actual control points in the images would help validating this). 

By the way , Update 18 of ContextCapture has been released, with an optimized AT engine. I believe it would be worth trying it.

Thanks,

Sylvain

If automatic tie points are not lost around GCPs after adding GCPs measurements then no idea.
But if you fly only Nadir then such distortion could be expected as it can't create proper calibration from Nadir only photos.
Reuse calibration from 3D flight and lock them(Keep camera calibration preset) in aerotriangulation.

This was after AT.  Since I am running and RTK drone using a network I typically import my photos.  Import my control.  Tag my control.  Run AT using Photo Meta Data Only.  Then run an AT using control on top of the Meta Data Only AT.

Is this before or after aerotriangulation with GCPs?
Also it is better to do 2-step process. At first run without GCPs and after that add them.
What camera did you use?