Here you can see the camera positions and GPS
And here the sloping roof ridge, which should actually be horizontal.
I have more extreme examples, but what could it be?
Is it a problem how I shoot the photos?
I actually do it identically every time and these crooked models only occur sporadically. But recalculate again comes back to the skewed result.
how can i prevent this from happening in the future ?
(I know about Groundlayer adjustment, or what ever the englisch name is 
)
Hi Tim,
are you using GCPās? If so have you tried setting the absolute pose to unknown? First select all of your images, then in the selected inputs panel in the 1D in āprior poseā set āabsolute poseā to unknown.
Hi Wallace,
No I donāt use GPCās the GPS tag in the EXIF information is my only source of information for dimensions.
I do use the markers you can generate with RC.
Here you can see the problem again with another object. Even if here the errors are somewhat stronger.
The drone is a Yuneec H520E and has flown a mission on actually exactly 25 meters (over starting point), the wave movement is perhaps to be explained by the fact that the drone holds the height possibly not exactly
Hi Tim,
Ā Ā Ā Ā Ā Ā Ā Ā Ā you can try going into Application Settings and clearing the cache, then try setting the Absolute Pose, as described above, for the drone shots to ālockedā, this way your cameras will not move on alignment.
Ā Ā Ā Ā Ā Ā Ā Ā Also for future shooting you may find using a grid pattern with small angle changes will work better than a circle.
Ok I will try that!
The grid pattern I am aware but rarely feasible, because I may only fly over the residential property and therefore it is difficult with an automatic profile to get the right pictures. There are most of the times only 15-40 meters diagonal space that I can fly in.Ā
For large buildings, however, I already use the grid pattern! Since it is definitively much better than the Olique photos. Or i use both kinds.
And by the way, I donāt fly a direct circle but along the property lines with the camera always pointed at the house and a photo is taken every 3 meters. Thus, only very small angle changes and high overlaps.
But I will definitely try the Grid pattern again again to maybe come to a solution for the problems that occurred initially with it. (too few photos or far too many, hardly any walls on small buildings)Ā
Thanks alot Wallace!
I wish you Merry Christmas and now already a happy New Year!
EDIT:Ā
When I lock the Absolute Position, Yaw, Pitch and Roll are also locked and I donāt know what these values are. AndĀ therefore there is no result.
Hi Tim,
Ā Ā Ā Ā Ā Ā Ā Ā We were hoping the orientation would be stored in the image file, the locked solution is not the one for you. Drone GPS is prone to inaccuracies, as indicated by the orange lines from your cameras on the screenshots. By far the best solution for you would be GCPs, and the second best to iron out some of the inaccuracy would be a grid flight path. Probably the easiest fix in your situation would be to set the ground plain to level the model out if geo-location is not as important.
Ā
In most cases it works wonderfully, only in rare cases it becomes crooked.
Just so I donāt misunderstand, by GCP you mean this(I already use this but its crooked with and without anyway):
Or GCPās from which I know/ measure the coordinates?
I think then it would be much faster and more cost effective to just continue to set the ground level plain.Ā
Thanks again for your quick response!
By GCP i mean Ground Control Points, which as you suggest are points that you know and measure the coordinates. Yes I think setting the Ground Plane may well be your best option. Have a fantastic Christmas