Model crashed after merging chunks?

[toffersan]

Hey there,
i was merging 3 Chunks of aerial photography made with a dji phantom drone. All pictures got GPS but no absolute elevation, only elevation above the ground(and this with some error). after merging the chunks it seems the model is gone and i can't find it with the trackball(Screenshot 1). I think it is some problem about the elevation (Z axis) because gps is given in this overlapping models.
I attached a screenshot of my problem. i tried the predefined views to find the model and i tried reset region. After merging i have a very high error of 6 000 000 m in the reference frame(Screenshot 2).
After optimize cameras i get the message: Some cameras have insufficient number of projections and will be reset. Continue? Then i can see the merged chunks as a sparse dense cloud. But not fitting together in the z axis(Screenshot 3).
So anyone can help me or has the problem?Where is the model and why all all cameras pointing to one point (seemingly)?
Thanks & Greetings toffersan

[Alexey Pasumansky]

Hello toffersan,

Can you please specify the following:
- are you using the chunk alignment operation or you are merging the chunks basing on their georeferencing information?
- are the coordinates applied correctly in each chunk using Update button on the Reference pane before merging?
- which chunks you are merging from your first screenshot?
- also provide the screenshots or text files with the coordinate information (source values) for each chunk before merging.

For mee it looks like there are some invalid coordinates in one of the chunks for the disabled cameras (like 0,0,0) so on the first screen valid cameras are placed to one point and invalid (light-blue sphere) - to another. The distance between such points in geographic system is very large, but by default PhotoScan is trying to display all the camera positions, so it looks like all the aligned cameras are stacked to one point.

[toffersan]

Hello toffersan,

Can you please specify the following:
- are you using the chunk alignment operation or you are merging the chunks basing on their georeferencing information?
- are the coordinates applied correctly in each chunk using Update button on the Reference pane before merging?
- which chunks you are merging from your first screenshot?
- also provide the screenshots or text files with the coordinate information (source values) for each chunk before merging.

For mee it looks like there are some invalid coordinates in one of the chunks for the disabled cameras (like 0,0,0) so on the first screen valid cameras are placed to one point and invalid (light-blue sphere) - to another. The distance between such points in geographic system is very large, but by default PhotoScan is trying to display all the camera positions, so it looks like all the aligned cameras are stacked to one point.

Hey, thanks for the fast answer! I merge the chunks bases on their georeferencing information without aligning chunks. I did not use the update button before i merged, i will try this now!
I merged the first three chunks ( Fly 1, fly 2, fly 3)
and here is a screenshot with the coordinate informations.
Thats sounds plausible that one camera with incorrect coordinates is the source of my display problems

[toffersan]

Hey there, okay updating the single chunks helped. But the merged chunk shows the the single chunks at different elevation, because i flew with the drone at the same "relative" elevation of about 120m above the ground. That while all camera positions are on one level, but the model itself is not consistent.

Anyone has an idea how to solve this elevation problem?