Torn dense cloud and DEM after update

[Ungrim]

Hello, few questions. I've been doing same work every month for over 18 months. Since last update of Metashape I have some issues. Maybe I'm just doing something wrong but I want to make sure.

First of all my GCP lost accuracy. It was never an issue and I can't really say what can cause a problem here. Maybe now they are more accurate and closer to truth than before .

Issue that really bothers me is torned dense cloud and DEM. It looks like at the screens. Any solutions? What am I doing worng?

[Alexey Pasumansky]

Hello Ungrim,

For me it looks like the alignment issues.

Was this sample dataset captured during the single flight? Which processing parameters have you used for alignment and optimization? Please also provide the screenshot of the Adjusted values tab from the Camera Calibration dialog.

[Ungrim]

Hi Alexey, thanks for reply.

Yes, it was a single flight and it was aligned without issues. I got another mission even in different zone of coordinate system from last week and it looks the same. I was wondering if that could be a problem of bright sunlight but all photos are exposed correctly and with perfect focus.  But I was already flying in such sunny conditions and it was fine (attachment from January)

Aligning parameters are: high, source, generic preselection, key point 40k, tie point 2k. I attached screen from camera adjustment tab as you said. Camera is from DJI P4 Advanced.

[Ungrim]

Any update with this issue? I got more locations photographed and more results. At three locations dense is torn and looks awful. One of the location was flown twice, first with bright sun, second with fully clouded sky. Results are the same. I am really out of ideas.

At all locations I got issue with gcp errors. It could also point at the align problem right? I took measurements with two different gnss receivers and compared the results from the previous several months and they looks good.

I am really concerned about this because one of location's dense cloud was out of any acceptance. I had to emergency install competitive program which given correct results. This is definetely something in Agisoft itself that I am doing wrong. But I don't really know what it could be, because it is first situation like this and I am using Agisoft for 5 years.

[Paulo]

Ungrim,

I am worried about your adjusted focal length of 3008 pixel which is almost 20 % off the initial value of 3648... Maybe you can look at distortion plot to see if residuals look good as well as Correlation tab...attached is screen capture of successful adjustment with FC6310 camera...

And also,if you can look at successful previous alignments and compare the adjusted focal...

[Alexey Pasumansky]

Hello Ungrim,

Is it possible to share the project with the alignment results (for any of the problematic projects) and the original image set with support@agisoft.com?

We will try to investigate the problem on our side then.

[Ungrim]

Thanks for reply Paulo. Unfortunately focal length in the newest project seems to be just fine but at the other hand distorsion is worse. I take a closer look at this topic.

Thank you Alexey, for taking a closer look at that. I would send aligned project and photoset tomorrow.