Forum

Author Topic: HELP!Almost loosing it, sparse cloud at -512m elevation after alignment with RTK  (Read 768 times)

Mr.Vain

  • Newbie
  • *
  • Posts: 31
    • View Profile
Hello forum, hoping someone can help me out here.
I have a flight with our Matrice300RTK with P1 camera and 35mm lens. When i look at the pictures the look the same as all other hundreds flights we have done so everything looks great. RTK data looks fine, flight height also looks great comparint to our gcps. BUT, after aligning phothos the result is really poor. Aligning eith highest will give med just some photos aligned and using loweat will aling all of them but i always get the sparse cloud at minus 500m. It does not matter what i do, what am i missing here?? This looks like it should work like its an ordinary day at work but now i have tried fixing this for a lot of hours without any succes, Posting ascreenshot also with results and settings.

Have a nice day :)

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14965
    • View Profile
Hello Mr.Vain,

Can you share the screenshots of Initial and Adjusted tabs from the Camera Calibration dialog?

If possible, please also share problematic project with the alignment results (in PSZ format) to support@agisoft.com.
Best regards,
Alexey Pasumansky,
Agisoft LLC

Mr.Vain

  • Newbie
  • *
  • Posts: 31
    • View Profile
Hello Mr.Vain,

Can you share the screenshots of Initial and Adjusted tabs from the Camera Calibration dialog?

If possible, please also share problematic project with the alignment results (in PSZ format) to support@agisoft.com.

Thanks for a quick answer, here is a link to the PSZ file https://www.dropbox.com/t/cxpeaVugew9BsaTJ and the screenshots is attached in this reply :).

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14965
    • View Profile
Hello Mr.Vain,

Seems like the coordinate information for the cameras is incorrect, which leads to the alignment issues.

You can try to align the images from scratch having the cameras unchecked in the Reference pane and apply the coordinates after alignment is completed - and you should see very big errors on camera locations.
For me is seems that coordinates are assigned to improper cameras (like 5 cameras shift between actual coordinate assignment and correct one).
Best regards,
Alexey Pasumansky,
Agisoft LLC

Mr.Vain

  • Newbie
  • *
  • Posts: 31
    • View Profile
Hello Mr.Vain,

Seems like the coordinate information for the cameras is incorrect, which leads to the alignment issues.

You can try to align the images from scratch having the cameras unchecked in the Reference pane and apply the coordinates after alignment is completed - and you should see very big errors on camera locations.
For me is seems that coordinates are assigned to improper cameras (like 5 cameras shift between actual coordinate assignment and correct one).

Thanks! I see that now, missed it. it should be 7 digits in the X coordinate but its only 6 before the decimal. Could you see why that has happend?

Mr.Vain

  • Newbie
  • *
  • Posts: 31
    • View Profile
Hello Mr.Vain,

Seems like the coordinate information for the cameras is incorrect, which leads to the alignment issues.

You can try to align the images from scratch having the cameras unchecked in the Reference pane and apply the coordinates after alignment is completed - and you should see very big errors on camera locations.
For me is seems that coordinates are assigned to improper cameras (like 5 cameras shift between actual coordinate assignment and correct one).

Thanks! I see that now, missed it. it should be 7 digits in the X coordinate but its only 6 before the decimal. Could you see why that has happend?

Just tried it but everything seems to be at the wrong place and when importing the gcp:s the wont even show. When looking at the coordinates for the pictures all pictures seems to have the correct coordinates, do we need to take new photos or am i missing something?

Paulo

  • Hero Member
  • *****
  • Posts: 1350
    • View Profile
Hello Mr. Vain,

The projected coordinates in SWEref99 12 are correct. As X camera coordinate are around 149 000 m which is around 11.99 longitude. So it is a problem with your RTK coordinates as photos were taken every second and there are big gaps in position. For example fotos 7, 8 and 9 were taken every second but distance between 8 and 9 is way greater  then 7 to 8...
Also looking ats fotos 8 and 9 side by side, it is clear that distance between them is a few meters...

If you can share the image dataset, I could have a look at it   :)
« Last Edit: March 29, 2024, 09:40:14 PM by Paulo »
Best Regards,
Paul Pelletier,
Surveyor

Mr.Vain

  • Newbie
  • *
  • Posts: 31
    • View Profile
Hello Mr. Vain,

The projected coordinates in SWEref99 12 are correct. As X camera coordinate are around 149 000 m which is around 11.99 longitude. So its is not a problem with your RTK coordinates as photos were taken every second and there are big gaps in position. For example fotos 7, 8 and 9 were taken every second but distance between 8 and 9 is way greater  then 7 to 8...

If you can share the image dataset, I could have a look at it   :)

Of course i can, here it is https://www.dropbox.com/t/pki1MvNhP1gaA2Ei Thanks!

Paulo

  • Hero Member
  • *****
  • Posts: 1350
    • View Profile
Hi Mr,.

Yes disabling the camera coordinates I was able to align cameras but at a tilted angle. Then enabling a few (3) cameras having a RTK flag of 50 (fixed) I got a correctly oriented tie point cloud. Then I measured 5 GCPs and optimizing (4 control and 1 check), I got a very good result. see attached pdf. https://1drv.ms/b/s!AmanbKe_DVntlmiSk8fFZgUlX2Sw?e=AvLtgA

Note your MRK file has a problem as first events has correct RTK flag (50) but then flag changes to 1 which seems a bug...all should be at 50....and following change to 1 all coordinates are the same...
Also alignment was done at medium accuracy, so at high we probably could get better results...

If you need more please send me a PM.
Best Regards,
Paul Pelletier,
Surveyor