Orthoimage reconstruction artifacts in DSMs from 9.0 - photo lines and pits

[andyroo]

I originally posted about this in the general forum, and another user has reported the same issues so I thought I would add a note here since it looks like it might be a bug.

When reconstructing in smooth/high (50M polygons for ~250 12 MP photos in my case), photo alignment artifacts show up in areas of high overlap, and it appears that overall alignment with GCPs is worse for the same photoset in 0.9 vs 0.8.5.

Original post has several image examples:

http://www.agisoft.ru/forum/index.php?topic=766.0

[Alexey Pasumansky]

Hello Andy,

Thank you for reporting.

We are currently investigating this problem.

[Matt]

Hi Alexey,

I can bundle a 50 photo precompiled project for you to look at if you like.
It has both the photo edges and the circular depressions.
Model is controlled by up to 20 1-2 cm accuracy DGPS control points.

[Alexey Pasumansky]

Hello Matt,

If you can provide us this dataset, it will be very usefull for us.

[Matt]

Hi Alexey, Project link emailed. Let me know if there are any problems.

[andyroo]

Any resolution to this? I am still using 8.5 to avoid artifacts.

[Matt]

It appears they are caused by removing to many points or doing to many iterations during the gradual selection/optimize process as we aim to increase accuracy. I have eased up on the gradual selection limits i was targeting (0.5 max initially, now about 0.8 max) and I have not seen the artifacts recently. Will have another round of photos to process soon so will let you know if they reappear.

[andyroo]

I am confused by this. I don't use any gradual selection tools in the point cloud process, and the only filter I use is the one in the Build Geometry... dialog, where I leave it at the default.

I have not spent much more time trying things in 0.9.0 because I am re-flying ~60 sq km every ten days or so and that keeps me plenty busy with the image processing, but I finally put together a second computer that I could try to reprocess some data with for comparison purposes if that would be useful.

I have projects that I completed in 0.8.5, complete with GCPs, and I could reprocess the "Build Geometry" step and compare DSMs if that would help with anything. Unless I don't understand the gradual selection process, I don't think that played a role for me.

This problem is still keeping me from switching to 0.9.0.

[Alexey Pasumansky]

Hello Andy,

And do you have any sample projects you can share with us for our internal studies?

[andyroo]

Finally got one uploaded that shows pits/lines in 0.9.1 and not 0.8.5.

Andy