we have a processing pipeline to process drone images.
However when we run the alignment with a certain image pattern the alignment fails.
In the images you can see the red dots which are the images taken by the drone.
The alignment fails, but if we remove the images with the blue outline the alignment worked.
Is it possible to make reality capture ignore these images that cause those issues?
I’m not sure what is being asked here. Are you getting an alignment that does not include those images? or are those images making it so that other images don’t align? do you know before alignment which images will cause the issues? (I am assuming you don’t or you could just not include them) Would you like RC to somehow detect such images and remove them, and if so what would be the criteria to differentiate between those images and the other images? Are the grid dots on the diagram images? Are the images inside and outside of the main circle the ones that you want to remove? we do not have commands to select images by any criteria other than their name.
No RC can only select images by name (regex). To select and remove images using any other criteria you would need to write your own script outside of RC, and it would not be easy. This issue could be avoided in capture. If you want to do a new ring of images at a different distance, start the capture very close to one of the images from a previously captured ring and then move to the other distance whilst still capturing; this should help tie everything together. The other thing you can try is to disable GPS for all of the images in rings; if you have GPS for the grid shots, then your component will still be georeferenced even if all of the images in it are not. you could do that by renaming the ring images and putting “ring” in their name before processing then you can select them using regex and turn off GPS for them