I sometimes have projects where there is a lot architecture in the background that tends to produce misalignments, because it is partially repetitive. The objects I want to model themselves often have too little texture to allow masking the background, as they alone lead to bad alignments, too. I should take photos with more overlap, but can't because I can't get far enough away from the object.
In these cases it would help if I could set markers - I have manually set over 100 in one project - and PhotoScan aligns the the photos based ONLY on my manual markers. The accuracy should be good enough for my task, because I usually have very small objects to set the markers, so my marker placement is highly accurate.
Can this be done easily?