Attempting to do photogrammetry on a 1/2 acre lot with a steep slope and dense trees on half of it. The neighboring buildings get modeled OK with a small number of photos (i.e. using just every third photo from the set), but the tree-covered half is just a few giant traingles. Adding more photos (e.g. using 2/3 of the set instead of 1/3 of the set) causes the image alignment to just fail (registering just two “camera poses” (i.e. registering just two images as alignable). Oddly, the “draft” mode of image alignment works fine, giving me a point cloud that looks pretty good. I’ve tried to change the settings for the regular alignment process to match those of the “draft” mode, but there must be some other details of the distinction between the two modes hidden away, as no matter what I try, “draft” works fine and the regular mode fails. What is going on? If I better understood what it’s doing, I think I could make some educated guesses about what to try to improve it. Even suggestions on what might help, without explanation of why… would be welcome at this point. I’m about 30 hours into this head-banging.
Hello James,
trees are always challenging for photogrammetry (as they are moving and it also depends how far you are from them during capturing).
Are you using the pre-defined alignment settings?
What kind of images are you using? Are they georeferenced?
Have you tried to align all of your images in one project?
Are you trying to do the alignment in the same project as draft one?
Try to follow these advices, but for the image downscale factor use value 1.