hey i wanted to share some features that i think would make this much easier and produce better results (for most people) and most of them should be fairly easy to implement.
If tracking is lost create a checkpoint in the process one that can be manually returned to in a list form so you can roll back/forward to get rid of incorrect data.
allow setting of constraints or ability to mark specific features/edges/boundaries prior to building something that can be returned to as a reset point posibally by taking a snapshot single image as a reference
allow switching between the 3 imaging types as the primary focus the main image is often distracting and can be misleading
allow manual reset of position (moving and turning the model to match what the camera sees and preventing a guess as to the location)
add some form of adjusting before a mesh is generated, marking reference points etc.
allow a photo mode (goes with last point) so its shows you the individual images and allows you to discard images that provide poor data.
the point before the mesh is generated allow trimming of random artifacts (floaters)