Instead of loading the images in a list, thus requiring all the extra zoom/scroll interface adjustment, we should show images in a thumbnail grid with a badge of how many GCPs, if any, are in a given image. User chooses thumbnail, which expands to fully zoomed-out image side-by-side with map. Then user adds point and moves it to the correct location on each side of the interface. When complete with that image, chooses some sort of completion button, which brings them back to thumbnails. Rinse. Repeat.
From Steve: