A couple of friends and I are working on making a copy stand style film scanner. Our goal is to make a set of open source plans so that photographers can DIY their own scanners. We use a dslr, a macro lens at 1:1, and a moving negative stage.  For instance, with my D200, I capture 16 frames in a 4x4 grid to cover a 6x7cm negative.  We'd then like to stitch the frames together to give us our final "scan".  We have manual prototypes working, and we've been able to get better scans than with consumer flatbeds.

One issue we have is that it's hard to align areas of a negative that are relatively featureless, such as a wide expanse of clear sky, and we'd like to come up with a reliable stitching solution for our situation.

The system involves multiple viewpoints with no changes in camera rotation between shots.  Thus, the multi-viewpoint feature should be helpful, right?

Currently, we're working on automating the movment of the negative stage using stepper motors.  As a result, we should be able to make very repeatable grids of scan.  It would thus seem that a type of align to grid feature would be helpful, but this only appears to be available if your images with captures with specific motorized pano-heads.  Is there any way to use this feature in our situation?

Any other suggestions as to how to improve our stitching workflow would be greatly appreciated.