here are some small (and perhaps unimportant) things you could change that would make, imho, a big difference in the feel of your program (mac version):
 
1. study the mac UI standards more carefully.
 
2. write a IM icon for IM data files rather than a generic one.
 
3. use standard mini menu bars for palettes (check adobe CS2)
 
4. spell 'navigation' correctly (view menu)
 
5. get a power IM user to flesh out your user manual
 
6. get the QTVR export feature working properly, or at least more intuitively (and in any case with better documentation)
 
7. get the reference distance thing worked out: if you don't happen to hit 'enter' after putting in the distance, your measurement won't be accepted. this could be a lot simpler. and, based on a recent discovery, even if you do hit 'enter' the measurement doesn't always stick. this is a bit of voodoo that has to be worked out.
 
8. saving: it's better that it was in 3.5 for sure, but still not perfect. if certain tools are selected you can't save, for example. and, when i reset my ref. calibration, IM didn't consider this as worth saving. i had to add a ruler to get it to recognize that something had changed.
 
9. the 'apply to all views' button: sometimes works, sometimes doesn't. i haven't figured out the magic sequence yet.
 
10. menus sometimes stay highlighted; i just had this happen with file menu, after using the keyboard command to save. I thought the program had hung, but it was simply that the file menu remained highlighted. Changing tabs between measuring and calibration had the same effect.
 
11. This is a big one, but i don't know much about it: would it be possible to implement aspects of single view metrology (i'm thinking of the work done by pierre macé at the ecole d'architecture at la villette) to allow calibrated spaces using a SINGLE 2d image? This would be really amazing.