Since the imaging approach seems incredibly complicated, and there are many (conflicting?) answers to the original question, I'm going with simplification. My backups do not generate an image (just file backups). When I want to do an image I rename the previous one and then manually request a new image. They take a while to do (since each is a full or base image) but there are always big periods in a day when the PC has nothing else to do anyway.
I also figure I only really need 2 images: a relatively old one (one or two months old) and a new one. I'm thinking I can restore from either one and then use the file backups to get up to date.
--Larry
But the problem with such strategy is the HDD lifetime. I mean it's not good if you write many TBs on your external HDD monthly. I'd rather do incremental/differential image. I try use my external HDD as less as possible because I want it to survive many years.
I think you imply 2 problems:
1. One should always have at least 2 external disks for every PC and alternate the images - just in case one disk breaks (which they rarely do).
2. If you have many GBs of stuff, most of it is data. That should be seperated from the OS into a data partition. Then you can image the system (which is usually between 20 and 40GB in size) and backup the data with a sync program.
The system images will be between 11 and 22GBs and you should keep many (I always keep at least 25 images). The data sync moves very little data around (just what you altered since the last sync) and should not be stressful for any disk.