Any windows-based imaging software must take a snapshot of the partition prior to imaging. If this is not possible, then the software will require that your reboot the PC so the system can LOCK ON the partition (going back to DOS). Sometimes, the snapshot in windows can go wrong, resulting in a bad image file.
I'm from the old school, which tells me that the partition must be 100% dormant DURING the imaging process. This method should yield the MOST stable image file because NOTHING is changing during the imaging process.
If you CANNOT shut down windows to backup your partition (24/7 application), then a windows-based imaging software may be of benefit. For all other users, it's best to do the imaging stuff in DOS or Caldera DOS.
The old Drive Image 5 can schedule a backup session, and it runs in DOS. I don't recommend incremental backup because if you corrupt the source image file, then subsequent incremental backups can be useless.
It is very rare for programs to go bad. Most of the time, it's the OS. That's why I set up a small primary active C partition ONLY for windows. All other stuffs are stored in other extended logical partitions. By keeping the OS under 1.5GB, I can image the OS partition C in less than 2 min. At maximum compression, the resulting image file is under 700MB.
Ask yourself why these windows-based software must require over 20MB of hard drive space to image a partition? Complexity is never a good idea in the software business.