You're right, sites do change images, so I should have phrased that more
precisely. What I meant is that there are a great many sites where images are
not *supposed* to change (and if they do, this is indicative of an error). 

I realize the idea of httrack is to mirror what it sees, but in the case of
temporary outages, even though capturing what amounts to an "error image" and
overwriting the original is "technically" correct behavior, from a user
experience point of view, I would think for most that is *not* the desired
behavior. I can think of ALOT of sites that use static images. tumblr, sites
that use amazon's s3 services, etc. 

Httrack would be able to function the same as it always has, and be able to
update anything it sees as "new". But there would be an extra option to be
able to exclude images that exist locally from an update. This would both
conserve a lot of bandwidth and while it might mean a mirror might not be 100%
accurate, it would also guard against unintentional data loss.

Httrack already has options that allow for the mirror to be different from
what it "sees". For example, the option to not redownload a file if it is
locally erased means that the mirror isn't 100% accurate.