I thought about trying to run httrack so it only updates html pages. And then
rerun it in "resume interrupted mirror" mode to see if it grabs the missing
image files, but no matter what I do it seems to always redownload images. I
only want to update html files, and then only download images if they *don't*
exist.

Alternately, is there any way to have download html pages and rewrite all
links so they are relative (even if httrack doesn't download the images they
point to). Then I could run a script and scrape all image links out of the
html files, compare that to the contents in the local image store, and just
grab the images with downthemall or something.