One of the DVD2One fellas online can probably confirm this, but the impression I've gathered over the last few weeks is this:
DVD2One identifies which track is the main movie, and allots a higher quality rate to that video. Extras and menus are considered more expendable, and so get crunched more to save as much space as possible for the main movie.
Now that VARIABLE and CONSTANT are introduced as options, thing have changed a bit.
VARIABLE is the original compression hierarchy as described above. Some scenes (or video tracks) get heavier compression depending on whether DVD2One thinks it can take it, and most of the time does a fine job.
CONSTANT more closely follows the source disks peaks and valleys of encoding quality. Instead of favoring just the main movie so much, everything on the disk is coded at more or less the same percentages compared to the original material (which is not to say that it's the same numeric encode throughout -- when the source codes high, so does DVD2One, and when low, it goes low, etc.).
It's a different ball of wax than the CBR/VBR bitrate options we're used to from direct encoding. Even movies compressed with DVD2One at CONSTANT setting show lovely graphs of peak-and-valley bitrates, if the source material had 'em.
So, why is VARIABLE recommended for full-disk backups? Because it's designed to donate more quality space to the main movie, and squash the hell out of the rest of the disk.
And CONSTANT for movie only? Because you'll want the best quality for that one track, and DVD2One relies on the wisdom of the original encoding rates to determine how much to compress each frame.
It's also very true that quality depends on the source material. I've made brilliant backups of 3-hour movies that look fabulous, because the video was prepared so well. And I've made (and rejected) awfully scrubby backups -- often from DVDs of older television sources -- that are only 90 minutes but are as blocky as, well, a really blocky thing.
ReneB, does that sound about right?