Chiefly a workflow thang … while I'm working top to bottom through the source file, it's easier to deal with each graphic as I come across it; including giving it a meaningful name derived from its context. With other documents, I've sometimes used the save as .html option to extract all the graphics.

I've ended up taking my own advice :-} save graphics on the fly to Windows' preferred location, add links using that location so I can check the result; end of session, move all graphics to my preferred location and s/r in .xml source to update references.