I think xmorph wouldn't be able to do a very good job of figuring out the motion vectors of the objects (or rather of the cameras in relation to the objects) and you'd get a very fuzzy morph.
If you talk to Timeslice Films or Snell & Wilcox you might be able to license their software for a few grand for film-quality results.
I have a 1982 reprint of the (original?) 1975 edition, with lots of descriptions of his experiences with the IBM OS/360 project.
I'm guessing the later editions were radically revised/had extra chapters with more up-to-date experience of software engineering.
Still, i've managed to ref. lots of pertinent chapters in recent essays.
If you talk to Timeslice Films or Snell & Wilcox you might be able to license their software for a few grand for film-quality results.