No, the real problem is the rise in voice acting. The resources and effort it takes to make a new character with full voice acting (and the more detailed graphics) is considered not worth it if the character is not going to be seen by most players.
Oh yeah, the FMV deal is just an extension of the real enemy.

Not really, the increased level of geometry, quality of special effects, shading and amount of objects in a scene can still (and will always) be much greater in prerendered video.
As I said, the same things just in greater detail. There is now way the FFVII engine could have more than approximated stuff like the plate falling or the sister ray firing. In FFXIII FMVs did nothing that the engine couldn't, it just did stuff with a marginally higher level of detail.