I'd go with "cinematic/close-up if a PC can answer". If it's just a greeting or a generic oneliner, there's no need for an intrusive close-up.
It's generally two types of "realistic" competing: cinematic allows for more nuanced acting and you can see the characters up close and personal, while "in-world" is seamless and doesn't feel like "now watch a movie".
That's pretty much how I feel about it. The flow of the game gets broken up a bit for me because of it.
As much as I appreciate the models and the work that's gone into the animation and the actors who pantomimed that stuff out, getting sucked into a dramatic dialog every time I'm speaking to inconsequential randoms kills my immersion.
