You might want to do a little research, first. Take a look at some of Philippe Bouyer's Vue show reels; he gets the camera effects because he is also a cinematographer, and knows how those physical cameras behave. People don't realize it, but we've been trained to deal with that behavior.....for instance, a pursuit cam. In real life, when you have, say, a chopper come swinging into the intended viewing area, there is a brief delay =after= the chopper appears before the camera starts tracking it. This is the neural delay involved when you have the eye receive the data. Brain processes. Nerves send signals to muscles. Muscles have to overcome intertia and friction to move the camera. And you are going to overshoot your tracking at first, simply because during the delay time, the chopper has moved. It can take a couple of seconds for you to 'catch up' with what's happening.

 

That is why so much CG can look fake; we're used to watching cameras manipulated by people, and that involves a lot of things behind the scenes You don't see much action footage that doesn't have lag and tracking issues, simply because it takes a second or three to get things timed right. Yes, that footage is usually edited out, but if you watch carefully, you can still see the neural delay effect, particularly on long shots with lots of motion.

 Each kind of camera has its own unique ways of behaving, and the closer you get to those behaviors, the more the audience's brains goes 'that's right'. CG cameras can do lots of things that physical cameras simply can not. The question is whether you -should- use a CG cam in a certain way...

 

You will also want to storyboard your intended action. This will save you a lot of aggrivation and wasted render time if you plan things out before hand.