For a very basic camera system, just translate the camera x units back and y units up from the player avitar and comupte a look at transform from this position to the players position. This should work nicely for a simple 3dperson camera. But it would move very riggidly and things would get in the way of the camera. I''ll outline a more advance camera system below to give you some implementation ideas for a more advance system, but note that this is a pretty tedious task, but the results are well worth it.




For more of an advance camera system i look at the camera like any other object in the world (similar to the player, enemies, or any other dynamic object in the scene). So the camera will interact with the world like any other dynamic object; it must obey the collision detection rules (with some exceptions to help maintian constant visiblity of the player). This type of implementation would require a dynamic object-to-static world collision detection system, although dynamic-to-dynamic would also be helpful. Collision is used to check for visiblity between the player and camera as well as forcing the camera to abide by the restrictions of the world (this maintains a level of emersion for the player) .

Additionalliy, Like any other autonomous object in the world the camera has a set of objectives it tries to achive. These objectives will outline the way our camera behaves.

-most of the time, we want it to always look at the player (focused object)
-some times we want it to look a bit in front of the player (so that we can see what the player is looking at....this is helpful when the player is moving forward and turning at the same time)
-we want it to try to aways maintain a position behind the camera.
-the further away from this ''desired'' position the faster the camera will move to try to reach this point.
-if there is something that is between the camera and the ''focused object'' (obstruction) then we must try to adjust the cameras position so that their is nothing between the ''focused object'' and camera.

noticie that the above set of behaviors for our camera assume that there is no initial obstruction between the camera and the player and that there is will always be some point where the focused object is visible from the current cameras position.