I am developing a 3rd person object viewer (for a CAD type app.). I am using quaternions to calculate the rotations around an object - applying these to the view matrix to move the camera around. I am a little confused as to how the LookatLH and the Translation matrices interact with each other. I have applied the correct rotation to the camera(view) but now have to compensate the camera translation to leave the camera in the correct overall position with the correct rotation (camera moving around a virtual sphere looking at the sphere centroid). The camera position can be affected by both the LookatLH matrix and the Translation matrix, which one should I use to control the camera position? Does it matter? Seems a bit confusing to have two methods to achieve the same thing!

Hope someone can give me some good advice, this is a CAD type program so it is important to be able to Pan, Rotate, Move etc to just about any position over a large area.

Thanks in advance.