Camera - A viewpoint from which the scene will be rendered.
OGRE renders scenes from a camera viewpoint into a buffer of some sort, normally a window or a texture (a subclass of RenderTarget). OGRE cameras support both perspective projection (the default, meaning objects get smaller the further away they are) and orthographic projection (blueprint-style, no decrease in size with distance). Each camera carries with it a style of rendering, e.g. full textured, flat shaded, wireframe), field of view, rendering distances etc., allowing you to use OGRE to create complex multi-window views if required. In addition, more than one camera can point at a single render target if required, each rendering to a subset of the target, allowing split screen and picture-in-picture views.
Cameras maintain their own aspect ratios, field of view, and frustum, and project coordinates into a space measured from -1 to 1 in X and Y, and 0 to 1 in Z. At render time, the camera will be rendering to a viewport which will translate these parametric coordinates into real screen coordinates. Obviously it is advisable that the viewport has the same aspect ratio as the camera to avoid distortion (unless you want it!).
Note that a Camera can be attached to a SceneNode, using the method SceneNode::attachObject(). If this is done the Camera will combine its own position/orientation settings with its parent SceneNode. This is useful for implementing more complex Camera / object relationships i.e. having a camera attached to a world object.