in 2D and 2.5D modelling you describe an object with points and lines. You "pretend" to show 2.5D objects in 3D via some "tricks" like perspective transformation and hidden line elimination.
In a true 3D modelling you describe an object with surfaces, i.e., via the light the object reflects. For example you say that there is a sphere at position x,y,z with radius r and color (r,g,b), relative to a light source and your eye (or the camera) by calculating the "normal" vectors coming out from the sphere to your eye, where you "see" a shiny object with some shading and degredation in color. There are no lines or points (unless you explicitely want to draw one).