When you apply a flat image, taken with a camera, onto a UV map that represents every face of a mesh model, you will end up with only the front looking right. The sides, top, and possibly bottom will have distortion as a single camera shot cannot easily photograph all sides of a face in a single shot.
Several years ago, we would have to take several shots all around the head then manually stitch them together in a paint package before finally bringing them to a 3D package to apply to an unwrapped model. Indeed, trying to match the distortion inherent in stitching flat photos to complex UV maps was very difficult.
Blender, however, allows us to create multiple UV maps, which can be mapped from different sides, and combined to create correct mapping all around a model. The human face is an excellent example of why this technique is so useful. We can use a camera to create shots of the front, sides, back, and top of a head, then map these to individual...