Return background images to 3D viewport, not camera

I don't see any advantage of having the background images in the camera rather than the 3D viewport. It does mean that if you want front, side and up reference images, you'll have to add 3 cameras, and switch your active camera, or otherwise, use images as empties, which is having extra objects in your 3d space, making your scene more crowded.

Maybe am I missing the whole point here, could any of you guys clarify this for me?