Facade vs Lumigraph/Light Fields

Both are image-based methods for rendering novel views of a scene from a set of photograph.
Facade assumes that we know a 3D model of the scene, but makes no assumption about the locations of the source images.
Light Field assumes that the image are taken from regular locations, but does not involve a 3D model.
How are these two related? Are they special cases of a more general method?

Both are image-based methods for rendering novel views of a scene from a set of photograph.