What interests me is the following: In order to model cavity (blackbody) radiation, one FIRST has to calculate, from the boundary conditions, what the allowed frequencies can be. This requires standing waves stretching over the whole cavity. Then you have two choices: (i) you can use Boltzmann-statistics on the standing waves, or (ii) you can use Bose-statisics to determine how the photons fill the energy levels. Now in the first case the waves outomatically "know" which frequencies are allowed, while in the second case, the photons are assumed to have a limited size; so how do they know which frequencies are allowed? Does this not imply that the radiation in the cavity does exist of standing waves, while the radiation, measured through the hole in the side, exits and then morph to form localised photons. If this is correct it means that quantization does not necessarily implies granularity of "light size", but only that a wave with a certain frequency, whatever it's size, cannot have less energy than given by the Planck's relationship between energy and frequency.
