Rather, the metadata are ignored and the bed is played as-is, with all elements present.
Correct. But what you are referring to is not "the bed".
I like your use of floor rather than bed. Bed has a technical meaning (not object), that technically doesn't belong in this discussion, but is used all of the time.
Correct.
Let's educate a little. Without acrimony, as a Spanish politician said.
The
"bed" channels is the
Atmos concept of 7.1.2 fixed speakers locations where sounds can be assigned without using "Objects". 7.1 are the known
floor channels and the .2 are the
Top Middle channels just above the listener. Top Middle channels
are bed channels.
The multichannel file TrueHD (7.1) or DD+ (5.1) is seen as having those number of "
channels": 7.1 or 5.1
Additionaly, those mch files can contain a
substream of metadata. Metadata describes sound content for the
Atmos "bed" channels (fixed locations of 7.1.2) and/or
Atmos Objects (whatever 3D locations).
When the processor does not decode Atmos it just
"ignores" the metadata, because it does
not understand it. It does not understand Atmos. Then it only "see" the mch file TrueHD (7.1) or DD+ (5.1) with those number of "channels" (not bed channels). Bed channels is an Atmos concept, and there are No "Bed Channels" if we do not decode Atmos, because we do not understand Atmos. Then the whole content 7.1 or 5.1 is played according to the AVR capacity, as in pre-Atmos era. That 7.1/5.1 substream, as well as the 2.0 substrem are generated when the TrueHD Atmos is rendered from the Master mix and contains ALL mixed sound. But this is done during the consumer TrueHD file creation at the production studio, not during the AVR decoding.
When the AVR processor support Atmos, it reads the metadata substream, and for each sound described in the metadata (3D location and size) (etither bed channel or Object), executes a
substraction from the mch file channels substream and an
addition to the available Speakers to image the sound 3D location (and size) as better as possible using the available speakers.
If the mix is done using Objects located in 3D space in locations other than bed channels, then as more speakers available the better is achieved as the mixer engineer intended. This way of mixing could be considered as "good" using the available potential of Dolby Atmos.
If the mix is done using only the bed channels, then the mixer engineer is not using the full Atmos potential, and not necessarily the mix could be considered "worse". It would depend on how well he has crafted the sounds for each channel to get the image effects that we know can be achieved, even from just Quad 4.0
Of course, the Objects potential of locating a sound at the 3D coordinates we want and with the "size" we want is more advanced and would get better results if well done, than with just single locations of bed channels.