The idea sounds nice and i'm sure the plugins sound nice. However, to claim that this is like masterbus processing is in my opinion not fully correct because the behaviour of these plugins is going to be different from what I can tell with the information provided here.
If I have 10 tracks playing at the same time and add 2 dB of 10kHz, applying it to 10 individual tracks will yield a different result than 2 dB applied to the masterbus (= the result of the 10 tracks summed).
If I have a compressor on the masterbus, the threshold is applied to the sum of the tracks. This means that the sum of tracks will trigger the compressor to duck the volume when that sum passes the threshold. However, with individual tracks, it is very possible that as individuals, they will either not trigger the compressor as the threshold on the individual track is not passed, or they might trigger at a different point because their peak might be at a different time than the peak on the "masterbus" (or in Atmos terms, perhaps the bed), or they cause a different amount of compression.
The way to design an Atmos (or any other format with objects) compressor which will act like a real masterbus processor (= meaning on the full sum of all tracks) is on paper easy to do but to actually pull it off is difficult and pretty CPU intensive as it involves a lot of tapping into feeds and routings:
1) You have to first make an output mapping to a channel. So if you are using a 7.1.4 setup, the compressor will be 12 channels
2) Some tracks are sent to the bed (or beds…), some to objects. You basically have to be able to fetch both the audio of the objects and the bed channel, sum those together, and then that signal is basically your compressor sidechain (as in: the sidechain used within the circuit to control itself with).
3) Then all instances should get fed that sidechain in order to have the same compressor behaviour applied identically on all instances of the compressor on that compressor channel (though you will only use these instances on objects, the bed itself is basically like a mix bus);
4) Now that poses a problem on objects which are in motion: if you do not apply the same compression to all 11 channels (LFE channel is taken out here as I believe objects cannot be routed to LFE) there will be a mismatch when the object pans around. To fix that, you need to feed the object information into the appropriate channels and get the right sidechain fed back in real time of the pan movement. In other words: the sidechain needs to pan along. Usually you make groupings (front, back, heights, something like that).
You maybe also need information from the render to determine how the audio is fed into the speaker setup based on pan because this strategy differs of course from setup to setup, meaning that if I now switch over to a 7.1 listening mode to see how it all folds down, the compressor probably should react appropriately? I was thinking that perhaps a compressor should react to whatever the render creates because that is the place where ultimately objects and beds come together and the signals are mapped to dedicated outputs. Ideally, the masterbus processing is after the render.
The other solution is using just the bed as the sidechain. For bed heavy mixes, that can work. But if you still apply different compression to speaker/output groups, then the panning thing still remains an issue.
Fwiw: I created a template in Logic to apply "Brauerizing" in Atmos (why? just for fun and because I could…). That works on the bed, it doesn't work the moment you start introducing objects for exactly the reasons mentioned. And I didn't get the sidechaining for objects to the appropriate A–D bus to work in the first place.
I'm not writing this to dismiss or bash your product. On the contrary, I expect that the plugin sounds great and can be very usable to match settings quickly. But based on the descriptions provided, the result will be different than a regular masterbus would give. Can still be a very usable and great sounding result, but this is a different approach.
That said, I think the notion of wanting to apply masterbus processing as if it were a fixed bus (2, 5.1, 7.1) is perhaps outdated in the context of object based mixing (Atmos being the prime example but not even the only format out there). I think it is up to the engineer to prioritise something similar to gluing a masterbus (meaning bed based mix with minimum objects) or create an object heavy mix but less glue.
