[Edited 23:56 Sunday 21st Oct]
Background
I'm somewhat underwhelmed by the results of binaural mixes I'm getting from Dolby Atmos in new Logic.
Professionals will of course do what they must to comply with their current and future contracts. But I'm not a professional music maker, and I plan to continue spatialising my mixes without Atmos, Ambisonics, or any other ready-made big-time solution. *Sings* - "We don't need no Dolby Atmos ..." Lolol.
I much prefer to mix with headphones, for listening with headphones. It's estimated that around 80% of music consumers worldwide listen on headphones of some kind (including earbuds, etc.). As a target market, that'll do just fine for me - I'm not greedy, Lol. In the early '70s I learnt by experience what kinds of loudspeakers I really, really appreciate; but they're certainly not for domestic use. And since I've never had any intention of having my own soundproof studio, I've happily settled for good headphones rather than using so-called "home-studio" monitors - I've no wish to upset others with noise nuisance, nor compromise my ideals of loudspeaker quality.
I'm guessing a lot of other music makers are also most interested in headphone mixes. Nowadays that will of course increasingly involve so-called "binaural mixes".
Moreover, it appears to me that "height" needs either a good 3D loudspeaker setup, or else heaphones with fast head-attitude tracking; otherwise, placement or movement of height in a binaural mix is, for me, too indistinct and ambivalent and too lacking in any definite "dimensional" advantage. So I'm talking about 2D spatialisation here - happily so.
As an amateur music maker who is aware of and already using spatial sound technology but with no intention of installing a set of 3D or even 2D surround speakers, as far as I'm concerned Atmos is much ado about nothing.
Doing it yourself
1. Binaural panning. Ever since Logic 8, Logic users have had access to a splendid binaural panner option built into the mixer. I've yet to hear any qualitative advantage at all from the new Atmos object panner over Logic's own binaural panner; indeed I prefer the qualities of the latter. Nowadays for users of other DAWs there's a fairly good choice of binaural panner plugins - some of which include height, some not.
2. 'Binaural' reverb. I tend to use a separate stereo algorithmic reverb plugin on the output of each binaural panner, and alter the reverb algorithm slightly for each. The near-far dimension is already partly taken care of by the binaural panner, and partly also by the dry-wet setting of the reverb. Reverb pre-delay can help with realism where instruments are at significantly different distances from their nearest main reflective surface(s). Some gentle balance-panning can be used on the output of any of these reverbs to fine-tune the general whereabouts of its reverb "corona" if and where needed. I'd caution against using any width-enhancing control in the reverb - it may actually spoil the binaural panning effect.
3. Best sound sources for Binaural panning. In Synchron libraries my choice of source for the larger orchestral sections is pretty much always the "Main" (Decca Tree) pair. When using this source for string sections I'll typically split Main-left into one binaural panner and reverb, and Main-right into another binaural panner and reverb, then set the two panners to give a wonderful 2-dimensional realism to that whole string section wherever I place it on stage. For an individual instrument or a small section I'll typically use a single mono channel, such as Main-C, or a stereo pair (such as Close) summed to mono; of course this can serve VSL's VI libraries very well.
If this is new for you I urge you to dive in and try it out. I'm pretty sure you'll quickly get the hang of binaural mixing without any further waffle from me or deliberations by you.
There's little need for the underlying science to get in the way of experimentation by the user, except perhaps just to remember these 2 guidelines: (i) the binaural panner introduces a small but precise and very necessary timing difference between its 2 output channels; so avoid adding any effect that can interfere with that timing difference; (ii) the binaural panner affects the spectrum of each side according to its so-called "HRTF" (head-related transfer function) mathematical model; so avoid adding any effect that can interfere substantially with that spectral difference. Reverb can of course mangle both of these aspects but that doesn't matter too much, just so long as there's enough dry binaural signal to satisfy the ear's innate location detection for each and every individually binaurally panned source. We have millions of years of evolution on our side; learn by experience to place due trust in your hearing's binaural capabilities.