M/S processing is transaural I suppose, and that's not without it's uses; but once something is fully panned to one channel or the other, it no longer contains any inter-channel information, in terms of amplitude, phase, or delay.
As a result, the approach really 'works best' only when things occur in BOTH channels to SOME degree, either from being "panned in from the edges" or acoustic crosstalk between microphones, whether in coincident or spaced arrays...
Which explains why with the old 'stereo widening' controls, mixes which were BLENDED a little too 'mono' tended to benefit, while mixes which were full-wide tended to become 'artificial' sounding.
There's also a lot of other factors; suggestion being one of them. For example; when the brain begins to have difficulty accurately localizing a source, it falls back on a simple assumption: The horizontal-plane 'circle of awareness' around the head, (which it builds up through a combination of sight and perceived angle of arrival) has a sighted side (the front) and a blind side (the back). -Since we are sight-dominant, it over-rides any angular vagueness of things it can SEE, using sight for confirmation/correction. -However, for things it cannot see, it assumes that things which it can be fairly sure of must be in FRONT, and things it cannot be so sure of may be BEHIND.
This -believe it or not- is a strong survival-linked instinct. If caveman heard a tiger roar, -but couldn't be sure if it was before or behind- positional vagueness or uncertainty prompts a head-turn to the side. -By this mechanism, anything which SEEMED 'central' (equal amplitude and/or arrival times at both ears) could of course be that way because it's front-and-center, OR rear-and-center... With a nearby saber-toothed tiger, knowing the direction would be VERY important!!!
So by turning the head, -for example to the left- anything behind would now be much louder in the left ear, anything to the front would now appear louder in the right ear.
Interestingly, any of you who have a pet dog can observe a related head-shift, but I'll come back to that later, if I remember...
So... here's where 2-speaker transaural stuff ALWAYS falls down: when you turn your head to try and pinpoint where things come from, your mind's 'circle-of-awareness' image 'collapses', because the confirmation cues aren't there.
Head tracking is thus the ONLY way to make surround work on headphones. Any devices which try to offer 'surround on headphones' without head tracking are utterly pointless.
VERTICAL position can be strongly suggested. -For example, -even though there is NO playback mechanism to direct signals upward or downward, people will ALWAYS hear birdsong as being 'up'. -Try as hard as you like, on a stereo system with your eyes closed, you can't MAKE birdsong appear as coming from down by your ankles... It's a mental conditioning, and just one simple example of how mental conditioning provides a fertile ground for psychoacoustic 'suggestion'.
Oh, and just declaring something as 'psychoacoustic' doesn't make it clever or mean that attempts to analyse, describe or explain what is happening or why should be abandoned, yet I note that many people seem to take the introduction of the word as an opportunity to drop some line of reasoning... it's most curious.
Okay... the dog thing. I've just been house-sitting and pet-sitting for my mom-in-law, who has a Cocker Spaniel, which actually used to belong to my wife and I.
Anyhow, if you watch what a dog does when it's very curious or interested in something and wants to know more, it will often cock its head to one side and raise its ears.
There are times when a dog (which like us is sight-dominant) is unsighted but can LOCATE things through listening. Take the example the other day of my wife going into my mom-in-law's garage to get something from the garage...
She was talking on the phone, as she walked to, through and back from the garage, but she left the door open a small amount, rather than fully closing it behind her. -The dog -who was very curious as to where she was going and what she was doing- followed her to the door, where it waited, with its head cocked to one side, listening to her voice as she walked away from and back to the door. -You could tell from the position of the dog's head that he knew EXACTLY how far away she was, on the other side of the door.
With the head TILTED, a dog can locate things along a DIAGONAL plane, instead of merely a horizontal plane. -With the door cracked open, this allows him to approximately determine how far away something which given that the source of the voice is around
five-feet-four above the ground. -the footsteps at ground level are also another clue.
The dog of course does this instinctively, rather than thinking it through consciously. -It's tied to food location and threat location, so -again- grows from survival over many generations.
Bear in mind that while my short 'parlour trick' description before does SUM UP my overall feeling of where this usually goes and what the practical applications are, I am actually a little more 'nuanced' in what I really think... It's hard to be accurate without getting VERY lengthy on the subject (as shown by this post!) though. -I'm still being fairly 'short' in my replies, so I'm still simplifying somewhat, though trying to be a little more thorough.
I've been interested in this since the earlier 1980's when I had some inspiration and tried to build a box to make FX returns etc. more 'spatial' without the usual drawbacks... I looked into it, I built a few prototypes, nothing really worked for ALL purposes, and I learned why. Since then, I've revisited the subject while working with location recording, while studying, designing and building B-format microphones, Ambisonic encoders and decoders (B-format and UHJ) and some other applications.
The one useful thing which I have been able to come up with -and which a few dubbing mixers now use- is a stereo-to-surround 'unwrap' technique, which gives something of the 'illusion' of surround, but is fully LT/RT compatible, so doesn't upset Dolby pro-logic... but apart from that, my investigations have led me to the view that there are -and always will be- drawbacks to any 2-channel-source 'surround' processing techniques, which render them useless in terms of being universal. -We make records too many different ways for anything to work universally.
The video has so many tell-tale signs which appear to support my notion that this approach cannot work universally either... the 'locked' listener head position (I have access to THREE of those Neumann heads here, and a fourth somewhat inferior head which I made myself for 'rough-testing) and the near-anechoic requirement for the listening environment being two big ones.
-But I'm not an 'audiophile', so what do I know!
fascinated by the subject though.