What I meant was that in reality, few musicians are going to release dummy head / binaural content that is mastered such that it only sounds correct if you're listening on headphones. That is: theoretical custom mix for headphones >= Dolby Atmos > traditional compromise stereo mix. However ~all content released is in the latter category. Pretty much all music released is designed to be listened to on the car radio, so all the details are compressed out. In that context, a headphone-specific mix seems to be asking too much.
It is, of course, completely technically possible to produce spatial audio for headphones, but there's a reason why it didn't really take off anywhere other than video games. DirectSound 3D, SB X-Fi and A3D back in the 90s had object-based spatial audio, though it wasn't great, and was an API not a storage or transmission format. Ambisonics has existed since the 70s, and when I tried it out a decade or so ago, it was very cool - but for various reasons, it never really was practical outside of a few tech demos.
Dolby Atmos isn't doing something that was physically impossible before, it's just a new codec that makes it practical to store and transmit spatial audio that can be rendered for the specific speaker configuration it's being played on. Unlike some previous attempts at it, it's a codec that actually has relatively widespread support now, with the force of a Dolby behind it to encourage manufacturers and producers to actually implement it in the correct way. Indeed, going back to my point about car radio dynamics, Dolby Atmos even has strict rules about the loudness of content, meaning if something has Atmos, you're probably going to get better dynamics than the standard stereo mix. My hope is that pushing audio rendering decisions to the decoder is going to force them to get a bit smarter than just dumping channels to the speakers. It's long overdue for car audio systems to have dynamic range compression built in.
I see it a bit like NVIDIA G-Sync. VRR existed for a long time, but with nobody using it. NVIDIA brought it into the mainstream and applied technical standards to it to the point that you knew if you buy a G-Sync branded monitor and GPU, you're going to get a consistent experience.
You really think if Eno released "Ambient 5: Music for Headphones" it wouldn't likely be his biggest seller? "...few musicians..." Eno for sure is one, and it could even be just a different mix and can be right beside the icon for the file with the regular mix on a streaming service. I guess these albums are just that, right? Only not as good.
I'm saying not much at all about Atmos other than Eno is uncharacteristically talking bunk in saying this is what he needs for 3D.
It is, of course, completely technically possible to produce spatial audio for headphones, but there's a reason why it didn't really take off anywhere other than video games. DirectSound 3D, SB X-Fi and A3D back in the 90s had object-based spatial audio, though it wasn't great, and was an API not a storage or transmission format. Ambisonics has existed since the 70s, and when I tried it out a decade or so ago, it was very cool - but for various reasons, it never really was practical outside of a few tech demos.
Dolby Atmos isn't doing something that was physically impossible before, it's just a new codec that makes it practical to store and transmit spatial audio that can be rendered for the specific speaker configuration it's being played on. Unlike some previous attempts at it, it's a codec that actually has relatively widespread support now, with the force of a Dolby behind it to encourage manufacturers and producers to actually implement it in the correct way. Indeed, going back to my point about car radio dynamics, Dolby Atmos even has strict rules about the loudness of content, meaning if something has Atmos, you're probably going to get better dynamics than the standard stereo mix. My hope is that pushing audio rendering decisions to the decoder is going to force them to get a bit smarter than just dumping channels to the speakers. It's long overdue for car audio systems to have dynamic range compression built in.
I see it a bit like NVIDIA G-Sync. VRR existed for a long time, but with nobody using it. NVIDIA brought it into the mainstream and applied technical standards to it to the point that you knew if you buy a G-Sync branded monitor and GPU, you're going to get a consistent experience.