[castleofargh]

I'm not as optimistic as you are about the realiser, but then again I'm also not as informed and would love to be wrong. the part going from my speakers to my headphone is measured by the same mic into my own ears so there is little that cannot be done aside from bodily felt bass and the headphone's own limitations when it comes to fidelity and being EQed. if I calibrate without moving the couplers at all into my ears between both measurements, that's just extra accuracy right there. from that I expect great things and the cues we feel with the speakers, should for the most part be felt on the headphone. but that's where the technology stops IMO. we're mimicking the speakers in the room with all their specificities and flaws. that's our reference. so the logical question is how good are the speakers and the room are reconstructing a space?  if they do good the realiser should do good.
I expect everything else to be based on standards themselves based on averages and we might be back to getting different results for different people and different rooms. given the price of the system I doubt that the microphones are perfectly calibrated high fidelity couplers, and just little changes in the way we insert them could alter the sound measured in an audible way. so I'm not sure we ever create a reference that can be used to convert a standard into our sound. our speakers aren't a reference of any standard and that's what we record.
if there is a reference room for ambisonic, going there and making a calibration with the realiser that would include vertical positions of the speakers is probably the closest one can expect to get of that tech. 
 
of course here I'm talking about really high level of accuracy compared to a desired reference in space, as you seem to really be into that. for typical movie watching and gaming, there are probably ways to make impressive stuff with what we already have and a few standards added to the realiser.
but again, that's just me guessing. maybe they do reference the microphones from the start and so they generated that one specific reference to adapt to other standards even before they sell the realiser to us? I just don't know if that's in the original purpose of the device.

 

[jgazal]

Actually I'm really curious, but not exactly optimistic. I just want to know which one has less errors.

After all the info I got from you here and in the accuracy thread, I now believe both binaural and Ambisonics are different cans of worms.

Dr. Stephen Smyth did not want his product advertised as a virtual reality device.

Now I understand why he was that cautious, because image and sound positions may not be rendered coincidentally.

If I record content with the in ear microphones in my own ear, I expect such chain to be the one with lower error.

In theory and without psychoacoustics decoding, Ambisonics need a lot of virtual speakers to have good spatial rendering of mid or high frequencies.

With a device that allow measuring BRIRs with one real speaker positioned in several coordinates and a believable interpolation algorithm (and the reviews indicate that Smyth has both), I would just drop binaural or ambisonics content to use object based content.

I was underestimating the usefulness of content with that kind of codec.

If our perception is more accurate in the horizontal plane and the perception of elevation can be represented not only by subsidiary ITD (lower end) and ILD (mid and high frequencies) cues fired by the horizontal channels, but mainly by personalized spectral cues, it makes sense to edit a horizontal bed and have objects varying in different elevations.

If you think about it, this is exactly what Dr. Choueiri mixer does, but he sells it as a software to mix binaural content instead of a personalized* binaural player of several objects.

What would happen if you get the output of an Ambisonics decoder and input those channels into the bacch-dsp mixer for headphones with each channel position assigned to the coordinates of the ambisonics speaker array?

I guess one limitation with current surround object based codecs is that you don’t have personalized BRIR measurements and thus personalized spectral cues, so you need to rely on only one elevation (the overhead speakers angle).

What would happen if you get the output of an Atmos decoder and input those channels into the bacch-dsp mixer for headphones with each channel position assigned to the coordinates of the horizontal bed and then you ignore the position metadata of objects and use the mixer to fly them around at your will or changing its proximity by altering the direct field versus reverberation ratio?

Anyway, Google VR development tools are based on Ambisonics. And Dr. Choueiri still advocates binaural. IMHO, that is more than enough to be curious about both methods.

Interesting times.

*to be sincere I don't know if bacch-dsp has BRIR measurements as refined as the Realiser. Maybe his player mixer uses a database with different HRTF and the personalization is only applied by filtering the binaural output with the dipole BRIR measurement. That reminds me deep neural networks could be trained to find In HRTF databases one that matches the user HRTF just by comparing his her dipole BRIR...

 

[jgazal]

Just for future reference:

 

[jgazal]

Keynote Speaker
This year’s Keynote Speaker is Prof. Edgar Choueiri of Princeton University. The title of his address is, "Fooled by Audio." How far are we from having reproduced or synthesized sound that is truly indistinguishable from reality? Is this laudable goal still the receding mirage it has been since the birth of audio, or are we on the cusp of a technical revolution - the VR/AR audio revolution? I will report on recent advances in virtual and augmented reality audio research from around the world, and focus on critical areas in spatial audio, synthesized acoustics, and sound field navigation in which recent breakthroughs are bringing us quicker and closer to being truly fooled by audio.
http://www.aes.org/events/143/specialevents/?ID=5522

If anybody goes, I would like to hear impressions.