Originally Posted by sdrucker
Well, it's only controversial if you consider extraction between front and surround to wides to be a norm
You got a point there. I will explain.
I still believe (until proven/experienced otherwise) that center extraction between mains and surrounds could, as a post-processing option, be 'the holy grail' for creating discrete content
for Wides when not supported by the used format or upmixer (e.g. stereo, legacy 5.1/7.1, DSU, Auro3D, AuroMatic). I do not consider merely copying Lf/Rf content to Wides to be creating discrete content
, and would prefer to call this matrixing
, referring to anything that is limited to arraying or mixing channels (incl. level and delay adjustments).
Also, IMO, this matrixing
L/R main channels to the wides is ONLY needed in cases where the ipsi-lateral side-wall reflections fall short in producing the right amount of envelopment, e.g due to the absence of a reflective surface (no side wall, absorptive treatment, or even Wides themselves mounted right on top of the reflection point) and/or extreme toeing-in of the mains (cross-firing). In that sense, this method could even be regarded as being part of the acoustical treatment strategy.
What I am getting at, is that the one method (creating discrete content
) can work together with the other method (matrixing
Lfr/Rfr). So, there is not necessarily a controversy between my current belief, and your observation that when compared separately the one method (applying Neural:X) works not as good as the other method (mimicking reflections). The question that remains in this example is whether applying both at the same time could produce an even better effect...
I originally came from the perspective that this was some sort of holy grail for upmixing, but after some listening, there's good reasons why Dolby did what it did with DSU (sending nothing to wides) and Auromatic did their approach to wides.
Are you referring to Neural:X up-mixing? Because Neural:X does more to the Wides than just center extraction, while center extraction by itself should only result in a more stable sound field between fronts and surround, and possibly some soundstage broadening as well...
It's just a matter of what degree of mimicking side wall reflections (if any) you want to live with. At first I thought it was odd but listening to content it works.
Yes, it certainly seems to be something that not lends itself to be a simple on-off option, but that needs manual tweaking by an experienced/dedicated listener to get it exactly right, and that also depends on the acoustical treatment of side-walls and the toeing-in of the mains.