Thank you for your comprehensive answer!
To get back to your question:
Practically for me it's important what happens when I listen to music while I'm sitting relatively still.
I'm not circling around all the time trying to pinpoint virtual speaker's positions.
I determin the virtual speaker positions by automatically switching a mono sound, step-by-step, through all 7.1 channel, while looking straight ahead, and I get this:
I tried the other way, turning my head to each virtual speaker, too.
But those virtual speakers do not stay in place when I turn my head.
In fact they move to about the intended positions of a standard 7.1 setup!
I take this as an evidence for the headtracking not giving a correct angular response, so I'd considered those results as worthless for determining the speaker positions.
This means, we completely agree with our opinion:
The 3-D engine is not placing the speakers on correct positions, partly contributing to your mentioned "too much shadow effect".
Now the wrong working headtracking seems to fix that, but not in normal listening position, only when you turn your head to "face" one virtual speaker, while moving other virtual speakers to even more wrong positions (LeftFront moves into the back when you turn right).
Changing the virtual speaker positions was my first request to Audeze, but I got a strong and distinctive "no, impossible".
At that time I thought, moving the virtual front speakers more to the center would fix the headtracking too.
Now as changes in the 3-D engine does not seem to be an option, working with the coupling between headtracker and 3-D engine is the only thing that's left.
Just one thing I want to make clear, which I didn't in my primary postings:
Scaling the headtracking angular response needs to be nonlinear, angular dependent:
those angles that already work correctly are not changed, mainly the front area's about 110° angular response would be different.
The whole HRTF 3-D processing would not be changed at all (it kind of works).
Just the angular mapping from the physical headtracker unit to the 3-D engine's simulted angle need to be changed to the perceived correct values.
It might even be that Waves company finds out that they have an error inside their own mapping and correct it on their side, but my proposal could maybe done by Audeze themselves.
The physical headtracker unit used in Mobius might not be fitting perfectly to Waves NX software, as Waves might have adapted their own hardware unit only and not the one from Audeze.
@KMann
These all are questions that only an Audeze / Waves interaction can solve.
I just can state, the angular response of Mobius 3-D headtracking effect is not correct.
I would like to ask for a fix, as it largely reduces the fun I have with Mobius.
I'm very happy that I am not the only one hearing it this way, thank you GalaxyMaster!