Possible = not a fact, its an assumption.
I observed that in general a messy GD (above 500Hz) is more common with open headphones than closed ones. That and the fact that deviations in GD point towards resonances are the only thing one can derive from such measurements.
I don't think (assumption on my part) that there is a link between GD and 'headstage' or 'imaging'.
Headstage and imaging is a highly personal thing (brain) and one's auditory system is definitely NOT the same as that of industry standard fixtures.
However, what about mild/barely audible ones?
That's what I meant.... what would be the threshold and is that the same for each person and with the same stimulus (music vs tones) ?
Let's call > 10% not 'barely audible'. Also very narrow peaks (especially when in %) may not be audible at all.
The 114dB SPL trace, with real music playing irresponsibly loud, will never be reached, bass levels would be around 130dB by then which cannot be achieved.
So in this case the distortion, even for loud music, is practically non existent as even 104dB at 5kHz is not going to be very audible.
On tonality:
This plot should be smoothed when one wants to say something about tonality. Research has shown that a small dip in the2-3kHz range is actually preferred by some (makes the sound more 'mellow' and less 'edgy' and perhaps even give more 'space' to the recording. Also well extended, but not 'Harman boost', bass is also preferred by some. When one happens to be one of those people and uses the headphone to enjoy music this particular headphone may well sound great... $ 6k great ??? That's in the ears/wallets of the owner(s).
This headphone does not measure nor sound poor nor uncomfortable to wear. It just is very expensive.
i don't think the majority of people here doing that. they immediately conclude, upon seeing high distortion at 104 or 114db, that a headphone is bad.
Oh yes I agree. No one will be listening to mid and treble at 114dB SPL. Amir just measures the whole range to see where problems occur. Which in itself can show possible issues. They may well come from time related issues and not actually be distortion. This depends on the used method.
I too think that many readers draw incorrect conclusions, also based on FR b.t.w.
For those only using headphones as low level background music it is all moot anyway. I get that point. That does not mean one should not test near the extremes. It does supply data, regardless if they are valid to said users.