Do note that the IEF targets were extrapolated against data from GRAS KEMAR, the pinnae of which causes a dip around those regions (notably for the HD6X0 and the HD800/S).
Noted, agreed and that is exactly my point.
The inaccuracy at these frequencies in HATS differs between them.
It also has to do with driver position and size IMO. Of course I don't have the vast experience of mad economist who has done tons of research on this while I am just a simple engineer.
I see measurements of headphones in a different light. All the research mad economist presented all point to rather substantial differences between HRTF's of human subjects.
While this is fascinating from an engineering p.o.v.in order to determine sort of an average so that a standard can be developped in order to get 'comparable' results. Of course this all is very useful and having a standard is paramount for this.
Here is the thing that keeps mulling in my head.
While research shows that there are substantial differences in HRTF which people often use to say 'we all hear differently' I don't really think that is the case.
I believe that the human brain calibrates itself continuously. This is a slow process. If one gets sudden hearing loss we hear it as such. When it is gradual (due to aging) we don't hear it as such. I still 'think' my hearing hasn't gotten much worse in the last 20 years. When looking at actual measurements of hearing I can see it is deteriorated. I mean if the difference between 30 years ago and now was 'sudden' I would be really alarmed.
When we hear acoustical instruments/music our HRTF is taken into account. We all hear those instruments as real aided by our eyes.
When measured at the eardrum all folks will have substantially different 'input' into our brains but we all hear the music as 'reference' to reality.
Now we measure at an 'average through research' pinna and above all ear canal/coupler. Those are bound to deviate from what people hear and also, depending on the research done to determine the 'average' pinna/ear canal they use to get DF and FF (which are standards) correct.
The differences between these 'differently constructed average simulations' is the cause of the errors when measured and above all when corrected. When one is to EQ on specific measurements (lets use your and Amir's) then the result will be that one HD650 will now have much less treble and the other one much more treble.
When one doesn't act on those measurements or 'averages' all known measurements (and it is the most measured headphone I know) then you may come to the conclusion you should not EQ on it.
I also don't think you need to EQ on it.
The best way IMHO is to listen to headphones first. EQ it compared to hearing (very difficult unless one is really experienced with this) and then measure to see if it works. It's the way I do this. Listen first and then measure to see if it correlates to my 'basic' EQ.
I am pretty sure, given the variance between human ears that there will be folks for whom an attenuation or boost at higher frequencies will actually be beneficial when their HRTF happens to be close to that of the measurement rig.
To me this is the best 'explanation' as to why some people say they don't prefer the EQ profile generated by this or that person and do with another one.
In the end we (well I do) measure headphones to EQ.
We measure/compare HATS to see how and where they differ.
Given the results and comparisons between 'compensated' measurement results it is painfully obvious there are substantial differences between HATS above 5kHz or so. To compound the matter the differences differ when different headphones are used.
To me, the real question is which of the 'standardized' methods is closest in reality with all headphones and with the vast majority of people's actual HRTF.
We will never get there is my opion but really welcome the ongoing research and improvements in this area.
To me it is an unsolvable engineering problem but applaud all efforts to get closer.
And yes, I am fully aware that my test rig and correction is flawed to the bone and does not comply to any standards.