Olive-Welti curve and summary of headphone research/presentation can be found here. Response of the adjustments made from flat response at the listening position. I wonder why O-W didn't use a mannequin to determine the response at eardrum-equivalent for the in-room speaker.

O-W rig with headphone use would be generally similar to AQ, but O-W used large hard pinnae in testing, whereas AQ lists three different sets of pinnae used in measurements (hard small, hard and soft large).

Yes, this is the presentation file for O-W target, and it is also a great example why trying to make consensus with subjective listening tests, even in ABX, is not really a good idea because of bias caused by preference

If you have noticed, first test with 11 listeners.... 8 of them are Harman employees, who are already very used to Harman's house sound. Of course, it is inevitable that they would prefer the sound that is similar to what they used to hear at their work.

This is how a company can claim "With ABX tests, our speakers come top against competitors! Buy ours!!!" While this can be true, but since the company obviously used its own employees to do ABX, the result is not really valid for general public.

Sure, that test with 219 people are probably more valid, but there are several problems too (Kinda strange there is no result posted for HD518 despite the fact that you can even see it from the pictures they take, huh. Make you think about the result indeed.....)

My potential issue with Olive-Welti: it doesn't explore levels in the ear-gain region; it starts with a full ear-gain measurement of an in-room system measured by their GRAS setup (plus some voodoo), which won in a shootout of just 6 response curves. Another paper then explores bass/treble shelf adjustments to this curve. The Olive-Welti curve may be OK compared to the simulations of other headphone responses (using dozens of biquads filters), but it is not the result of a particularly exhaustive search. [OJneg may be right: the O-W curve could drop with a thud in headphonedom.]

And as regards the Nighthawk "white paper": would be funny if they developed a headphone that had all the technicalities right, but had a wrong-headed balance/response-target that drowned it all out.

And as regards the Nighthawk "white paper": would be funny if they developed a headphone that had all the technicalities right, but had a wrong-headed balance/response-target that drowned it all out.

Nothing some EQ couldn't fix in that case. 

neither free field nor diffuse field 'corrections' give an accurate compensation curve for headphones.
Both curves compensate for the ear canal + Pinna + room effects but differently because the stimulus is different.
With a headphone the sounds come from the SIDE only and not via a room.
Also the Pinna works differently when sounds come from the sides than when listening to something in front.
So... the only compensation needed for HP's is ear canal + infinite baffle + Pinna but NOT like when coming from the front.
This means that when you 'undo' the room correction (that is enclosed in DF and FF) you end up closer to what a HP 'emits' into your ear canals directly.

Closest way (they figured) was to undo the room by adding compensation for a small room (in which diffuse field is measured) to the DF which sort of 'corrects' the diffuse field compensated signal to become 'flatter' again so they 'undo' some of the room 'acoustics' in essence.
They use this curve to compensate the too high treble response which is caused by reflections:



This results in a 'nearer flat' response from 50Hz to 20khz and a too low response (due to the small room not able to do low frequencies) for bass response so they add some extra bass again.
In the earlier plots they showed the rising bass response as a target rather than apply it as compensation to the plot and displaying the result as a 'flat line' where the flat line represents equal perceived (but not measured) FR response.

At least that's what I can deduct from their limited amount of info on this subject.

And then even the X-curve (for theaters = large rooms) is heavily debated over the years... here is some thoughts about it:
http://www.hps4000.com/pages/general/the_mythical_x_curve.pdf

That roughly equates to a flat response on a flatbed measurement rig with compensation for bass.
There is a little snag though when using capsules that measure flat to 20Hz in free air.
When these are mounted on an infinite baffle (closed headphone) the bass response also will be elevated in the mic output signal but somewhat differently then just some added subbass.

But this is about how accurate the AQ response is and compared to what 'known reference' and if that target curve is more correct than others.
Then there is the question of how CLOSE the actual headphone is to that curve.
As I saw deviations of over 4dB in various places they aren't close enough (IMO) to their own curve.

From what I saw (would like to see it measured on the CUNT or UB's rig) there is more midbass and less 'clarity' in the NH compared to their own target which certainly seems to be perceived as darker than what 'ideal' sound should sound like.

disclaimer:
Above is all my opinion and not scientific proof nor does it have to be the real truth but is the way I see it.
I can be completely wrong about all this.
Don't really care about compensation curves for HATS as I don't use it.