I am somewhat skeptical about that for a part of the spectrum where HRTF matters. Below 3kHz or so these larger over-ears tend to have a somewhat stable response when varying the position (front/back/high/low) and react quite linearly when varying pad compression (SPL rises linearly), so there is quite a bit of insensitivity to where exactly the driver is positioned relative to the ear and what happens in the front volume.
Besides, what looks like an angled driver effectively becomes, at least for me, a parallel driver, because my head isn't shaped like a hammer-style ear simulator (it's rather that the parallel drivers are angled slightly backwards when on my head, designing the drivers to be slightly angled merely brings them closer to a parallel plane).
For example, with blocked ear canal mics, the behaviour of the HD560S and HD650 under pad compression / pull (with potential breach of seal), with the blue trace as they naturally sit on my head, compensated to a flat line, and the red traces showing the difference. Up to 3kHz or so the FR of these two headphones (or most fully open HPs that I've tried for that matter) tends to be quite invariant :
View attachment 178395View attachment 178396
These results are quite similar to DIYaudio's :
https://diyaudioheaven.wordpress.com/headphones/earpads/
With these headphones the FR stays quite similar up to 3kHz even when the several variables that are modulated when compressing the pads, well, vary.
You can contrast that behaviour with more esoteric designs such as the Hi-X65, or most ANC headphones, which tend to be quite a lot less linear in that band under pad compression / pull :
View attachment 178397View attachment 178398
In regards to the front/back/up/down positional variation, with the HD560S the largest deviations I'm noticing are above 3kHz :
View attachment 178399
Averages of five individual traces at each position, blue trace as they naturally sit on my head, solid red traces the front / back variation (as far as it's feasible without deforming my pinna), dotted red traces the up/down variation (two headband "clicks" up or down vs. normal on both sides).
The difference below 3kHz is what I'd consider audible given the very large bandwidths involved, but not that significant either. Also, because my head isn't a flat plate, some of the difference could come from seal (the HD560S doesn't seal optimally on my head in general because of its yoke design and pads stiffness) - although that should have only a moderate effect with these headphones, or coupling, so we're not strictly observing the front / back / up / down variation here. Isolating the effect of front / back / up / down variation from pad compression / seal tests is better done on an ear simulator with a flat plate around the pinna (although that also won't be representative of the actual, on-head behaviour then).
The thing though is that HRTF modulation already occurs above a few hundred hertz, so above that frequency and below 3kHz or so I'm not certain that angled drivers have anything to do with whatever people call "soundstage" for these larger, fully open over-ears. They won't magically produce a FR at your drum that will ideally correspond to your own individual HRTF in that band for example (which isn't varying
that much between individuals at these frequencies anyway).
Above that I believe that Rtings tried to propose an hypothesis but I'm not certain that it's very conclusive.
Also, what do we make of people who find that some of their in-ears achieve better "soundstage" than some of their over-ears ? Personally I have little understanding for the notion of "soundstage" with stereo recordings, but I've found various forms of binauralisation processes of object based formats to be more convincing with some IEMs / earbuds than some of my over-ears. In general for me it's just about reaching a certain basal FR and then hoping that the binauralisation process / object based format will perform well for my own HRTF.