Wanted to post my HD800 IEM graph of flatness, but since I'm still tweaking (and IEM measurements aren't super reliable), I will hold off for now. No back damping on mine. One thing that I've found is that the way the pads are mounted is very important: The distance between driver and ear changes the midrange and the stock plastic ring seems to (at least partly) cause the resonances in the treble. One other graph for now though: This is the difference between my coupler measurements and my in-ear measurements. The in ear measurements obviously show a lot more upper mids and treble. HD800 was modded. This graph highlights how the interaction between headphone and ear vary by headphone and shows that headphone measurements should generally be taken with a grain of salt. The HD800 difference curve is over 3db higher at 4kHz (where I put the cursor), while the HD600 is higher around 1kHz. On another note: Look at how flat the curve is in the bass. This essentially means that my coupler has a very realistic seal, unlike most creatology foam couplers or even dummy heads. Yes, HD800 isn't a bass extension king and never will be. The seal is only slightly better on my coupler than on my head (graph below 0). Because these curves show how the headphone interacts with the ear and head, I think that it could be possible to attribute differences in soundstage and imaging to differences in these curves. The inverse of this curve on a coupler would translate to a flat response with in ear measurements, which is my target for now. I've been measuring speakers at my ear under varying conditions and it seems like a 1db/octave rolloff at the listening position is a relatively flat response at the ear canal with the typical bump at 2-6kHz. Tyll's curves also show a similar thing: compare the response below 2kHz with after 9kHz. This also depends on the room and many other factors. The Harman room actually seems relatively reverbant, I get a higher speaker L/R crosstalk with my measurements. As Tyll mentioned in his articles, it's also difficult to decide whether to use both or only one channel for speaker measurements. (One other thing that these measurements showed is that a resonance around 8kHz with my in ear measurements seems to be a normal effect of my ears and may actually be a desirable feature, which is good because I've been going crazy because I couldn't really hear this huge resonance that my in ear measurements showed.) Remember how Marv said that the B&K target tends to work well with the coupler measurements? Essentially this is true, but I think the target is a little different. Looking at the curves above, the target varies by headphone and essentially looks like an Audeze curve, but with only 5db rolloff instead of 10.