[Andolink]

I've had the opportunity for the past 4 months to experiment with the addition of cross-feed to my headphone rig (see my signature).  In my case it's a custom designed 6 level balanced cross-feeder.  I'm extremely pleased with this major enhancement to my listening experience and I find several theoretical issues keep swirling around in my thoughts as I learn more and more about just what cross-feed does to the sounds I'm hearing through my headphone.  I'll address just one of these issues below.
 
First,  just so you know at least one of my biases up front, I'm a dedicated 2-channel audiophile.  I've never particularly enjoyed any surround-sound type home theater set up I've ever auditioned finding the multiple sound sources confusing and unnatural. 
 
I'm also aware of the rather strong negative view toward cross-feed in general among Head-Fi members and by the headphone gear industry as well, as is made obvious by the rarity of cross-feed implementation with either headphone amps or stand alone cross-feed devices such as mine.  (I had to have one custom made only because nothing like it exists in the marketplace that I've been able to find.)
 
So the issue,
 
First, there's no question that cross-feed does indeed narrow the width of the soundstage and this seems to be the primary objection to it's implementation by most listeners.  The more cross-feed, the narrower the soundstage until you reach maximum-- a mono signal. Cross-feed, in my experience, has no noticeable effect on the depth of the soundstage.  
 
The question that arises:  Is having the widest possible soundstage with your headphone gear always desirable as seems to be the opinion of virtually everyone contributing to Head-Fi?
 
I've come to the conclusion that the answer is definitely not. 
 
The first inkling I had that extreme soundstage width might not be desirable, indeed quite unnatural sounding, was when I got a chance to audition an HD-800 rig using the Decware Taboo MK II.  The accuracy and clarity were amazing to be sure but the extreme width of the soundstage was quite disorienting and unnatural to me.  
 
Using cross-feed these last several months has demonstrated to me quite clearly that this kind of presentation, where the sound sources are coming at your brain from so far around the sides of your head is not only contrary to what one hears at a live performance but confusing and disorienting to the brain in a way that is actually fatiguing over time.  I've noticed that adding cross-feed not only sounds much more like an actual live performance but is a much more relaxed presentation without the fatigue with extended listening.
 
No doubt all this is irrelevant to many of the genres of electronic music which deliberately use channel separation effects as a major element.  My listening however is primarily acoustic classical music and, of course, when I do listen to electronic music, I turn off the cross-feeder. 
 
I suspect that the generally negative views I've seen by most Head-Fiers to the idea of cross-feed stems from the saturation of our culture with the implementation of surround-sound in movie theaters and home theater set-ups.  For many, that dizzying effect of the sounds coming at you from all directions at once sounds normal.  It never has, though, to me.

 

[ProtegeManiac]

Depending on the position of the headphone drivers relative to your ear canals as well as the frequency response, it may have a tendency to have a "wide" stage with clumps of sound sources that have gaps between the clumps as opposed to all those sound sources more evenly distributed along that "wide" stage. Cymbals can be very loud and on the far left and right of the stage, with the rest of the drums concentrated in the center - you can visualize that as Reed Richards with stretchy arms hitting the cymbals and snare at the outer edges of the stage while his feet are there in the center along with the bass drum, then he somehow is fast enough to hit the toms still there too. Then, along with this strong L-C-R signal, in some cases when there are drum hits between L-C and C-R they tend to sound farther away, so it's like Reed swung his arm backward to hit some drums behind and to his left and right.
 
In such cases Crossfeed can be beneficial - you lose absolute width but you put instruments in more believable positions relative to each other. Note that when some people say there is a lot of "air" between the instruments, you'd have to consider whether the person who said it is actually hearing a lot of believable, relative space between each sound source, or you have that strong L-C-R signal cluster with barely any sound between L-C and C-R, basically a lot of air between three clumps of sound.

Another tendency of that absolute space psychological issue is how people interpret soundstage in general for headphones. You either have some raving about a huge soundstage (which, again, may be disproportionate to the relative space between each sound) or the counterclaim that there is no such thing as soundstage in headphones so screw it and gimme exciting forward sound that can easily be squeezed from an iPod. Both extremes ignore the more important bit about imaging in audio, which is to easily separate each instrument tonally and spatially, and beyond absolute space/size relative spacing is more important. 

I mean, if absolute width and depth are of absolute importance, then even speakers will fail at that unless you have a room that is the same size as an actual stadium stage for example, or the same size as the jazz club you heard that artist perform in.

 

[ToddTheMetalGod]

Most of the reason I chose the HD800 was it's imaging, rather than most people who chose it for soundstage. The HD800's imaging is very convincing because the images are larger than other headphones and more accurately placed in the space (not for all music, mostly orchestral). It's similar to how people say Magnepan speakers are, being panels they can represent the instruments in their full authority and scale.

I do feel that most Head-Fiers try to hard to get a large soundstage, to me the Q701 soundstage is very unnaturally wide as you described the HD800. I think this obsession is an attempt to make headphones sound like a band playing in front of you or around you, which just won't physically work in a pair of headphones. Speakers are needed to spread the image and give them proper scale (which I feel the HD800 is the best attempt of in a headphone, but I haven't heard the K1000).

I personally dislike Crossfeed because I haven't heard an implementation that didn't alter the frequency response and severely alter the image, collapsing it and taking away the illusion. The software Crossfeeds I've used cause the sound to have too much bass, and while for a while the imaging sounds nice eventually I realize some images disappeared in order to apply the Crossfeed. Detail decreases because parts of the song were made too small to hear in the soundstage, or are covered up by other sounds.