what speakers do the best "disappearing act"??

Magneplanars, 3.6s are the ones to go for.

Martin Logans do. So do Thiel 2.3's with proper placement.

My experience is that whether or not a speaker disappears is dependent on room acoustics, location of the speaker in the room and on the remainder of the system. Any one of these facotors may limit the ability of the speaker to disappear. Similarly, these factors will effect the ability of a speaker to image beyond its outside edge. Having said this, some speakers are obviously better in this than others. The best examples that I have heard are small stand mounted two ways, point sources like the Quads and planar drivers like the Magnepans and Martin Logans.

A related issue that is also revealing of the quality of the system as a whole is where the system places noise. Much of my listening is analog. In a really good system the record noise is localized on the front of the speaker and is not a part of the soundfield. In lesser systems the noise meshes with the sound field. A related phenomena is the ability to move around in the room and have the location of instruments remain stable. For many years, I thought that this was solely an artifact of the speakers and their set-up and that it was impossible to have a sharply focused image and not have it collapse as one moved away from the ideal listening position. Now I am beginning to think that this may be an artefact of the phase accuracy of teh system as a whole. Given phase accurate speakers and electronics it is possible to have sharp image focus which is stable with movement away from the listening position. I had previously experienced this with Beveridge speakers many years ago, but not since until recently.

I'll proffer a vote for the Parsifal Encores (but with Alephs!) too....

Many interesting posts, folks, but my first thought relates to the inaccuracy in the pair-matching process. Image flotation specificity is wildly affected by acoustic loading AND left/right variations in amplitude response, no?
ANY carefully-matched pair of good-sounding speakers, all other things being equal, will image well.

Some manufacturers are exceptionally careful about specing matched pairs (Verity certainly excells here; Snell, and even Boston (!) are famous for tight control of driver response tolerances, whereas some are notoriously poor.

When I realized that even good manufacturers (SEAS, Vifa, et al) will only control tweeters and mids to +/- 1-2dB in non-custom lots, I knew that amateur attempts to make cloned pairs would be prohibitively costly.
Although process controls are improving all the time for driver specing and sorting (crossovers, too), I always wonder if the "critically reviewed" pair actually sound like the ones one is apt to get. I think the most dangerous thing to do is to decide on a speaker pair after home-demoing a dealer's pair, and then insisting on a brand new pair! Most of us (between bouts of tinnitus) can hear
1/3 dB shifts in midrange response over only an octave or two, right? xcept for only SOME of the best manufacturers, there's practically NO WAY you can get a second pair to sound that close to another!

I know that arguments re room loading, driver smoothness, diffraction and phase issues are pertinent here, but I don't think the mirror-imaging inaccuracies should be ignored because they're too difficult to think about!
Just a wee hours thought....Ern

Excellent points on quality control, Subaruguru!

Ultimately, I found driver-to-driver inconsistencies to limit what I could accomplish on my own during my years as an amateur speaker builder.

Now that I think about it, the three speakers I mentioned above all excel in this area. The active Meridians used individually-tweaked amplifiers to set the exact levels of each driver, and if memory serves me right, when ordering replacement drivers they would ask for the serial number because they matched the replacements based on their database at the factory.

You mention Snell as a company that pays special attention to driver-to-driver matching, and in retrospect I'd have to credit the Type A's excellent imaging in part to this.

The Sound Labs have a bias control for each panel that allows precise level matching. I have found that sometimes, especially with tube amps, one channel may be a tiny bit louder than the other. By tweaking the bias controls on the Sound Labs I can dial that vocalist right in to dead center, and the result is much more three-dimensional than using a balance control in the signal path would be.

Fcrowder, I remember well the way the Meridian and Snell placed the analog noise at the speakers, instead of back in the sound field. This made it much easier to ignore the ticks and crackles, much as it's easy to ignore minor audience noises at a symphony concert because they are not spatially intermingled with the music.

You also spoke about being able to move around in the room and have the location of the instruments remain stable. This is something a line source inherently does better than a point source, because the decrease in volume with distance is much more gradual with a line source than with a point source. Of course the speaker would need a wide, uniform radiation pattern to give decent soundstaging from well off-axis. Also, a tall line source will have the same tonal balance whether you're sitting or standing, which I think is kinda neat.