Doesn't work with either OM-D -- compatibility list on their website says something to the effect of the structure of the camera blocks the radio signal. Doesn't the EyFi broadcast OK from an OMD?
I'll question the blanket low light superiority of the DSLR. I have a Lumia 920, and the night/low light shots it takes is amazing - often as good as my E-M5 with a 2.8 lens. Of course, as they point out, there's no zoom, no real flash, etc. But it is with me ALL the time. I have to deliberately bring the E-M5.
I can't talk about all four, but I can tell you why I chose and am very happy with the Nokia 920 -- the stabilization.
LIke many of us, as we get older, our hands shake more. WIth my iPhone 4s (which I've read as having equal or perhaps a better output than the 5), I was having more and more issues with motion blur. Pictures as high as 1/60 were blurred and everything under about 1/30 was unusable.
After reading initial reviews last fall, I got a 920 and am VERY pleased with the quality of the photos. I generally prefer no-flash available light photography, and the 920 fits the bill. Existing light down to 1/8, blur-free, and with my shakes, that's impressive.
Issues? Colors can be fooled sometimes, and for contrasty scenes, placement of the focus point is really important. Flash can be iffy from an exposure aspect. Lots and I mean LOTS of accidental pictures due to the screen going all the way to the edge.
But for ME, it is without a doubt the best phone camera.
I had a Chinese Seagull folding 2-1/4 rangefinder camera for many, many years. It took some of the best pictures I've ever taken, so I don't have any doubt they can be very competitive.
Looking around on the various camera blogs on the web, it seems all the talk about the X-pro1 is comparing it to the Leica M9. But, I noticed in the image comparisons pages of this review that the image reviews defaulted to other cameras. I selected the M9 in the image comparisons, and I'll have to say that throughout the ISO range, the X-Pro1 images look better in almost all cases, in both jpg and raw.
I don't have either one, so Im not biased one way or the other -- just thought it was interesting that the M9 image comparison wasn't a default choice...