Summary: I'd not buy the 4900 again. There appears to be no recipe (that I can find) that allows me to focus on printing - rather than problem solving with the 4900. As others have said, the output is very fine when coupled with good processing habits and accurate custom profiles. Head clogs and paper handling issues are, on the other hand, numerous and show stoppers - even in our shop where the 4900 rarely sits longer than 24 hours without printing in a 40-50% humid environment.
Summary: I bought this printer a couple of years ago when they were deeply discounted following the Japanese tsunami. I am not a professional but can afford to indulge my passion for photography by working hard at my day job. Given the deep discount I splurged on this machine. It is large requires two people to move it, uses expensive ink $100 a color, but produces beautiful images. Right out of the box I had some issue that I finally called Epson about.
Summary: When it works, it's a great printer. Unlike the 7600 which has been a workhorse for me for years, the 4900 is temperamental at best. Ink head clogs plague this printer. I cannot recommend the printer for someone who plans to use it on an occasional basis. There is a long weblog discussion about head clogs in the x900 series on luminous landscape. Read that carefully before purchasing one of these printers.
Summary: It ought to be simple, with a multi-thousand dollar printer, to print an 8x10 print, no bleed, on 10in roll paper. Right? Not with this printer. It doesn't do full bleed, top or bottom, period. For that, save yourself some money and buy a cheaper printer. And if you try to print an 8 x 10 - heaven help you!
Summary: Update: 11/15/2013 - NEVER BUY EPSON PRODUCTS IN THE FUTURE. September 2011: First, the good news. The Epson 4900 Stylus Pro is an awesome printer. It is built like a tank. The images are truly remakable, as it is designed as a photographer's dream come true. The colors are crisp, the high resolution is quite remarkable. I am impressed.
Summary: Wirtschaftlicher Totalschaden nach 2 Jahren bei knapp 5.000 gedruckten Seiten. Eine Reparatur der defekten Druckköpfe wäre ähnlich teuer wie ein Neugerät ohne Spectrometer. Das ist natürlich inakzeptabel.