I just uploaded a series of test-scans on my Epson V700 of a 4x5 shot of a rail-road bridge (the same shot that Frank Petronio graciously scanned for me on his 4990).

Maybe it's just me (maybe I'm easily impressed), but I'm pretty impressed. Quoting from what I wrote on the tabblo:

These are scans of a very small region of the rail-road bridge shot from a 4x5 piece of film that I posted in my previous Tabblo.

Each crop represents 0.5 x 0.5 inches of what would be a 50 x 40 inch print. The numbers on them indicate the dpi-resolutions at which they were scanned in on my Epson V700. I did not make any height / focusing adjustments of the film on the flatbed, nor did I do any image-sharpening. All were upscaled to the same size as the 9600 dpi image for this comparison (so each image is 480 x 480 pixels, which corresponds to 0.5 x 0.5 inches at 960 dpi).

I see clear distinct improvement going from 1200 to 2400 dpi, and some improvement going from 2400 to 3200. From 3200 to 6400, there are minor improvements. I am not sure I see any improvement at 9600 dpi (despite claims Epson makes about "micro-step" features). It seems possibly even a little bit less sharp at 9600 than at 6400.

It seems like all of these would make perfectly acceptable 50x40 inch prints, imo (my monitor is 128 dpi, so they measure about 4 inches on my monitor, 8 times what their print-size would be in the 50x40 inch print). The 1200 dpi scan looks slightly less sharp on my monitor when down-sized to be 0.5 inches. I can only clearly see differences between 1200 and the higher scan-resolutions when I resize the images on my screen to be 1 x 1 inches (which equates to a 100 x 80 inch print). Although of course, actual prints are at 300 dpi or more, so the differences would be more clear in a print.

None-the-less, I conclude that it is very difficult to see the differences between a 2400 dpi and even 6400 dpi, even looking at these images at a 1:1 pixel ratio on my monitor.