Eamon said this morning that "the E-P1's autofocus has taken a lot of flak," which is true. I distinctly get the sense, though, that "web wisdom" has exaggerated the E-P1's alleged slowness of focusing. It's not superfast, but once I tried it for myself I thought, "Oh, well, that's not so bad." I was pleasantly surprised because, after all the teeth-gnashing and hair-tearing and garment-rending over this issue, I had expected so much worse. I don't think I'd find it annoying in the long term. Maybe once in a while.
Since I have both cameras and both lenses (i.e., the 17mm and the 20mm) here at the moment, last night I ran a bunch of seat-of-the-pants trials with both lenses on both cameras. My sense is that the Panasonic 20mm might be just a little faster focusing on the E-P1 than the M.Zuiko is; it's tough to tell without the means to measure it. It's not a dramatic difference, though.
The difference is a little more pronounced with the two lenses on the GF1, where the 20mm focuses just barely noticeably faster than the 17mm. Both are still in the same ballpark on that camera, though.
The most distinct difference is between the two cameras themselves. While the focus on the Olympus just plain isn't as bad as it's painted to be out on the pixel-peeping web, the GF1 is unmistakably better. It's faster and locks focus more positively. (The E-P1 seems to have as a default a little in-and-out move that it goes through on every focusing attempt, with either lens. It does it briskly enough, but it uses up a little time that the GF1 doesn't seem to need.)
What really impresses me about the GF1, though—and might be getting overlooked in the general tendency to compare these two cameras to each other—is how remarkable the GF1 is at focusing a) in very low light and b) on featureless surfaces. With the AF area on the smallest setting, the GF1 will lock focus immediately on a featureless white wall:
This is the white wall immediately above my viewing board, at night, with the only light in the room being my 27-watt Verilux fluorescent, aimed at the corkboard below the framed picture seen at the bottom of this shot. With its AF area on the smallest setting, the GF1 repeatedly locks focus immediately on the plain white wall.
This is the texture of the drywall it's grabbing focus on. I can't make the E-P1 achieve focus at all in this situation.
With the same light as the only light in the mostly darkened room (I compared this JPEG with the actual scene to match its apparent brightness level, although bear in mind my night vision is not as good as a 20-year-old's), the GF1 repeatedly and immediately grabs focus on the darkened bookcase, just past the much brighter computer monitor—without using its focus-assist lamp (it pretty much has to be pitch dark before the AF-assist lamp will kick on). The E-P1 will focus here too, just not quite as quickly. (Metered exposure for the bookshelf area is 1 sec. @ ƒ/2, ISO 100.)
In fact I had a hard time finding situations in which the Sony A850 DSLR will focus where the GF1 won't. I did find a couple, but it wasn't easy.
The autofocusing ability of the GF1 is pretty remarkable, overall...compared to anything, not just compared to the E-P1.
P.S. You might be amused by the maxim I've had taped to my computer for the past couple of years...
ADDENDUM: With apologies for the slight off-kilterness of the setup (it's surprisingly difficult to get exactly square to a rectangular subject just by eye), here's a quick test of distortion correction with the two lenses, each of the top two shots using the E-P1:
As you can see, if the camera's JPEG engine is applying distortion correction to either of the lenses, it's applying it to both, although neither are quite perfect. The GF1, similarly, looks nearly the same with both lenses.
Best distortion correction of the four combinations, as far as I could see, is from the GF1 with the 20mm. —Although you'll note that I couldn't quite hold the camera steady in this shot! I don't know if you can see that in the small JPEG, but there's a little motion blur here. I could've tried again, but even though that has nothing to do with distortion, I figured it shows the main advantage of the E-P1 over the GF1 quite handily, so I should leave it alone.... —MJ