Despite an almost identical appearance to its predecessor, the X100, the Fujifilm X100S offers enough enhancements to deliver a significantly more fluid shooting experience and notably improved photo quality. It still suffers from a few annoying issues that I don't expect from a camera in its price range, including autofocus inconsistencies, middling lens performance at f2 and an irritating control dial. But overall, I like the camera quite a bit and find it more recommendable than the X100.
One note on the rating: if it seems low -- the same as the X100 despite all the improvements -- that's because it now has competition that it didn't have two years ago when the X100 debuted. That includes the vastly more expensive but full-frame Sony Cyber-shot DSC-RX1, but more importantly a direct challenger in the Nikon Coolpix A. If necessary I will adjust the rating on the X100S in mid-April after I've gotten a chance to test the Nikon.
Two years ago, I considered the images delivered by the X100 as class leading with some caveats; I still think they're excellent -- and visibly better than those of the X100 -- but with more qualifications. The new version of the X-Trans sensor and updated image processing really shines on JPEGs. Thanks to a lack of color noise even as high as ISO 1600, they're clean up through ISO 800 and very good through ISO 6400; while the higher-sensitivity JPEGs don't stand up to pixel peeping, they do stand up to 13x19 prints. I wouldn't venture into expanded high ISO territory except at small sizes, though; they're soft and grainy, and the camera doesn't support raw for ISO 12800 and ISO 25600 (or ISO 100, for that matter), which decreases the tweakability quotient.
One of the big qualifiers I have about photo quality stems from the lens. When I reviewed the X100, I blamed myself for the poor results I got at f2, but I see it again here: f2 is close to unusable, bad enough that if I hadn't seen the same problems after looking back at my X100 shots I would have thought I had a defective unit. It's not too bad beyond about 10 feet, but for anything closer it's fuzzy in the center with what looks like convergence issues and extremely distorted at the corners of the image. At f2.8 it almost looks like a different lens, snapping into sharp focus in the center; you have to bump up to f4 before the edges are all relatively sharp. It does manage to maintain both center and corner sharpness through the top of the aperture range, f16. That said, the camera didn't display any of the sticky aperture blade issues that seemed to plague the X100 (though I didn't experience them). The lens also seems a bit more susceptible to flare than I'd like.
|Click to download||ISO 200
The camera produces extremely pleasing colors, though getting accurate colors is a bit tricky, since there's no really neutral color simulation and the defaults seem to push the contrast to the point where you lose a lot of shadow detail (though it's all there in the raw file).
Video looks improved over the X100 as well, mostly because it's been increased to 1080p, and it's extremely sharp. But there's tons of edge artifacts -- crawling, aliasing, jitter -- thanks to the downplayed filtering. Low-light video actually looks better than daylight, I suspect because it drops the shutter speed, which decreases the appearance of some artifacts (you can't control the shutter speed during video on the X100S, just the aperture). It crushes blacks in low light, but it has an appealing tonality nonetheless.
Though the X100S won't break any speed records, it's a lot faster than the sluggish X100, with much improved autofocus performance. Fujifilm incorporated phase-detection autofocus for a hybrid phase/contrast AF system. That said, the AF has consistency problems: even when there's no change in subject or position (on a tripod facing our test scene, for instance) it will frequently fail to achieve a focus lock after a successful lock. I suspect that accounts for the large variation in results for autofocus-related tests, including shot lag and shot-to-shot times. In cases with a significant variation among the values, I averaged the fastest three results.
Time to power on, focus, and shoot runs about 1.5 seconds; time to focus and shoot in bright conditions is about 0.6 second, rising to 0.7 second in dim light. Two sequential JPEG or raw shots averages about 1 second (though it rose as high as 1.4 seconds), increasing to 1.5 seconds with flash. The continuous shooting rate zips along at about 5.5fps as long as you stay under the buffer limit -- it slows at 8 raw frames but didn't slow as high as 36 frames JPEG with a 95MB/sec SD card. However, continuous-shooting mode only works by fixing exposure and autofocus on the first frame, which doesn't meet our performance-testing criteria (and therefore isn't in the chart).