|Fujifilm S5 Pro Review
Is this the "better" D200?
|Add a comment or send Thom feedback on this article.|
When Fujifilm announced in 2006 that they'd be making their next DSLR using the Nikon D200 body, quite a few serious event shooters started salivating. The D200, after all, is essentially Nikon's small-scale pro body (similar to the F100 in the film world), and as such could fix most of the perceived faults of the S3 Pro's body in almost one engineering design gulp. Since the S3 Pro's strength was its imaging (sensor and JPEG rendering), the marriage of best-of-Nikon and best-of-Fujifilm was heavily anticipated.
For a few days.
Then details started to slowly leak out. Instead of 5 fps, the max speed would be 3 fps, and even slower if you used the extended dynamic range feature. This conjured up visions of the S3 Pro's biggest weakness: slowness during operation. Next, the sensor appeared to be unchanged from the S3 Pro. At 6mp, this seemed like a non-step forward, especially since we've already seen 9mp sensors in the Fujifilm compact cameras as the next step from 6mp SuperCCDs. Those two things--slowness and lack of more resolution--became the discussion points on virtually every Fujifilm DSLR forum, with the usual conclusions tending to be "oh, oh, we're in for more of the same slow-as-molasses" versus "just giving me an S3 Pro in a D200 body is good enough for me."
When the camera finally appeared in March 2007, the discussions had all died down into the two opposing camps, and mostly only the one camp (S3 in a D200 body is good enough) opted to actually try the camera. Here's the thing: neither camp was right. The S5 Pro has some real pluses and some real minuses, and whether it ends up being the right camera for you will have a lot to do how you value those specific things.
This review is a little late (as will be the Complete Guide) because of a number of unplanned trips I had to make due to a family issue that came up. But at the same time that also meant that I was carrying an S5 Pro in my bag and using it for a bit longer than I usually do before I wrote my review. Put another way: I believe I have a strong sense of what is right and wrong about this camera, which I'll try to present in this review.
Since the S5 Pro is so close in feature set to the D200, some of this review will simply assume that you've read my D200 review. We all know the D200 is a fine camera with an impressive set of features, and I'm not going to take the space here to repeat all that I said about those features in the D200 review; what you'll find here are mostly the things that are different about the S5 Pro. Thus, if you haven't already done so, please read my D200 review before proceeding here.
*I'm going to assume the flaw in my camera isn't common.
The D200 was a completely new Nikon body design that seemed almost a perfect hybrid of the D50/D70 and D2 series. It had the size, pop-up flash, and battery heritage from the Nikon D100/D70/D50 progression, but the feature set, button layout, and menus of the Nikon D2 series. The Nikon D200 body construction is more like that of a Nikon D2, including metal framing and gaskets for weather protection, but in your hands it feels much more like the lower-end cameras due to the smaller size and some external construction cues. So the Fujifilm S5 Pro starts with a base specification that looks and feels good. The two things the S5 Pro does not share with the D200 are essentially these: the menu system and the imaging system. (Again, for the rest of the D200 feature set description, see my D200 review.)
The sensor in the S5 Pro is a SuperCCD that appears to have the same specifications as the S3 Pro. While Fujifilm's early marketing materials for the S5 Pro made some noise about it being significantly different than the S3 Pro's sensor, those claims simply seem to have disintegrated over time. For example, the current main statement they make is "The new FinePix S5 Pro continues these advancements by coupling the proven Super CCD SR Pro with the companies [sic] new RP (Real Photo) Processor Pro for even more outstanding film-like results." This implies that the S3 Pro SuperCCD SR Pro sensor is left pretty much intact but the main change is the RP Processor Pro imaging processing ASIC. Based upon what I can see, I'd have to say that's true.
Right up front we have to get to one of the key controversies of the Fujifilm DSLRs: are they 6mp or 12mp cameras? Fujifilm says: "12.3 MegaPixels (S – 6.17MP, R – 6.17MP)." The truth is that the sensor is a 6mp sensor that happens to have two photo diodes at every Bayer position. The "S" photo diode is larger than the "R" photo diode, and it is this difference in size (and thus receptivity to light) that allows the camera to conjure up a larger dynamic range than a traditional sensor. But in terms of real resolution--the thing you usually associate with megapixels--the camera is no different than a 6mp SuperCCD would produce without the extra photo diodes.
That's not to say that it operates like a 6mp camera, though (this is where people get confused). First, one of the unique aspects of the SuperCCD designs Fujifilm uses compared to the Sony and Nikon sensors you find in the other Nikon mount bodies is that the Bayer pattern is angled at 45 degrees. While this doesn't sound like it should produce a difference, it does. Essentially, by angling the Bayer pattern the Fujifilm engineers get a "boost" in horizontal and vertical resolution at the expense of diagonal resolution. Traditional Bayer patterns (as used in all the Nikon DSLRs) get a "boost" in diagonal resolution at the expense of horizontal and vertical resolution. So what, you say? Consider gravity and its impact on how detail is organized in both the natural and man-made world: if you keep the camera perpendicular to the gravitational force (i.e. hold the camera level ;~) you get the maximum resolution ability of the Fujifilm bodies on the axis that has the most detail. This is not a trivial increase. Technically, it works out to as much as a 1.4x difference, so the 3000x2000 native pixels for the Fujifilm 6mp SuperCCD are acting more like a 4200x2800 sensor on the primary axis. The drawback is that diagonal lines are going to be more likely to show stair stepping than on a traditional Bayer sensor of the same number of pixels (which puts its highest resolution capability on the diagonal). I'll have more to say on this in the Performance section, as the theoretical and practical limits don't actually align perfectly.
The second benefit that Fujifilm uses apparently has to do with the demosaic into 12mp. The default JPEG size for the 6mp S5 Pro is 12mp! In other words, the camera interpolates to a larger image size by default (you can turn this off). But as you'll learn, the 12mp image is slightly better at rendering detail than you'd expect from taking a 6mp image and using traditional interpolation routines. I conjecture that Fujifilm has a unique demosaic that takes into account something about the unique spatial relationship of the photo diodes. All the other demosaics I know of that convert Fujifilm raw images essentially rotate the Bayer data 45 degrees and then run a normal demosaic, and I've never seen any of them get quite the same level of detail that Fujifilm seems to achieve (though many come quite close--the operative words in this paragraph are "slightly better," and I do mean slightly, as in many people won't notice a difference).
I wish we could move on from the sensor, but I need to address something that keeps coming up in discussions. Essentially, some people hypothesize that because the S and R photo diodes in the Fujifilm sensor are equally spaced (i.e. there's an R photo diode essentially positioned in the middle of every diamond of four S photo diodes, and vice versa), that there really are 12mp worth of photo diode positions. The conclusion these people then reach is that you should be able to retrieve 12mp worth of resolving power if you just put the two pieces of data together correctly (and no one, including Fujifilm, has done that). The problem is this: microlenses. Each S and R pair of photo diodes apparently lives under a single microlens set. Put another way, each S and R pair gets essentially the same light. Thus, there's really no "extra" resolution to be had.
So let me restate all the sensor stuff simply: the S5 Pro has a 6mp sensor that does produce horizontal and vertical resolution more akin to a higher-specified sensor. I'll speak to how much more than 6mp in the Performance section. But this is not a 12mp camera.
The one thing that Fujifilm does talk about a lot in its marketing materials for the S5 Pro is how much better the noise tendencies are. I have good news and bad news on that front. The good news is that, as I'll report later, the noise levels in the S5 Pro are reasonably well controlled. The bad news is that it comes as a result of the RP Processor Pro imaging ASIC's noise reduction routines. Noise reduction applied downstream from the sensor (in the imaging ASIC or in raw conversion software) always tends to take away detail as it does its job. The Fujifilm S5 Pro has no magic get-out-of-detail-reduction-free card. Yes, you can produce images at high ISO values with no objectionable noise, but it comes at the usual cost.
Outside the sensor, the S5 Pro feature set is essentially the same as the D200, which is to say, state-of-the-art. Here's a fairly complete list of the differences between a D200 and an S5 Pro:
Shocked? Funny thing is the D200 and S5 Pro are very, very close in feature sets. As I noted earlier, the primary changes come in the menu system and the imaging chain--everything else is pretty much a direct one-for-one between the two cameras.
Heck, the S5 Pro even adds intelligent battery life tracking ala the D200 (but, unfortunately, using an identical-looking but different battery, the NP-150; this means that you can't use Fujifilm batteries in a D200 nor Nikon batteries in an S5 Pro, a significant drawback, in my opinion. You can, however, use either charger with either battery, though, meaning you only have to carry one charger with you if you carried both a D200 and an S5 Pro).
Before moving on, I should speak a bit about the difference between the S3 Pro and the S5 Pro. Uh, night and day? The S3 Pro was essentially a true consumer Nikon N80 put into a slightly gawky and larger body. As such, the feature specs and build quality are considerably lower than what you find in the S5 Pro. Essentially, the S3 Pro is to the N80 as the S5 Pro is to the F100.
But there's more. With the S5 Pro Fujifilm has abandoned the unique four-button FUNC panel used on their previous DSLRs. Most of the things you'd want to set that had to do with the image were set through this convenient button-and-supplemental LCD array on the S3 Pro. As such, that made going from one set of imaging characteristics to another quite easy. The S5 Pro abandons this unique design and awkwardly grafts it into a dual-natured menu system (I'll describe this more in the Handling section that follows). The net result of the change is that, if you graduated from an S2 to an S3 you'll now find that the next generation of camera has a complete different personality when it comes to setting things. Fortunately, ISO, Quality, and WB have dedicated buttons on the D200 body, so the S5 Pro simply gloms onto those for part of the FUNC panel replacement.
The gawkiness of the S3 Pro is gone, too. The odd shapes and slightly deep body design of the S3 Pro--the result of taking the guts and top panel of an N80 and pasting on a digital back, vertical grip, and new battery--aren't repeated in the S5 Pro. The S5 Pro looks and feels like a Nikon D200. Same size, same shape, essentially same weight. The only giveaways that anything is different are the Fujifilm labeling on the front and the fact that four of the buttons on the back are labeled differently. Otherwise the S5 Pro looks exactly like a D200. Thus, it also feels like a D200. That's a good thing, because Nikon got the D200 design quite right.
The US$1899 price tag (as I write this) puts the S5 Pro significantly above the D200 in price. Thus, you're paying a ~US$400 surcharge for essentially the difference in sensors (and getting some handling differences thrown in for free ;~). This is the key element you need to consider, I think. Is what is different on the S5 Pro than the D200 worth the extra money? Answer that question and you'll know whether you want one.
A couple of paragraphs back I mentioned that the S5 Pro looks and feels like a D200. That's a little deceptive, and the truth comes out in the handling. Indeed, I've been shooting with a D200 and an S5 Pro side-by-side for several months, and I have to say it's not fun switching back and forth between bodies.
It all boils down to four buttons, basically. On the D200 to the left of the LCD we have: Playback, MENU, Thumbnail, Protect/Help, and Zoom/ENTER. On the S5 Pro that becomes: Playback, MENU/OK, DISP/BACK, SETUP, and Zoom/Face Detect. On the D200 the MENU button leads to a complex but well-organized hierarchy of all the things you can set on the camera. On the S5 Pro the MENU/OK button leads to a hodgepodge that is labeled SHOOTING MENU and tries to perform most of the functions of the now-gone FUNC panel plus the button tries to act as the "complete this action" button.
Why do I say "hodgepodge?" Because the ordering isn't particularly well thought out. I think that Fujifilm thought that since there were dedicated QUALITY, ISO, and WB buttons that they could put those functions on the second scrolling page, yet those are things that people set a lot. Meanwhile, NOISE REDUCTION, which you'll also want to change semi-frequently is the last of 15 functions on the SHOOTING MENU. But the real hodgepodge comes in naming. If you aren't confused by Fujifilm's naming, I'd like to know which planet you originate on so that I can see if I can find the translation dictionary that'll help me understand. For FILM SIMULATIONS we have STD, F1, F1a, F1b, F1c, and F2. Clear as pie. For D-RANGE (which really should be labeled DYNAMIC RANGE) we have 100% (STD) and numbers up to 400% (W2), plus AUTO. Still clear as pie. Our choices for COLOR are HIGH, MEDIUM HIGH, STD, MEDIUM LOW, ORG, and B&W. Get used to that "ORG," which really means "sort of off" in Fujifilm parlance.
When we get to AWB/PRESET WB FINE TUNE (no, I don't make these things up) you're presented with a cryptic R-Cy and B-Ye continuum that you can set. At first you might think this is the traditional CIE Lab Color two-dimensional representation, but it isn't. Well, half of it isn't. The R-Cy (Red-Cyan) is a normal axis of the CIE Lab Color space, but the B-Ye (Blue-Yellow) is a 45 angle to the other normal axis. Put another way, there are places in the two-dimensional CIE color space that you can't shift to with the control, though you won't find any meaningful explanation of what these controls do in the manual (or in the camera--help is not surprisingly missing for this item).
Meanwhile, image size is labeled RECORDING PIXELS (are some not recorded?). Yes, I'm being overtly sarcastic here. The point that should be coming across is that Fujifilm's naming and organization of the menu system is not only unique, but potentially highly confusing. Now, if you come from a previous Fujifilm DSLR, some things might not be so confusing. The "(W2)" after the 400% in the D-RANGE menu, for example, is a pickup from the terminology used on the S3 Pro. But to the Fujifilm newcomer--and the camera is good enough that there should be more Fujifilm newcomers than upgraders--all this will seem cryptic and puzzling. It also is part of what makes it difficult to move back and forth between a D200 and an S5 Pro. Not only are the menu systems quite different, they use very different terminology (and even different capitalization--the Fujifilm menus SHOUT every command at you with their all-caps design).
Another aspect of the menu design some won't like is that the > key of the Direction pad isn't used to select an option as it is in the D200 (and other Nikon body) design. Instead, you're theoretically supposed to press the MENU/OK button, which, of course, dismisses the menu system from the LCD simultaneously to setting your option. If you should want to set more than one thing in the SHOOTING MENU at a time, you'll be pressing the MENU/OK button a lot, slowing you down, or you'll learn (not from the manual) that you can simply press the < key of the Direction pad to enable a setting and return to the next higher menu. Starting to understand why it's difficult to move between a D200 and S5 Pro? And I'm just getting started ;~).
The SET UP menu (accessed by pressing the SETUP button instead of the MENU button--i.e. the Fujifilm menu system is dualistic in nature compared to the holistic Nikon approach) has 11 categories of settings, some of which have multiple pages of menu items to choose from. And people complained about Nikon's Custom Settings menu with it's grouping of six categories. Indeed, many of the S5 Pro's SET UP items are duplicates of Nikon's custom settings, though others are things that appear on other menus in the D200 or are unique to the S5 Pro. Here's just one questionable choice (in my mind): the FORMAT option is in the last category (SYSTEM). And frankly, I don't quite understand how things were chosen for MISC., MAINTENANCE, and SYSTEM. BATTERY INFO doesn't seem like MAINTENANCE to me, for example. And battery things are spread over several menus (the battery type for the MB-D200 accessory is put in MISC, for example). Again, my primary comment is that shooting with a D200 and S5 Pro side-by-side will send you into confusion as you look for where Fujifilm put the feature that's so nicely organized on the D200. Meanwhile, the one thing Fujifilm could have done to help minimize your dropping into the menu system--have banks of settings for quick switches of multiple items, as does the D200--is missing completely.
In short, Fujifilm needed a good user interface designer to come in and simplify, organize, and better name their options. I suppose it means my eBook will sell well, as some of the names are so cryptic and unexplained anywhere in public that any morsel of information will be welcomed by S5 Pro users. Frankly, however, I'd rather be out shooting than reverse engineer what a user interface term means.
If the menu system were the only thing Fujifilm got wrong on the changes to the D200 handling, I'd just spread my sarcasm a little less thickly and move on. Unfortunately, there's more.
Image review is something that you'll come to dislike on the S5 Pro. First, if you have the camera set to automatically show an image after being shot, you'll be flummoxed to discover that you can only review that image. Heaven help you if you shoot a burst of shots and want to go back and review any but the last one. You'll have to cancel the display and hit the Playback button to do that. If instead you opt to not have images show up immediately after shooting, you end up waiting longer to see them, as you have to wait for the card to stop writing first. Neither option will have you happy.
Moreover, Fujifilm apparently didn't read my reviews of previous cameras: the histogram display uses white graph data on a white grid (note that Nikon uses yellow or colored data on a different color grid). This makes seeing what's happening at the edge of the histogram (really, the only important part!) next to impossible to decipher, especially in bright light. True, Fujifilm does now offset the edge data by one pixel (on a 2.5" display), so if you know what to look for you can sometimes see it. But it's decidedly less easy to see and use than any Nikon histogram has ever been. I and other Fujifilm users have been complaining about this for almost five years now.
And I still don't understand the Fujifilm zoom functions. We've got two dials, two buttons, and a Direction pad we really could use (as does Nikon) to simplify the process. Instead we get the same slightly awkward zoom function that was in the S3 Pro.
Yes, I'm harping on handling a lot. Fujifilm has had a full two years to improve so many things in the details of the S5 Pro over the previous model, but it really doesn't seem like much engineering effort or time went into that. Coupled with the non-sensor change, one has to wonder just how much engineering resources Fujifilm is putting into their DSLRs. If it's a lot, it's not showing up in the details that users will encounter every day in using the cameras.
Which brings us to another issue: speed. The menus and review system will slow you down a bit from what you can do setting and reviewing on a D200, but the S5 Pro imposes some other speed penalties you need to be aware of.
The biggest one comes with that D-RANGE setting: get out of the normal setting (STD) and the camera drops further down from 3 fps to 1.5 fps as a maximum operating speed. The buffer sizes go down considerably, too. Should you also decide to record that extra D-RANGE into a raw file along with a companion JPEG, you'll encounter yet another penalty: the huge 30MB size of the data being recorded will start having an impact on speed due to write times. At worst case, it'll take four or five seconds to flush that out to the card and clear space for the next image with reasonably fast cards. Since you can only get 7 or 8 such frames into the buffer before that happens, you can turn that 5 fps D200 body into a 0.2 fps S5 Pro body if you're you're a shutter masher. And more slowness: if you fill the buffer and then need to change to another card because you're getting low on room, you'll be waiting over 30 seconds before you can turn the camera off and remove the card.
The good news, I suppose, is that the S5 Pro has more buffer room and seems a bit faster in writing to fast cards than the S3 Pro was. This is especially true for when you don't expand the dynamic range. But the D200 shooter who moves to an S5 Pro will feel uncomfortably slowed if they're using continuous shooting settings with raw files and expanded dynamic range.
Overall, the S5 Pro is a more awkward-handling camera than the Nikon D200, yet most of its controls are identical! It really boils down to the choices of how to use those four renamed buttons, and Fujifilm's choices seem simplistic and not fully fleshed out. I can understand Fujifilm's wanting to keep terminology the same for S3 Pro users, but that could have been done through an S3-Pro mimic mode (which most S3 Pro users would probably abandon had another better mode been available).
My S5 Pro came from the factory with a significant defect: it can't detect batteries. This turns out to be both a plus and a minus. Because the third pin seems to be disconnected, I can actually use D200 batteries in my S5 Pro! The downside is that I get no battery status and the camera stays active all the time. Rather than send the camera back to the factory to be fixed, I just learned to live with it (I turn the camera off when not shooting with it, plus just carry an extra battery at all times since I can't reliably predict when I'll run out of juice).
Here's the thing: this defect seems almost certainly due to the re-manufacturing process (e.g., grafting the Fujifilm stuff into the D200 body Nikon supplies). I can't say whether or not you're likely to encounter a similar issue, but it does bother me to have gotten a camera that obviously wasn't QA'd well off the remanufacturing line.
Writing to Card
Curiously, expanded dynamic range settings slow down even JPEG writes, even though the eventual file size is the same. For example, a normal dynamic range JPEG shot in a burst writes at a little over 1.25 shots a second while expanded range JPEGs write at about 0.9 shots per second on the same card. Obviously, the RP Processor Pro imaging ASIC is taking up some of that time and slowing things down. With RAF (raw) files, things get dramatically worse: a normal dynamic range RAF writes at about 0.9 shots per second while an expanded one writes at a much slower 0.3 shots per second (all times with a SanDisk Extreme III 4GB card). Of course, the expanded dynamic range RAF is twice the size of the normal one, which explains a lot of the difference.
Essentially, if speed is important, you need to shoot JPEG, if for no other reason than the the file that needs to be written is one-fifth the size of a RAF. Because the S5 Pro's write circuitry is still well behind the state of the art, file size becomes a critical issue for determining overall speed. JPEG has higher buffer sizes, too, with the smallest being 19 (JPEG FINE LARGE). Thus, if you're a JPEG-only shooter and don't mash the shutter release forever (19/1.5 fps = ~13 seconds), you're unlikely to feel constrained by the camera's write speed at all.
I will again repeat that the S5 Pro is better than the S3 Pro. Not only is the write mechanism a bit better, but the buffer seems more efficient, as well. If you keep your RAF bursts to three or four shots and don't push for more than an overall average of a shot every eight seconds or so, the camera won't "stall out" on you and make you miss shots. But the S5 Pro is not a fast action camera, despite being built on a body that was indeed well-suited for fast action.
Tracking flying birds, for instance, works about the same on the S5 Pro as it does on my Nikon D2x unless the tracking mechanism starts to use the outlying sensors instead the central ones. The Nikon D2 series will track motion across a wider range of the frame, and more accurately. But assuming your bird pretty much fills the central area of the frame and you can keep it there, the S5 Pro holds its own against the D2 in bright light. For the wedding and portrait photographer that is most likely to be interested in the S5 Pro, this level of autofocus performance is more than fast enough, and the group AF options can be highly useful.
Like the S3 Pro and the Nikon D50/D70/D100, the S5 Pro works best in low light if you can keep the central AF sensor the active one. The outlying sensors aren't quite as bad as the ones in the earlier cameras, but there is a decided drop-off in performance as you move outwards from center in low light. Again, however, wedding and studio photographers won't be displeased with the performance. I'd also suggest that such photographers take a very close look at the Wide option for the Autofocus system (7 sensors instead of 11). They should also look at using the AF Assist on their flash units instead of the camera body.
Overall, I'll stick by my initial statement: very good, but not excellent. The demotion comes primarily due to the fact that AF on the S5 Pro works better the closer to the center of the frame you're trying to focus. This actually may not be a problem for many photographers, as they either center their subjects or focus-and-reframe. But for the more sophisticated user AF performance degrades as you move outward, and compared to the focus-almost-anywhere aspect of the D2 series, is obvious.
Note that at the camera defaults in AdobeRGB the camera produces slightly less color saturation than "ideal," a slightly unexpected result. Meanwhile, in the F2 film simulation, the yellows and oranges are drifting more towards red and some of the blues a bit more towards cyan, but overall the color is pretty accurate and saturation is near perfect. All the film simulations differ slightly in a few of the color drifts, but there is no dramatic shift as I've seen on previous Fujifilm bodies. Moreover, at ISO 1600 the results are very close those of ISO 100; many bodies--the Canons and older Nikons being notorious for this--start to lose saturation quickly at higher ISO values. Update: a few comments on various Internet forums have asked if I don't see the "red" bias in skin tones that some do. You can actually see that in the above charts if you look. I didn't say the S5 Pro was "accurate," I said it was "pretty accurate." In general, the yellow/orange/amber colors drift towards red in most settings, and that impacts skin tones. You do have the white balance adjustment to potentially improve that, but because Fujifilm hasn't provided the CIE a and b color channels directly in that setting and the blues, greens, and magentas are more accurately captured but would be altered by changing that setting, you can't exactly fix the problem. You either like the Fujifilm colors or your don't. Most people find them pleasing, though potentially a little overly red on skin tones. I will note that the color has shifted slightly from the S3 Pro, which was less accurate, but more "pleasing."
White balance is the usual Fujifilm white balance: you either use AUTO or set custom white balances for everything. AUTO--at least with the latest firmware--seems to work well for outdoor and flash work. Indoors, well, I'm not so sure. The S5 Pro is better than the S3 Pro was in AUTO under incandescent, but my D2xs still is unmatched in getting it right indoors. Every now and then the S5 Pro not only gets indoor white balance wrong, but very wrong; I've got one compact fluorescent bulb that my Nikon and compact cameras don't have a lot of problem with but my S5 Pro does. I suspect some of the issue is mixed lighting situations. I will note that the S5 Pro gets outdoor white balance much better than the S3 Pro did; basically I came to trust the S5 Pro in almost any daylight setting with AUTO. Fortunately you have lots of custom white balance positions you can set (enough for a reasonably complex indoor wedding shoot, I'd say), plus you have the ability to fine tune the white balance with that faux CIE Lab Color control. What you don't want to do is use anything other than a preset Kelvin value, AUTO, or the custom settings. The other labeled white balance settings (e.g. Shade) simply seem to produce wrong results too often.
As I've written before, I'm really getting leery of reporting numbers for noise tests, instead trusting what I see more than a number can tell you. On an absolute numbers standpoint, with NOISE REDUCTION set to STD, the S5 Pro generates only decent values at ISO 1600 and even at ISO 3200 (much less so at ORG). By "decent" I mean that numerically, they're okay but not particularly low in value (the black patch on the ColorChecker chart generated a slightly higher noise level than the D200 at ISO 1600, though the patches above middle gray measured better than the D200). Interestingly, all three channels measured very closely in noise, showing that Fujifilm opted more for color noise reduction than luminance noise reduction. Thus, I find a slight bit of "graininess" in the ISO 1600 images and a slight bit of detail reduction, both of which remind me a lot of a well-controlled high ISO film. The question is whether or not you'll like the image that results, not whether you've got "better numbers" than the guy shooting next to you. Here's one place where what you use the camera for will determine your like or dislike of what Fujifilm has done.
For example, if you're shooting weddings and portraits, a bit of detail reduction actually tends to improve skins (at the expense of hair detail). The extra highlight detail of the dynamic range expansion can pull in detail that would be blasted on on other cameras, so the net impact for these shooters is generally positive (and you can always dial down the noise reduction by setting ORG). This, of course, is Fujifilm's target market for the S5 Pro, so the noise reduction settings don't really hamper you. Okay, you might worry about hair detail, but that's about it.
Now let's take the camera out into the woods and shoot bear. All hair all the time. Oops. Now we will start to object to the noise reduction removing detail, as it starts to make for fuzzy bears, not sharp ones. Of course, we probably aren't shooting at ISO 1600 or 3200 any more, so we're not encountering the worst of the noise reduction handling. I'd say that up to ISO 800 the noise reduction is tolerable for detail shooters, especially at the ORG setting.
Overall chroma noise is held down impressively except at ISO 3200, where it starts to creep in.
Once again I decided to run some worst case images in the basketball arena where I play, which has horrid light conditions. As with my D2x, this is a worst case image: ISO 3200, underexposed a bit (the light is dim and terrible and trying to keep any sort of reasonable shutter speed is impossible), Auto White Balance, no in camera NR turned on.
Sort of a good news, bad news scenario. First, color saturation has stayed very, very high (the orange rim). But you should be able to easily see both the graininess of the noise and the blurring of detail (look at the edges of curves lines). Running post processing noise reduction makes the problem even more visible:
Basically, the expanded dynamic range generated by that extra set of photo diodes is a reliable two stops. Most DSLRs really max out around seven stops of dynamic range, but the Fujifilm easily attains nine. Depending upon how you evaluate noise in shadows, you may find that you can produce images with more than that.
Most interesting is that the expanded dynamic range all happens above middle gray, which is exactly where most DSLRs have their least capability (I've written before that most DSLRs are asymmetrical in their ability to capture values below and above middle gray). If you're dealing with wedding dresses, white bird plumage, or anything else that has detail at the top of the highlight range, the S5 Pro has the ability to pull in two more stops of that.
But it's more interesting than that. If you shoot raw files you have the ability to overexpose and recover in ways that you didn't before. Indeed, most of us shooting with the S5 Pro are starting to think like we did with negative film and intentionally overexpose. Here's why I harped on the channel histogram display before: generally you can overexpose any channel by up to two stops before you can't recover detail. Put another way, you could push exposure of the most exposed channel right up to that two stop barrier and still recover both detail and color information correctly. Go beyond that and you often can still recover detail, but you'll get uncontrollable color shifts.
Unfortunately, Fujifilm's converter only allows a one stop exposure correction. And the camera itself can't tell you the histogram of the S and R photo diodes separately (it should--then we'd be able to dial up our exposure right to the boundary or recovery).
JPEG shooters should be warned that using the expanded dynamic range will almost certainly result in the need for post processing (it's one of the reasons why the film simulations are built into the camera, I think, as they attempt to make other setting changes that bring the final photo pixels into something more "printable"; F2, for example, does a strong contrast boost, which can take out that dull look that expanded dynamic range sometimes produces. Simply put, nine stops of image capture exceeds what you can put down on paper, so you end up having to compress or restructure the dynamic range in post processing to get snappy, pleasing images in some cases. (This is where D-RANGE AUTO comes in: don't use more dynamic range capture than the scene has!)
Overall, the dynamic range capabilities of the S5 Pro are impressive, and, at the time of this writing, unique to the digital camera world. Between this feature and the "Fuji color," you'll find the primary assets of this camera that compel so many wedding, event, and portrait shooters to purchase it. Update: people have asked me to compare the S3 and S5 Pro's dynamic range. For the most part, it's the same. At the highest ISO values the S5 Pro might have a slight advantage (at the expense of detail due to the noise reduction), but it's not significant enough to worry about.
Part of that is that the D200 tends to produce slightly soft (overly aliased) JPEGs. Part of that is that Fujifilm's 45 degree sensor trick does indeed produce more visible resolving power for most situations than 6mp would suggest. But, overall, I'd put the S5 Pro a slight notch below the D200. Just as with the S3 Pro, my conclusion is that the Fujifilm 6mp SuperCCDs really resolve most detail at about the level you'd expect an 8mp camera to.
It's next-to-impossible to prove or disprove that, unfortunately. Just measuring resolution charts doesn't work because you essentially get better results for the S5 Pro on the vertical and horizontal axis and worse results on the diagonals. Vice versa for the D100 (a standard 6mp camera). So we almost immediately get into an arguments about whether improvement on two axes is better than improvements on the intermediary axes. And that'll depend upon what you photograph. I've noticed, for example, that S5 Pro renders city scenes very nicely (as long as the camera is level ;~). Not many buildings have diagonal detail. Surprisingly, many landscape scenics do well with the S5 Pro, as well. You'd think the randomness of, say, sand grains in a dune or leaf details would render any axis-dependent improvement null, but it seems that even on truly random detail I perceive some benefit to the Fujifilm sensor arrangement (that may be because our brains scan for horizontal detail over diagonal, though).
What you won't like is that every now and then a diagonal component in your shots will render with some stair stepping Here's an example of what I mean:
This is a 400% blowup of an 115 pixel wide portion of an image taken in Bodie Historical State Park (the chromatic aberration is due to the Sigma 50-150mm lens being tested). While it's a little difficult to see here, there are no problems on vertical or horizontal lines in this shot, but there's obvious stair stepping on the diagonal. Personally, I can live with that. But it does go to show that you're not dealing with a 12mp camera, you're dealing with a 6mp one. The fewer the pixels in the sensor, the more likely you'll get those jaggies in certain situations. With traditional 6mp cameras you get that problem on near vertical or horizontal lines, but on the S5 Pro the problem most often shows up on lines that are near 45 degrees.
I should probably repeat something I've been saying (and writing) for quite some time now: if you can't get good quality images at the maximum size of a desktop inkjet printer (that would be 13x19") with the current crop of 6mp and 10mp cameras, it's not the camera that's the problem. That applies here. I have no problems getting very nice prints at 13x19" with my S5 Pro. That jaggy image I just showed you represents about a third of an inch in the final print. Let's go ahead and reduce that down for you (I can only approximate it--I don't know what dpi the monitor with which you're viewing this uses):
Still see any jaggies?
But, like the S3 Pro and the other Fujifilm DSLRs, the S5 Pro will certainly have its supporters, and it has a place in my gear bag, as well. As a wedding and portrait studio camera, it has plenty going for it. Most notably that Fujifilm color and the extended dynamic range. As long as you don't push shooting long bursts of images, you may not even encounter the dreaded Fujifilm wait-for-some-buffer-space pauses that were so frustrating with the S3 Pro. They're still there, but you have to push harder to hit them. Overall, the S5 Pro is a faster camera than the S3 Pro, but not hugely faster.
But that dynamic range. If you've shot HDR to get dynamic range, you know how much of a pain that can be sometimes. Any kind of movement in the scene can put a kabbash on your dynamic range intentions. The S5 Pro delivers a real two-stop advantage over the other Nikon mount DSLRs, and perhaps a one-and-a-half stop advantage over the best of Canon's DSLRs. That can make a lot of difference to some (obviously, most photographers would like to kill the person who thought of dressing the bride in white and the groom in black). Moreover, if you shoot RAF (raw) files, I actually think the camera should be overexposed for the best results, and I never thought I'd say that about any brick-wall-at-255 DSLR. I've seen recovery of images that seemed impossible to recover (and apparently Fujifilm thought so to, otherwise their raw converter software wouldn't stop at a -1EV exposure recovery point). By impossible, I mean apparently three and four stops overexposed. The trick is to blow no channel out by more than two stops (or to blow out a channel that won't impact the color of the highlight), otherwise those R photo diodes can't do their recovery job properly. Trust me, nothing else comes close to the S5 Pro in this respect--it's the first DSLR I've seen where I'd say that there is some real exposure latitude capability (though only with raw files).
Should you get a D200 or an S5 Pro? (Or perhaps both?) The answer to that question will lie in your priorities. The D200 is a hands down winner in speed, handling, and all-around decent image quality with plenty of megapixels for the maximum size of any desktop inkjet printer. While I normally shoot with my D2xs (as I write this in mid-2007), I have no qualms at all about instead using the D200 for some shoots. For example, my climb up Kilimanjaro, where I sacrificed a teeny bit of image quality for a big reduction in size and weight. The D200 isn't a great high ISO camera, but it's also not a bad one. Thus, it takes something mighty fine to dethrone it from a trusted place in my bag.
Meanwhile, the S5 Pro is a hands down winner in (un-neutral but) pleasing color and dynamic range. Boy could I have used that shooting white arctic foxes on snow earlier this year. But the lack of speed, some handling issues, and slightly less usable resolution make the S5 Pro more of a specialist than the all-around D200. If you need a specialist that does what the S5 Pro does, you'll love the S5 Pro. That's why, it, too, has found a place in my bag. I don't use it as often, but when I need what it does, nothing else can match it. If I were an event shooter or portrait specialist, I'd probably use it more than I do--there's something quite pleasing about the color rendering of most skin tones (once you get it dialed in with the white balance).
Think of it this way: do you need a Toyota Camry (generally all-around decent wheels) or a Mazda Miata (more limited in function, but great at what it does do)? Nikon has produced one heck of a Camry (D200). Fujifilm has given us a Miata (S5 Pro). More people will buy D200s than S5 Pros, but a lot of those S5 Pro owners will be perfectly happy with what they get.