initial review: 5/3/05
Corrections, added conclusion: 5/4/05
Corrections, ACR 3.1 5/6/05
Update: 10/05/05
D2xs update: 8/30/06
WT-2 comments: 1/22/07

  Nikon D2x and D2xs Review

Was the wait worth it?

  Add a comment or send Thom feedback on this article.

I've added D2xs-specific comments in red. The D2xs is enough like the D2x that the bulk of the review is still valid. I haven't bothered to replace "D2x" with "D2x and D2xs" in the original portions of the review, but you can safely do so as you read.

Announced at Photokina in September 2004, the D2x didn't show up in user hands until February 25, 2005. The wait was agonizing for some. Part of that is the fact that the Canon 1Ds had 11mp in user hands back in late 2002 and nothing had come down the pike from Nikon to match that. When the 1DsMarkII shipped with 16mp and Nikon's response was still missing, a lot of folk decided to jump brands, some making claims that Nikon couldn't keep up with Canon (wrong, but let's not get ahead of ourselves). Part of the frustration was Nikon's usual practice of no sample images, no user samples, no real comment on the product until the product was released. Thus, Nikon aficionados spent fall and early winter fending off Canon troll posts on forums solely on expectation, not on any real information about the camera.

This almost backfired on Nikon. As late as January 2005 I was getting calls from multiple highly visible, influential, pro Nikon users asking if they should switch to Canon. (This was additionally prompted by the revelation from Nikon that user shipments would be pushed back at least a month from the original January 25th date.) There was great panic in the Nikon pro camp in late 2004 and early 2005, and the only thing that was going to quiet that was an absolutely perfect D2x. As it was, a few last minute switchers made the jump to Canon; three of the four pros I talked to decided to stay with Nikon until they could try a D2x.

Fortunately, Nikon delivered something very close to perfect. At least for all those pros who had their hands on their wallets ready to pull the trigger on the C word. Is the D2x perfect for you? Maybe, maybe not. There's much to discuss and the evaluation is not going to be an easy one once we get down to the image quality and white balance encryption discussions. Indeed, I fully expect to get flack for some of my comments. On the one hand, every review to date that has said positive things about aspects of the D2x's image quality--especially in reference to the Canon 1DsMarkII--has gotten a barrage of Canon defense in their In box. Frankly, if you're a Canon user, you should be thankful that Nikon is putting up such good products. Without the competition, Canon would be charging whatever they feel and probably slowing their design cycle a bit from the current frenetic pace. On the other hand, the white balance encryption gives the target user for this camera something very serious to think about.



You may wonder why the D2x and D2xs don't get a Highly Recommended rating. The D2x and D2xs have modest liabilities in image quality, most notably noise in high ISO NEF images.

The Definitive D2x Book Is HERE!

Click here to find out more.


The Basics

The D2x is much like the D2h. Both are a complete redesign from the D1 series cameras. While the family heritage is obvious, virtually every significant aspect of the D2 series cameras has been rethought from the D1 series. The D2x is not a "some new features in a D1 package" type of design. As I said in my D2h review, I was impressed that Nikon took the time to try to improve, refine, and innovate in as many places as it has. Every control, every body curve, and even every part deep in the body (frame, shutter, etc.) seems to have been touched from the D1 designs. Even more impressive is that they didn't rest with the D2h UI and specifications; the D2x has additions and refinements over the D2h all over the place. Thus, we have our work cut out just running down the feature set and changes to it.

One thing that's had the Nikon throng puzzled is why the D1h and D1x appeared together, but the D2h and D2x were separated by over a year. Interviews in Japanese periodicals with some of the Nikon executives all point to the sensor decision. While we've come to expect sensor developments on a continuous basis, we're still very early in the digital sensor era, and sometimes you get caught in the cracks. The sensor used in the D2x had its design origins in 1996 (the D2h's sensor goes back further), but the actual decision to go forward with it came pretty much coincident with the D1x's release. These two sensors share almost nothing in design or even design teams, unlike the D1h and D1x sensors. Once you start making commitments to specific and divergent designs, even minor changes in development schedules can have ripple effects. In short, I don't think Nikon wanted the D2h and D2x intros to be separated by as much time as they were, but with the cameras now in hand, I believe they probably made the right decision.

It seems clear to me that Nikon really wanted to get to this sensor, but didn't get there as fast as they wanted. One rumor indicates that they initially considered the Sony/Nikon CMOS mix for the D1x but decided they couldn't turn it quick enough, even at a lower pixel count. There will be those that pounce on any comment like the one I just made as a sign of weakness on Nikon's part. Their claim will be that Canon doesn't have any problems churning new sensors. Sorry, I'm not going to buy into that claim. Canon showed weakness in sensors until they created the 1Ds sensor, and it appears that the technology under that sensor forms a basis for all their subsequent tweaking. In other words, they aren't currently making radical changes to their now common base technology, but merely refining elements of it.

Nikon has simply taken a different approach with sensors, using at least three different base technologies over the same time period that Canon has stuck with essentially one. And with the exception of the D2x's CMOS sensor, which is new, the other two have been undergoing constant refinement. I don't see any glaring weakness in Nikon's approach (though Sony's apparent use of the same sensor technology in their high-end consumer camera raises eyebrows). Indeed, if anything, they seem to be building a wider set of technologies instead of relying upon one. At this point in time, Nikon seems to have a wide range of choices of how to proceed with future DSLRs and seems to be pushing at all edges of sensor design, partly in partnership with Sony. I wonder if that won't have a payoff at some point in the future. I'll be curious to see which approaches they pick for future products.

Well, I've been talking about the sensor without really detailing it, so let's get specific. The D2x uses a 12.84mp CMOS sensor produced by Sony with Nikon design elements. My understanding is that this sensor is exclusive to Nikon, unlike the basic Sony CCD sensor that was in the D100, though I haven't been able to confirm that. [In September 2005 Sony announced a new high-end consumer camera, the DCS-R1, which appears to use a slightly smaller (10.4mp and 1.65x) version of the same photosite technology as the D2x. It doesn't appear to have the high-speed access that's in the D2x, though, instead substituting video grabs, so there's a fair number of changes in the on-chip support electronics.] CMOS is a lower power technology that can be mass produced in a wide range of low cost factories (fabs). CMOS has two properties that meaningfully differ from CCD: first, CMOS supports additional on-sensor electronics and direct photosite addressing; more important, CMOS is inherently "noisier" than a CCD, all other factors being equal. Nikon doesn't seem to be taking much advantage of the first key difference, and the second is one that has worried Nikon loyalists ever since the camera was announced. (CMOS also has lower power requirements than CCD, but the primary current drain in modern digital cameras isn't the sensor.)

The worry about noise in the D2x stems from a multiplicity of design elements: CMOS, the smaller APS frame size, the smaller overall size of the photosites, the lack of in-sensor NR electronics, and the fact that Sony nor Nikon haven't produced a high-quality CMOS sensor before. That sounded like a perfect storm to some, and the Canon trolls let lose on all those things on the Internet forums. Specifically, the reduced frame size of the Nikon D2x (compared to the full frame Canon 1Ds or 1.3x frame Canon 1D) leads to a photosite size of less than 5.5 microns. That works out to slightly more than 30 square microns available for light collection. Compare that to the D2h's 9.5 micron size, or over 90 square microns available for light collection. Smaller light capture area means fewer light photons are collected; fewer light photons mean that inherent noise properties of the underlying silicon are higher in relationship to the photon count, which means more noise.

Nikon chose to address noise in a number of ways. You'll note that the chip is claimed to be 12.84mp, but the actual number of useable pixels is about 12.2mp. There appears to be a larger frame mask in use on this camera than on some other Nikons. Second, data is kept in 16-bit space right up until the final compression to an 8-bit JPEG (or uncompressed TIFF). Third, data is manipulated in the analog space (electrons) prior to amplification and digital conversion. This is most prevalent where individual channels are amplified prior to the ADC to adjust white balance, but I wonder if it applies to higher ISO values, as well. Finally, the internal digitization engine has a noise reduction component in it (well, actually two different components, one for long exposure noise and another for high ISO noise). When we get to the results section of the review, we'll talk about whether or not Nikon was successful, but note that multiple techniques are being employed here rather than a single one.

Beyond quality, sensors have speed implications. Nikon used a two-channel readout on the D2h; on the D2x they've upped that to a four-channel readout. Considering how many pieces of data we're pulling off the chip and have to perform multiple processes on, the D2x is moving electrons hither and yon at pretty amazing rates. So much so that the 12.2mp files are supported at 5 fps and a subset of them (6.9mp) at 8 fps. D2h users took notice of both those things, asking themselves could they live with only 5 fps to get 3x the pixels or could they live with a "mere 6.9mp" at the D2h's 8 fps frame rate? If the answer to either is yes, then they D2h is effectively supplanted by the D2x. We'll come back to that in a moment.

Other sensor-related items have changed, as well. Once again the anti-aliasing filter seems to have been tweaked. It doesn't seem quite as relaxed as the D70 and D2h, but it isn't as aggressive as the D1 series or D100. Moreover, the infrared and ultraviolet passage has been more highly filtered--the D2x isn't a particularly good IR camera as the exposures get very lengthy, indicating very little near IR is getting through to the sensor. Nikon also seems to note that IR blockage is no longer a separate layer in the filter stack, but has been incorporated into the Bayer filtration layer.

Outside the sensor, the D2x feature set is also state-of-the-art. I'll not repeat the full list of features here, but instead point out some of the things that are changed or added from the D1 series (this first list applies to both D2h/D2hs and D2x bodies):

  • The vertical shooting grip gets a front command dial.
  • The color LCD is bigger and brighter than before.
  • Voice annotation is built in.
  • White balance can be set directly in Kelvin values, and white balance now has its own dedicated sensor at the front of the prism; custom white balance is simple, can be done in two different ways, and you can save settings with names you provide.
  • A new, user-assignable FUNC button has appeared on the front of the camera. The list of possible assignments is shorter than I'd like (you should be able to assign it to depth of field for the vertical release, for example), but it still provides some flexibility that wasn't on previous models.
  • The viewfinder now features additional informational displays--virtually every significant setting can be seen without taking your eye from the finder.
  • Speaking of the viewfinder, the blackout time during shots is so minimal that you'll easily be able to follow action at 8 fps. Coupled with a shutter release lag of as little as 37ms, this has got to be the fastest (non-pellicle) SLR ever made.
  • The autofocus system is new, quite complex, and versatile. An 11-sensor array covers all but the corners of the image area, and can be configured to operate in enough different ways to confound even those that study the manual carefully.
  • The Autofocus Direction pad now has a center push ability, which is used to advantage in a number of different ways (it brings up the histogram when an image is displayed, for example).
  • NEF and JPEG files can be written simultaneously for each image, and you can choose the size and quality of the JPEG file to be written.
  • An interval timer has been added to allow time-lapse and unattended photography.
  • AI and later manual focus lenses can be used with the matrix meter (after a bit of data input, and only in Aperture-priority and Manual exposure modes).
  • The camera can track vertical position and auto-rotate images in Nikon View.
  • The camera uses USB 2.0 for computer connections and connection to the optional WT-1 or WT-2 wireless transmitter. The wireless transmitter allows 802.11b (WT-1 and WT-2) or 802.11g (WT-2) ftp transfers from camera to computer at ranges up to 500 feet (with the optional extension antenna; otherwise 100 feet). (The original D2h can only use the WT-1; the D2hs and D2x can use either the WT-1 or WT-2.)
  • A new flash technology has been added, which Nikon's schizophrenic marketing department can't quite decide whether to call i-TTL or CLS (creative lighting system). The details of this addition are too varied and complex to describe in this review; I'll have more to say in my review of the SB-800, which is required to implement the new flash system fully on a D2 series body.
  • A new EN-EL4 lithium battery is used, and we can see exact battery life remaining. The camera even tracks battery charges and recommends when new ones should be obtained! Unfortunately we get new battery chargers and AC adapters because of this.
  • Non CPU-lenses can be used with matrix metering and TTL flash.
  • Three letters of the file name can be changed (DSC -> XXX).
  • Images can be automatically rotated (both in camera and off with software that supports the image rotation tag).

The D2x (and D2hs) adds to the D2h:

  • Separate RGB histograms (and highlights!).
  • High ISO noise reduction.
  • Support for the WT-2 (802.11g), which allows control of the camera wirelessly via Nikon Capture.
  • An improved matrix metering system (300k patterns instead of 30k).
  • JPEG compression levels.
  • Separation of Color Space and Color Mode for more accurate and controllable color.
  • A new tab (RECENT SETTINGS) provides nearly direct access to the last eight things you set on the camera.
  • (D2x only) A high-speed crop mode with a reduced angle of view of 2x from full frame (as opposed to the usual 1.5x) and reduced number of pixels (6.9mp from 12.2). The viewfinder is etched with this frame line and the camera can be set to use the FUNC button to go back and forth between this mode and normal.
  • (D2x only) Multiple exposures and overlay of two preshot exposures.

This is obviously a wide range of (new) features, almost too much to comprehend in a quick walk around the camera. The list of specification changes or additions I'd still ask for is therefore relatively minimal:

  • I'd like even more control over the patterns used by the autofocus sensors (e.g., "detect only horizontal motion").
  • I'd like more control over the self timer settings.
  • Since the battery charge is known, the cleaning mode really should allow cleaning with a battery (if remaining power is greater than XX%).

Finally, GPS support is back (missing on the D2h, but present on the D2hs), though it requires the US$95 MC-35 Connecting Cable.

Okay, the D2x contains a bevy of features that easily matches anything any competitor has thrown at Nikon. The sole point of contention any Canon user could argue about, as far as I'm concerned, is 12.2 versus 16.1 megapixels and how those images look. We'll get to that, but we have many more things to discuss first.

The D2xs doesn't change much from the D2x. The primary change is that the high-speed crop mode is now easier to see because it uses an LCD overlay to show the crop area rather than four small LED brackets at the corners. More about that when we get to Handling. Other changes include:

  • Changes to the metering that reflect High-Speed Crop when selected.
  • An improved EN-EL4a battery (30% more milliamps mean far longer stamina).
  • ISO values about 800 can now be selected in 1/3 stop values.
  • The maximum continuous burst numbers have improved dramatically, to as many as 60 frames (up from 35).
  • AF has been tweaked a bit, with new Lock-On options.
  • The color LCD is viewable from a larger angle than before.
  • You can store up to three Custom curves.
  • You can crop images in the camera, after they've been shot.
  • A B&W ability has been added.
  • Camera settings can be saved to the storage card and restored, allowing swapping of settings between cameras.
  • GPS data now appears on image playback, if active. VR status appears, too.
  • The D2xs supports Nikon's new Image Authentication software.

Overall, no big changes, but lots of little things that do help getting the most out this camera.


(Much of this section is a repeat of my D2h review simply because the D2x is much like the D2h. I've only altered or added information here where it is something specific to the D2x--yes, there are a few things that are different. Especially note the part at the end of this section on white balance encryption, which is very important to understand).

First, let me say that this is a stylish body design. Lines are clean, markings are clear, and there's less of impromptu design feel to it than we had in the D1 series. It appears that curves and placement of ridges was more carefully considered to help handling characteristics.

Unfortunately, if you have smallish hands, as I do, you'll find some of the positions of controls are wrong. In particular, if you hold the camera primarily using the front grip, you'll find the thumb reach to the Autofocus Direction pad is a little long. The center of the pad ought to be where the right edge currently is. Likewise, reaching to hit the AF-ON or AE-L buttons with your thumb tends to move your hand position back around the grip, which means you don't have quite as solid a hold on the camera. If you have large hands (can get more than an octave spread on the piano) or are shooting with gloves, these positions are less problematic, but be forewarned: Nikon isn't designing for small hands.

The vertical grip continues that problem. The reach for the Autofocus Direction pad is even greater when holding the camera vertically, and the Depth of Field button is no longer reachable without moving your hand position off the shutter release (one of the reasons why I said the FUNC button needs to have other options, such as Depth of Field--you can almost reach that button from the vertical release with small hands). With the D2x Nikon has given some thought to the buttons and Direction pad. Custom Settings allow you to change a number of functions so that you can have "buttons" and "pads" you can reach, especially with the vertical grip, but this comes at the expense of things being labeled with what they do and some variance with "normal" Nikon UI.

On the flip side, the large number of buttons required for a digital body are all well marked, large, and nicely recessed on the D2x. It's both easy to activate a function as well as not easy to accidentally activate a function. That's just the way it should be. The AF-ON and AE-L buttons have been slightly offset at a diagonal so that they're easier to distinguish from one another (the AF-ON is the upper button). The ISO, QUAL, and WB buttons are larger than on the D1 series, and again nicely recessed yet accessible.

The two new buttons on the back are a good news, bad news pair. The voice annotation button is deeply recessed (you have to really want to use it to access it--thus you don't accidentally waste space recording noise due to accidental pressing). The Autofocus Area Mode Selector (hey I don't make these names up) is going to confuse many, however, at least at first. That's because some of the selections also add Closest Subject Priority when selected, some don't. Some show the active sensor in the display, some don't. And they all interact with Custom Settings, which control the release-priority versus focus-priority choice, and the pattern that's used in Group Dynamic AF. If you were confused by Single Area AF versus Dynamic Area AF, you're going to need to carefully study these new controls--they add a great degree of choice to your autofocus options, but they are not obviously intuitive, even to veteran Nikon users.

Personally, I like all the new autofocus additions, but trying to teach them to others isn't as simple as it used to be. The end result, though, is a camera that focuses the way you want it to, so slog through Nikon's manual (or get my eBook!), test, and use all the settings to refine the focusing to the way you want it.

One handling aspect that's welcomed is the new information in the viewfinder. Not only will you know how many shots are left on your card (or how many you've taken), but you can see white balance settings, ISO, image quality settings, and more. If you learn where all the controls are, there's no reason to take your eye from the finder, exactly the way you want a camera designed for action photography. On the D2xs, the new High Speed Crop masking is very welcome. The big difference is that you can now frame even when the meter isn't active, because you always see the frame limits. Before, the HSC brackets only showed up when you activated the meter (partial press of the shutter release), and dimmed when the meter went off. Beyond that, the new masking is full (you see all the frame edges), where the D2x only has corner brackets. It's just a heck of lot easier to see what your final frame will look like.

White balance is more flexible and more easily controlled than before. Custom white balance can be done from a gray card or the incident white balance meter in the camera. Kelvin settings can be set directly (though at pre-determined values). White balance settings can be named. White balance settings can be "obtained" from images already on the storage card. White balance itself is measured via both the viewfinder CCD and a dedicated incident sensor at the front of the viewfinder. A nice step forward overall.

Custom Settings have been grouped in a meaningful way, getting rid of the old scroll-through-30+-functions method of locating what you want to set. Custom Settings banks can be named, making it easier to remember what's different about Bank B than Bank A. Being able to save and load settings via the memory card is also very useful. I carry a 128MB card with me that I save all my settings on. If I get carried away and make too many changes to settings so that my basic sets get destroyed, I just restore them in the field from that card (with the D2x that had to wait until I got back to my computer and could use Capture Camera Control).

The new battery is nice in a number of ways. First, it's smaller than the EN-4 that was in the D1 series. Second, it doesn't have any exposed contacts, so you don't really have to worry about putting that little plastic end cap on when you stick multiple batteries into a pack or pocket. Third, you get relatively accurate charge remaining information, should you need it. No longer do you have to guess whether the half-full battery indicator means you're about to have a dead battery or you've still got some life left in it. And I mean accurate: I routinely can run my battery down to 1% before the camera shuts off. On the flip side, the silly removable camera end cap is a detail I could have lived without. Essentially, you have to "assemble" the end cap and battery every time you change batteries. Why not just give me a door to insert the battery through?

Most connections to the camera (video, AC in, USB 2.0) all attach via connectors on the left side of the camera. Unlike the D1 series, the rubber doors that sit over these connectors seem to stay closed, and provide a pretty good weather seal. Moreover, a side location is better than the previous front location if you hold cameras in the traditional manner (left hand under lens, right hand on grip).

Sensor cleaning is no different with the new model than previous, though there's plenty of "landing space" for swabs on either side of the imaging area. Nikon disclaims all but blower bulb cleaning, and you can't clean without an AC adapter. Yuck. This means that I have to travel with my extra cost AC adapter. Have Nikon engineers actually traveled with their products?

One small touch that has gone unnoticed in other reviews: the eyepiece won't unscrew unless you first close the viewfinder eyepiece shutter. For years, we Nikon pros have been slowly losing our eyepieces when they rattled out from vibration or whatever; no more. Thank you, Nikon.

Overall, handling aspects have improved slightly from the D1 series (which was already quite good), though those with small hands need to be aware that they might be uncomfortable with the reach to the Autofocus Direction pad and AF-ON/AE-L buttons.

A couple of other nits: first, the LCD cover isn't very secure and tends to fall off. I've already lost one. Get one of the Delkin LCD covers if you're really worried about scratching your LCD (less convenient, but more protective). Second, when connected to an external monitor, the color LCD no longer functions. Some studio shooters find that a bit of a problem, as they have to look to the external monitor to change some camera settings or navigate through images for a client.

The D2x has a couple of handling issues that are a bit different than the D2h, so I've separated them out here. First, we've got the Hi-speed crop mode. Basically, this is a 2x field of view crop from full frame that captures 6.9mp at up to 8fps (as opposed to the normal 1.5x field of view crop that captures 12.2mp at up to 5fps). Nikon markets this as a way to get the action capabilities of the D2h, but that's not what any photographer I know is really using it for. Instead, the typical use seems to be for in-camera cropping when you can't get closer to the subject (as in wildlife and sports). If you're going to crop those extraneous pixels later, why not just do it in the field? I just wish I didn't have to dedicate my FUNC button for this, though. That FUNC button is so useful, but you can only pick one thing for it to do. The Hi-speed crop is also valuable for use as a "sports finder." Because the framing is smaller than what the viewfinder shows (the edges are marked with etches and the corners light in red when the mode is active), it's very easy to follow action. One word of warning: the etch markings are at about the 96-98% point, not 100%. Nikon claims they did this for a reason (to keep things in frame when they're moving fast), but I'd prefer to have exactly 100%. I don't like the camera having different behaviors in different modes (the normal mode is 100% viewfinder).

As for being useful for action, the Hi-speed crop means your angle of view is changed 2x from 35mm full frame. That means your 70-200mm functions like a 140-400mm from the same position. Some sports photographers might like the extra crop, some might not. There are DOF implications in a smaller capture size, and some really long lenses now start to become too long. It means that you might have to change the lens arsenal you carry yet again. I have mixed reactions about that, and the change works better for some sports (e.g. baseball) than for others (e.g. basketball). So my advice is to not automatically assume that the 2x will or won't work for you--try it first. What I've found talking to other D2x uses is that some like the option and use it frequently, others don't and don't.

Multiple exposures and overlay exposures work pretty much as advertised. It's rare that I want to do either in camera, but it's nice to have those functions back in the camera. In particular, doing moon montages is a snap with the overlay function, though you need to have a truly black sky to get a nice look on the moon.

As an aside, I do think Nikon should seriously think about making a version of this camera without a few features: drop the vertical grip (ala the F6) and you have a more comfortable camera (smaller and lighter) for long periods of shooting while traveling. I'd be willing to part with a other features to get the price down some more (High speed crop, WT-1/WT-2 support, etc.). In short, a D2x Jr. would be an incredible addition to the lineup, and every pro that's got a D2x would be purchasing one as a backup body. Are you listening Nikon?

The WT-2, since I just mentioned it, needs one small bit of clarification. I've now heard from multiple folk that expected to get 802.11g speeds out of the WT-2 but could only get it to connect at 802.11b speeds. Here's the trick: you don't get 802.11g going point-to-point between WT-2 and your Centrino-equipped laptop. You need to have an Access Point, such as a Linksys router. But the added benefit is that you can now control and access your D2x from Capture Camera Control. That's a fair tradeoff, I think, especially now that laptop-sized (e.g. small) Wireless routers are available. There are some other things to be aware of with the WT-2:

  • Capture supports only the D2x/WT-2 combination. If you use Capture NX with the D2xs/WT-2 combination (which is required, as Capture doesn't support the D2xs) you lose tethered support (you can kind of make up for it with Nikon Camera Control Pro, but the new combo isn't quite as versatile as the old).
  • Battery life runs at about 25% an hour.


top | home

Encrypted White Balance

Once Photoshop's 3.2 converter became available and fully supported the D2x, things changed for the better since I first posted this review, so I've moved the discussion of encrypted white balance here and revised it to address the situation as it exists today. On the other hand, the D2xs intro reintroduced some issues (see the section in red at the end of this column). Let's get started:

White balance information tags for most cameras live in the standard EXIF data recorded with a digital image. Recent Nikon DSLRs have been putting that information into the Maker's tag section of EXIF instead. That could have been a problem, but almost all serious digital photography software has adjusted and knows where to look for the white balance data in a Nikon NEF file. Remember, NEF is a raw format, so in order to render an image from it, you have to apply some camera setting information--in particular, white balance--before you do your demosaic to generate the final pixels.

On the D2x, the white balance data is still in the Maker's tag section, but it has now been encrypted. The public "keys" are the camera serial number and shot number. But you need private information to unlock the value, and until early September 2005, Nikon did not document or provide it to third parties directly. Beginning in September 2005, Nikon has provided a mini-SDK to developers who want to access the information.

Nikon's approach caused major issues for software developers, and this trickles down to impacts on us photographers. The 3.1 converter Adobe introduced supported the D2x but wasn't able to access the "as shot" white balance information (it used an guess instead [Auto white balance]). Eric Hyman, the author of Bibble, reverse engineered the encryption, so his raw converter software was the first to fully support the D2x raw files directly. Dave Coffin also built in white balance decryption to the latest version of dcraw. Others are promising that they'll use the mini-SDK. Adobe, for instance, claims ACR 3.2 uses it. Overall, the problem until the mini-SDK appeared was this: if a developer decrypted the white balance value in software they were potentially liable under the US's DMCA (which is why Adobe initially didn't reverse engineer the encryption). Photographers, on the other hand, found their batch processing workflow crippled if they used software that didn't support the "as shot" white balance values. In short, not a happy situation for anyone.

Now that Nikon has addressed the problems that encrypting white balance caused, the issue is a bit off the radar screen, but still lurking. White balance isn't the only data that's encrypted on a D2x, though it is the only piece that software developers and photographers regular need access to. A better solution would have been to simply open up the format and document it fully.

And Linux (and perhaps PDA) users are still left in the lurch, as the mini-SDK only supports Macintosh and Windows. This introduces a potentially new issue with white balance encryption: what versions of each OS is the mini-SDK going to work with? Note that in the Macintosh world, OS 9.x is no longer supported by most developers, while in the Windows world many are drawing the line at Windows XP. As we move forward in OS versions, I doubt that Nikon is going to invest any energy in making sure that the mini-SDK works with older, legacy OS versions. Like I wrote in the last paragraph, it would have been better to simply document the NEF format. This would let individual software developers figure out which OS versions they want to support. Essentially, Nikon now controls which versions of an OS a D2x user should be using.

But in terms of being a crippling blow to some photographers, white balance encryption no longer is.

Encryption Redux

One sad aspect of the D2xs update is that Nikon once again fumbled on the software side. The D2xs appeared before Capture NX did, but Capture 4.x didn't support the D2xs (and still doesn't fully support it, and may never). This left early D2xs purchasers once again in the converter lurch.

At first I thought that this was merely because of the lack of an S in the camera model tag in the EXIF data. You can, for example, replace that S with a 00 hex value in the file and suddenly Capture 4.x can understand and convert D2xs NEF files. But there's more change happening than at first meets the eye.

My newsletter readers will have noted my use of UniWB when shooting NEF. Short explanation: you use UniWB to get accurate histograms--histograms generated with normal camera-settable values are deceptive, and may cause you to not expose as "hot" as you actually can with raw. Given the D2x's noise properties, you want to "expose to the right" as much as you can once you move past the base ISO value.

The problem is that to load UniWB into the camera, you need Capture 4.x--Capture NX doesn't support writing a white balance to the camera. But Capture 4.x doesn't support the D2xs. Oops. Well, okay, maybe I can trick it by taking a D2x UniWB file and updating the model number tag to D2xs. Nope, doesn't work. Something in the D2xs NEF format keeps this from working. Perhaps it's just that pesky encryption (the sensor key might have changed to a new batch number), perhaps it's a change to something in the format. It doesn't matter, really. Nikon has once again broken something useful.

Moreover, we're still waiting for converters to catch up to understanding the D2xs NEF. Not all do, and while my tag trick gets me by for the time being, that's a real pain in the butt to do with large numbers of files.

In short, on a pro camera Nikon has once again made work harder than necessary for a pro using NEF. If they keep doing that, they won't have ANY pros left by the time a few more models come out. Pros don't have time for silly file format frustrations and changes in workflow. Yes, I know that Nikon would really like to sell us Capture NX, and yes, I think Capture NX has quite a few very nice aspects to it. But, no, you can't tell a pro who's been using a D2x and Capture 4.x who buys a second body, this time a D2xs, that he has to change software and workflow because he bought a second body. NIKON. SHOOTS. SELF. IN. FOOT. AGAIN.

Does Nikon really think we won't notice these things? That we'll ignore them? That they don't impact our work? That we don't notice that their primary competitor hasn't made similar mistakes? If so, I have news for Nikon: you're wrong. Very wrong.

There's an old axiom that applies here: fool me once, shame on you; fool me twice, shame on me.






Battery performance is superb. Want even better performance? Buy the EN-EL4a and use it in your D2x. The extra juice in the new version of the battery makes an already great performance go off the charts. I know pros who no longer bring their charger with them on week long trips--they simply bring two or three EN-EL4a's. The camera seems to power down gracefully, for example, much like the old film bodies. If you accidentally leave the camera on overnight with a full battery, the camera still claims >90% battery charge 24 hours later, and that figure seems accurate. About the only thing I could find that draws down the battery quickly is very long bulb exposures, tethered use, GPS or wireless connections, and PictBridge printing. A 20-minute exposure can pull the battery down 25% or slightly more. This shows just how much power conservation is going on normally; a lot! Still, I can't imagine a situation where I'd need more than two batteries for a day's shoot, though there may be a few prolific photojournalists out there that disagree. [Just as an example: I filled 12gb's of card space in one day with my D2x, most of it with VR active, and the battery still read 30%.] Indeed, if you shoot less than 500 images in a session, you'll probably do fine with just one battery, regardless of how much chimping you do. I can routinely take my batteries down to the 1% level before having to recharge them (I don't recommend that), which means that in a pinch you can really use all the power that is stored in the battery. Batteries also charge modestly fast. All charges I've made from under 30% power remaining have taken just a bit under 90 minutes; Nikon claims about a 100 minute full charge time, and I've yet to have any charge go that long.

CompactFlash write performance can be superb. I use the words "can be" because I have encountered a few older and oddball cards that don't achieve the ultrafast write performance the D2x is capable of. But with a WA enabled card or other state-of-the-art card, write performance on the D2x is faster than the D2h and faster than any other DSLR I've encountered to date. It appears that burst write speeds on the D2x can be more than 2x the D100 with a WA card (you can't directly compare file write speeds, as the file sizes are different). Rob Galbraith has tabulated CompactFlash write performance on his site, and my results are right in line with his--with the right card the D2x is state-of-the-art fast. With the wrong card, it can be mid-pack in speed, or perhaps even fast, but not state-of-the-art fast. If you shoot RAW+JPEG and need the camera to stay responsive as possible, get the fastest card you can--you've got to save a lot of data (as much as 27MBs), and even at the very fast sustained write rates of the D2x that can take measurable time. It can be frustrating when the buffer fills with a slow card and fast action happening in front of you--you'll miss shots. And by the way, you'll want 2GB cards or larger. With as much as 27MB per image (pair) being stored, you'll find that 512KB and 1GB cards get chewed up film canister fast. I've been using a 6GB Microdrive in my D2x, which works fine for the more static types of shots I do. With fast action and the camera set to RAW+JPEG I've encountered buffer waits, though. Like I wrote earlier: get the fastest card you can if you shoot this way. And make sure that it's at least 2GB in size. One of the joys of digital is not having to stop and change film every 36 images--be careful not slip back to that by using small cards.

Autofocus performance is excellent, just like the D2h (though it requires learning to use well, as I mentioned in the Handling section). Once you learn how all the controls interact and find the settings that work for the situations you photograph, you'll find that the new autofocus system is fast, sure, and reliable. But you have to learn the nuances to get the most from it, especially since both the group dynamic autofocus and default autofocus positions re-introduce closest subject priority as an option. Be prepared to spend some quality time with a convenient moving subject practicing with the various autofocus settings. Nine of the 11 autofocus sensors are low-light, low-contrast capable, which makes it hard to find a situation where the camera won't focus (though it's still possible). And if you are having problems with that, use an SB-800 set up to only provide AF-Assist illumination (you can finally set it up so that no flash triggers). I wish the reach to the Autofocus Direction pad was better for my hands, but even though I have to stretch to reach it, I still like this system a lot. The few out-of-focus shots I've gotten have all turned out to be my fault. If you're using the more sophisticated lenses like the 200-400mm, you'll also want to master the focus memory functions of the lens. In short, you absolutely must spend time to learn how autofocus works, what controls you have over it, and how lens/camera interactions work. Failure to spend the time to do that will just have you frustrated with focus when that grizzly bear keeps moving in and out of the low willows.

Color integrity is the best of any Nikon DSLR to date. On Imatest's ColorChecker test the colors are well saturated and there is very little color drift from expected. In real life, if you nail the white balance you nail the color. White balance itself is a bit more accurate in Auto than on the D2h, but still not terribly reliable in anything but daylight and non-mixed lighting conditions. In my "wicked test" of mixed lighting (incandescent, reflected ambient off colored walls, indirect daylight through a window, plus flash), Auto does a quite decent job, getting within 200K for the brighter values, 400K for the darker. Still, Auto white balance seems to have the Nikon tendency of straying to the blue side on occasion, and that won't do for skin tones. The direct Kelvin settings don't exactly match my Minolta Color Meter, but they seem to be consistent to it, so once you've figured out how to set the D2x, you'll be okay (hint: I'm setting consistently higher Kelvin than I or my meter sees).

I show the ISO 800 test from Imatest here for a couple of reasons. Most higher ISO tests I've ran on other cameras show color saturations significantly lower than 100% when you use higher ISO values (here it's 97.29%), and some color errors start to drift quite far from ideal. The D2x is holding up quite well at ISO 800. The blues and most of the reds and greens are very close to dead-on accurate. A few colors (11, 16, and 15 particularly) are drifting significantly, but not enough to get excited about. The only thing that gives me any concern is that the plus values (towards yellow) in the b-channel (11, 16, 12, 7) are all pushing towards the center, and even that tendency isn't likely to show up significantly in shots. Let me put it another way: at ISO 800, the D2x shows better color handling than virtually all previous Nikon DSLRs did at their base ISO.

Okay, that was the easy stuff, now we're on to the contentious items: resolution and noise.

Let's start with noise. There's good news and bad. Or good news and good. Or bad news and bad. Eek! What the heck is Thom talking about now? I think how you view the D2x's noise performance is going to depend a bit upon what your expectations are, how you use the camera, and what you do and don't like in your images. That's not going to be easy to explain.

First, some good news: when the D2x produces noise, it doesn't tend to produce chroma noise when shooting JPEG (well, it does, but the chroma changes are so low and buried in the data you usually won't see them). In essence, as you increase ISO values (without the in camera NR turned on), you'll see an increasing "graininess" to the images. At ISO 100, as they say here on the New Jersey border: forgidaboudit. Tain't der. What noise ya talkin' bout, what?

Not that there isn't any noise at ISO 100, but the D2x runs with the best of the best at ISO 100: noise values are all low enough to warrant considering ignoring. That's assuming, of course, that you exposed properly. Here's what Imatest reported with a slight underexposure (my worst case noise test):

ISO 100:
Gray zone Y-noise R-noise G-noise B-noise
1 0.481 0.528 0.512 0.811
2 0.438 0.529 0.473 0.776
3 0.489 0.605 0.506 0.783
4 0.63 0.856 0.653 0.917
5 0.67 0.736 0.692 1.039
6 0.968 1.013 1.012 1.154
ISO 3200:
Gray zone Y-noise R-noise G-noise B-noise
1 1.163 1.276 1.175 1.354
2 1.355 1.525 1.366 1.564
3 1.961 2.193 1.977 2.198
4 3.295 3.656 3.294 3.601
5 4.524 4.778 4.508 4.778
6 3.734 3.874 3.716 3.833

Those numbers are actually remarkably low, which is where some of the reviews that have appeared are coming up with the "less noise than competitor X" comparisons. One interesting thing to note in the charts: at low ISO values the blue channel tends to be the "noise culprit," while as you go higher in ISO this evens out, with red and blue being nearly identical in noise production, and neither very far from the green! In my experience with other DSLRs, that's an unusual result. Put another way: be aware that the blue channel will produce significant noise in incandescent light at the base ISO; at higher ISO values, noise production balances out amongst the channels.

As you crank up the ISO, the noise levels do go up and noise becomes visible. Like the D1x, I'd say that the line between "ignore" and "pay attention to" is somewhere between 400 and 800. I'm very comfortable with ISO 100 to 400 without making any changes to my shooting, overly worrying about small exposure or white balance misses, the level of in-camera sharpening, and so on. By ISO 800 I would be thinking about adjustments to camera settings if I was shooting JPEG. At ISO 800 the camera kicks in a bit of High ISO Noise Reduction, and you can start dialing it in more aggressively should you desire.

But--okay we're up to the good news good news or bad news bad news part--I decided to do some worst case tests. And I mean worst case. What that meant was indoor (hideous lighting), underexposed by one stop, auto white balance, in-camera sharpening, no extra noise reduction, ISO 3200 (HI-2). Here's what I got (this is a 100% crop):

You're going to be in the good news or bad news camp after looking at that. Remember, that's the worst possible case I could manage with the D2x. Worst as in no other JPEG image I shot had as much noise, and I could manage camera settings to produce less noise in the same situation. Note a couple of things about this image:

  • There's almost no visible chroma noise: we're at the worst the camera can do and you're not seeing the in-your-face green, red, and blue blotches that come from chroma noise and usually appear at the highest ISO value. What we have here is noise that stays "on color."
  • Edges are getting destructed. You can see this on the top of the rim, where we're getting an interaction between the in-camera sharpening (Normal in this case) and the noise, which appears to be exaggerated by the JPEG engine.
  • Colors are remarkably good for ISO 3200, especially considering Auto white balance, underexposure, and some sort of vapor-type of lighting.

You're probably wondering what that looks like cleaned up (again, remember you're looking at 100% view: this a 0.5mp bit of a 12.2mp image):

So, are you in the good news or bad news camp? Personally, I'm in the good news camp. I never expected my D2x to be able to do ISO 3200 in a dimly lit gym with suspect lighting sources in any manner that would be usable. Yet the worst I could provoke out of the camera turned out to be certainly usable for photo journalism purposes. (The few "bright" pixels in Jon's arm aren't hot pixels, by the way--they're artifacts from the Neat Image noise reduction on a sharpened image. Again, this is WORST POSSIBLE case I could produce with JPEG. If I were doing this type of shooting for a living, I'd be running the camera much differently than I set it. It wouldn't be perfect, but it would be perfectly usable IMHO.)

NEF is a different matter. Even at ISO 400 you'll see noise for which you'll want to use some form of post processing noise reduction (Capture's noise controls are often enough). At ISO 800, I'd say this becomes mandatory, and you may prefer what a dedicated noise reduction program can do. ISO values above 800 aren't usable for NEF shooters, IMHO--the destruction of detail you'll get from using post processing noise reduction begins to rob you of all that wonderful resolution you bought the D2x for.

Which brings us to resolution. You just looked at what would be about a 2" square section printed at 300 dpi. Less than 1/24th the total pixels the camera produces. If this image weren't at ISO 3200 you'd be reading the print that appears on the ball below the "2000" (yes, I tried that--I can read it at ISO 100 with the ball not moving from that position). There's a serious amount of resolving power in this camera, and some folk are actually going to have a difficult time with the D2x because of that. (What do I mean by that? You need to watch for mirror vibration, wind vibration, sloppy hand-holding techniques, and be using excellent quality lenses or else you'll find that you don't get "sharp" images.)

But here again we have a contentious situation to deal with. When Bjorn Rosslett published his review (and several subsequent tests have said nearly the same thing), he made some claims that seemed outrageous on the face of it. Essentially, that a 12mp camera was in some ways outperforming a 16mp one (the 1DsMarkII) resolution-wise. That's not quite what he said, but that's the way everyone read it.

What's happening is that we're seeing the inherent differences that two specific design choices result in. With Canon, it was full frame and 16mp. With Nikon, it's 1.5x frame and 12mp. The photosite densities are different: the 1DsMarkII ends up around 140 p/mm while the D2x ends up at around 180 p/mm. Assuming that your lens is good enough to out resolve the sensor, this could come into play in, say, wildlife photography, where you might not be able to get as close to your subject as you'd like. Indeed, I'd rather have the D2x's 12mp and 1.5x crop sensor for wildlife photography than the 1DsMarkII's 16mp and full frame cropped to the same capture size. In terms of what I'd be "resolving" in my images, the D2x would outperform the 1DsMarkII in those situations because of the higher pixel density on the distant subject. While the Canon has more pixels to play with initially, by the time you crop down to 1.5x or 2x, the differences in pixel density start to be a real factor in what gets resolved. Bjorn's original barn examples were correct (though with a non-moving barn we could simply change lenses or distance to remove the D2x's advantage), as are his conclusions about the difference.

When we move to wide angle, the advantage reverses, as we're always going to be able to frame both cameras the same by simply changing lenses and moving forward or back. Then, the Canon has a higher density of recording elements on any given subject in the scene. On the flip side, the Canon is sometimes straining in the corners of the images due to the usual lens design issues, though not as much as some contentious forum posts seem to indicate. Very few of the Nikkor lenses, even the DX ones, have the same tight image circle to sensor capture as the full frame cameras and lenses do. So the Canon's advantage here is less than it at first appears, at least if you value edge-to-edge resolution and acuity. Still, it's an advantage. Indeed, it's easier to see comparing the APS-sized Nikon D2x against the full frame Canon 5D, both of which have a similar number of pixels but differ in sensor size. With quality wide angle lenses on each and with both framed the same, the Canon has a slight advantage.

But let me tell you something: all this deep ending into detail isn't particularly useful. The D2x, 5D and 1DsMarkII all outperform 35mm film and easily show the differences between good lenses and bad. So unless you're considering a DSLR in substitute of a medium or large format camera, I think any discussion of how much or how well a D2x or 1DsMarkII resolves is simply overkill. It's like discussing the relative merits of the top speed of a Porsche Boxster versus a Chevy Corvette. Frankly, the D2x is more camera than many amateurs can handle--they simply don't have the shot discipline (or the lenses) to get all the bits of resolution of which the camera is capable (pardon the pun). Pros shooting for large formats may find a deeper discussion of resolution slightly useful, but when I have to get out a magnifying glass to examine differences printed at 13x19" on my Epson 2200, I think that we're beyond the point where 90% of the market will find that discussion useful. For the pros: you'll need Nikon's best lenses to fully achieve what the D2x is capable of capturing, and, yes, we really are in the realm of better-than-35mm resolution. You'll want to avoid using the D2x on light tripods (such as the Gitzo 1228), you'll want to use Mirror Up or delay for shutter speeds in the 1 second to 1/250 range, and other techniques that keep vibration and movement from robbing that last little bit of resolution.

Which brings me to one more slight liability for the D2x: diffraction. Trying to calculate exact diffraction values for digital cameras is complicated by a wide range of variables (antialiasing filters, Bayer pattern demosaic, noise reduction, photosite well size, etc.). I don't know of any way to reliably make a perfect calculation of when the airy diffraction disc begins to rob acuity from edges. But I can usually visually see the break point. On a D2x it seems to be around f/11. Up to f/11 and the D2x's acuity appears as you'd expect. Beyond f/11 and you'll likely start saying to yourself that the results don't look like you're getting the full impact of more DOF. And at certain settings, the diffraction, noise reduction, and sharpening all start to make hard edges look soft. So stick to f/11 or wider if you can.

These things all add up, though. Put a D2x on a light tripod, set a shutter speed of 1/15 without mirror lockup and an aperture of f/22 and though you might think you should get plenty of depth focus, you instead get acuity results much less than you expected. As I said: the D2x is more camera than some casual users will be able to handle.

Suffice it to say that the D2x packs plenty of pixels, the camera acquires and produces them well, and unless noise levels are high enough to be visible, you'll have more than enough detail to work with in your D2x images as long as your shot and work discipline is high.

If you're noticing that I didn't say anything about D2xs image quality, that's because it's identical to the D2x.

Was the wait worth it? Absolutely yes. The D2x pretty much deals with every issue that's been brought up with previous Nikon DSLRs. Better-than-35mm resolution? Check. Improved noise handling? Still has noise, but improved, so check. Better color? Check. Better metering, AF, and white balance? Check. Better battery? Double-check. Improved handling? Check.


  • Potentially long reach for small hands. This is a big camera, and shooters with small hands won't find all the controls quite where they want them.
  • Buffer could be bigger. Yes, you heard me right. Because some will be tempted to use the D2x instead of the D2h for sports and action, the 16 to 22 frame buffer starts to sneak into play. Yes, you can switch to High-speed crop and get a bigger effective buffer, but once you've shot action at 12mp and seen the stunning amount of detail that can be captured, you won't want to switch to 6.9mp. D2xs owners get a bit of a boost here, as the burst performance of the new model is better than the original D2x.
  • Requires SB-800 for best flash. If you're going to get a D2x, be prepared to get at least one (and probably three or four) SB-800 Speedlights, as well. Flash is much better with i-TTL than it was with D-TTL, and you have more control and multiple flash TTL with the new system. But you need an SB-800 to get TTL above 1/250 and to control wireless, multiple flash TTL.
  • Wide isn't wide. If you do scenic work, as I do, you need DX lenses to restore your wide angle due to the 1.5x angle of view change. We've now got enough pixels that the flaws of wide angle lenses, particularly chromatic aberration, are more apparent. The 12-24mm is a nice mate with the D2x, but you'll see that it has a bit of CA you weren't seeing with a D1x.
  • Noise at higher ISO values. While I'm perfectly happy with the ISO 100 to ISO 400 performance, some will want even better results. High ISO noise is definitely visible, though JPEGs are free from chroma noise, and edge detail is hard to retain if you turn up the noise reduction or use post processing noise reduction.
  • White balance encryption. Despite being dealt with by the mini-SDK, it still is a lurking thorn.
  • Diffraction. Being diffraction limited at f/11 is a bit limiting.


  • Fast. Fast frame rate, fast mirror return, fast shutter lag, fast autofocus, fast write to card, fast, fast, fast. Effectively renders the D2h (and D2hs) into a lower-cost, lower-resolution solution.
  • Flexible and Controllable. New controls, more options, and improved user control all around. Some of the stuff is modest (like the Most Recently Used Settings tab), but it still adds up to making an already flexible camera even more so. Even the minor changes to the D2xs have positive impacts on flexibility and control--the good got better.
  • Rugged. Nicely built. Well weatherproofed, nothing sticking out to get clipped off in abusive handling, controls that are designed for use with gloves, magnesium alloy body with padding in the right places. Nikon has built the SLR hammer for the 21st century (the F being the original 20th century SLR hammer).
  • Flash. i-TTL works, and it restores multiple flash ability. No more needs to be said. Wedding photographers are going to swoon over the flash capabilities.
  • Excellent Image Quality. 12mp is a lot of data, and puts us into the realm of "that might be all we need." Acuity is good, color is excellent, and noise performance is excellent at low ISO values and acceptable at higher ISO values. With the right settings and discipline, this camera performs at the state-of-the-art.

top | home

Don't forget my quarterly newsletter on DSLR issues, supplied on CD. The Nikon DSLR Report is the only way to learn more about getting the most from your camera.

Cleaning the Sensor

I've posted an article on how to clean sensors.
 | Nikon | Gadgets | Writing | imho | Travel | Privacy statement | 2007 Thom Hogan. All rights reserved.