|Nikon D800 and D800E Review
A D3x successor in a D700 body.
|Add a comment or send Thom feedback on this article.|
I suspect that this review will produce a great deal of feedback, both positive and negative. The D800 is the best of cameras, the D800 is the worst of cameras. It really can be both things. Which it is depends upon your skill set, your expectations, your ability to adapt, your lenses, and even your luck of the draw in terms of quality control, it appears. We've got a lot of ground to cover, so let's get started.
The D800 announcement in February was both expected and a surprise. Expected in that a D700 replacement was overdue. A surprise because of one number and one letter: 36 and E. There were additional sub-surprises in the details: same focus system and performance as the D4, uncompressed video output via the HDMI port, even the US$3000 price. The general reaction to the details was overwhelmingly favorable, so much so that overnight tens of thousands of pre-orders piled into shopping carts of online retailers and every camera dealer suddenly found themselves with a wait list of customers eager to buy.
Overall, the D800 is not much more than a D700 made current and then supersized at the sensor. Let's take both of those things in sequence.
The D700 to D800 transition starts with a complete rethink of the shapes and controls. At first glance, the two are twins (see pictures, below). At second glance, you see that they're not identical twins. The D800 is more rounded in a number of areas and has a completely redesigned hand grip, complete with some rethinking about the shutter release position. A record movie button has been wedged into the area behind the shutter release. The Shooting Method dial and the buttons up top have had a modernization, and we get a BKT button out of the deal. The flap covering the connectors on the left side of the camera (as you hold it) is now a door, and inside we've lost the standard AV out, gained a microphone in and headphone jack, and the USB connector is now 3.0 mini. The Focus Mode control on the front of the camera is now the D7000-style button+lever. Out back we've got a slightly larger display, a new Live View control, and the Zoom buttons have swapped.
Inside, we still have a CAM3500FX focus module, but it's been tweaked to support f/8 lenses (not with all sensors, though, only 11). Likewise, the metering sensor has been updated to 91,000 pixels (ala the D4) and is much more involved with the focus system, as well. The optical viewfinder is now 100%, though the eyepoint was shortened a bit to accommodate that. The D800 models take two cards, one CompactFlash and one Secure Digital. The battery is now an EN-EL15 (shared with D7000 and V1 models).
The D800 models require a new vertical grip, the MB-D12. The grip has a Direction pad (though small) and AF-ON button, but no AE-L/AF-L button. You can stuff an EN-EL15, 8 AA, or an EN-EL18 (D4) battery in the grip, but you'll have to take the grip off to get to the battery in the camera for charging if you're thinking that you'll go two-battery via this route.
But you're probably mostly wondering about that number and letter.
The D800 models use a new 36mp FX sensor. Put another way: 4912 x 7360 pixel images. That makes the D800 (temporarily at least) the highest resolving FX camera available, and by a fairly wide margin. Just so we put that in proper perspective, jumping from 12mp to 36mp does not mean 3x the resolution. Resolution is measured linearly. The actual resolution increase over a D700 is a bit more than 70%. The D800 models also resolve about 20% more than a D3x. Those numbers probably deflated you a bit. They should. While I'll never turn down resolution (all else equal), the big visible gains were in the ramp from 3mp to 12mp. When you consider it in print terms, you start to see why. At 360dpi (the native level of the Epson inkjets), a 3mp camera can print a 6" wide print, a 6mp camera can print an 8" print, a 12mp camera gets you to an 11" print, a 24mp camera nets 17", and the 36mp D800 models get you to 19". Typically, you can squeak by with 1.5x to 2x those sizes, so those old 3mp cameras were laboring to fill a magazine page, while a 12mp camera did so with ease and could be stretched to a two-page spread without too much trouble. The "sweet spot" for most photographers would be between 12mp and 24mp, depending a bit upon what subjects they shoot. The 36mp of the D800 models is outside most people's sweet spot.
Since most people don't print big, once we hit 12mp we got into a good comfort zone for them. Except for the Big Croppers. The lens/composition challenged have problems with only 3000 x 4000 pixels (12mp). Which brings me to the first of the controversial things in this review: I'm finding a lot of folk picking up the D800 models fall into the Big Croppers group: they're buying a D800 because it allows them to crop dramatically. As in pulling out 12mp pieces from a 36mp source. I'm not sure why these folk think that's any different than owning a D7000. If you're always cropping that much, the D7000 is actually slightly better than the D800 in terms of pixel density (very slightly: 16mp DX on the D7000 versus 15mp DX on the D800), plus you'll save enough money to buy some lenses that'll let you crop less.
Update: it seems that my use of the term "Big Croppers" is being confused with "cropping." I capitalized the words Big Croppers for a reason, but apparently some reading this review think that I meant "any cropping." Cropping happens. The 36mp of the D800 models is indeed useful when you encounter a situation where you can't get close enough or want to change format from 3:2. That's far different, however, than someone who just uses the camera as "capture everything, crop later." I truly don't think that group is going to be all that happy with the D800. They're paying for a high-end camera and then creating lots of post processing work when they would be better served by the right equipment and a little more compositional thought up front in the first place. Now read the following paragraph with that as context...
The problem is this: those that want the D800 models primarily because of the cropping ability are a different kind of shooter/customer than those who want a D800 model because of resolution. The former skew highly consumer, while the latter tilt more towards highly prosumer and professional. This is the first of my "best of cameras, worst of cameras" points. The latter group knows what they're getting into, have clear expectations that are met, and are well served by the D800 models. The former group is going to find things like diffraction cause them issues; I'd argue that they're better served by other cameras (or more training first).
To E or not to E, that is the other question. The snarky Thom wants to write "if you have to ask that question you don't need an E." He'll be back for some more words later in the review, but let's sub in the helpful Thom. Nikon has given you two choices: a weak AA filter or no AA filter (but a bit of a filter, nonetheless). Both models begin by doing anti-aliasing on one axis. The D800 model then anti aliases the second axis before delivering the light through to the sensor. The D800E model undoes the single axis anti-aliasing before passing the light on.
Let me be clear about this: both models will produce moire and color artifacts. All the moire examples in my Complete Guide to the Nikon D800 and D800E were produced with a D800, not a D800E. The D800E is more prone to producing moire, but only slightly so. Thus, the question becomes: will you benefit from the lack of anti-aliasing? I'll have the answer in the performance section below.
The D800 models are strong performers that can produce knock-your-socks-off results in the right hands, with the right lenses, with the right technique. They are not cameras for everyone, however.
Review product source: two purchased bodies, one D800, one D800E, several borrowed bodies.
Support this Site
As noted earlier, at first glance the D800 is an updated D700. If you're considering updating, here are the major changes worth noting:
From a feature standpoint, the D800 doesn't break any truly new ground other than at the sensor. If you've used a D300, D300s, D700, D7000, D3, D3s, or D3x (and maybe even the D200/D2 series bodies), you're going to find yourself in comfortable territory in terms of feature set. You might find a few additions or deletions in the feature set depending upon which model you're coming from, but with one exception I think it's doubtful that you'll be concerned about any of those changes from a feature-only standpoint (hmm, that bit in italics sounded ominous, didn't it? Stand by for Handling ;~).
The feature exception is frame rate. The D800 maxes out at 4 fps for the full FX frame. If you shoot DX crop, especially with the MB-D12, you get a bit more (max 6 fps), but if you're using the D800 model as the FX camera it was intended as, that 4 fps is probably a bit disappointing. The D700 was 5 fps, after all.
Of course, we're moving 3x the pixels off the D800 model, so it's actually performing pixel movement faster than a 12 fps D700 would. That is to say, Nikon has snuck some bandwidth increase into the D800 model, but not enough to bump the maximum frame rate. I'm not a big fan of fast bursts of images, so the slight change doesn't bother me an iota, but from emails I've received I know it bothers many of you. So let me put it in a different context: the 24mp D3x was 1.5 fps at 14 bits raw. From my viewpoint, the D800 models are producing more pixels and can shoot more than twice as fast doing it. I'm happy. You may not be. Another of those "best of cameras, worst of cameras" things, though not a deal breaker.
Update: it seems that a lot of people are focusing on this section of the review when they try to characterize my review. The also think that this section is negative. Note the title: Handling. My point is that handling has changed. Whether you think those changes are a problem for you or not is up to you, not me. As a long-time user of Nikon gear, I can only report them and how they impact my shooting. I'm also surprised at how many people don't get the obvious attempt at self-mocking humor (next line).
Ba-ba-bum (ominous orchestral tease).
Who moved my cheese? Why is it a different flavor? Why are there holes in it? (For the poor Japanese translator trying to make sense of that: I'm referring to a best-selling book that Nikon engineers really need to read (Who Moved My Cheese?))
One nice thing we Nikon users have appreciated over the years is Nikon's dedication to legacy support. If you grew up on a Nikkormat FTN as I did 50 years ago, you'll find things remarkably similar in any current Nikon camera. Not so much with a D800 model. Nikon moved a lot of cheese.
Let's start with a simple one that'll catch everyone for a bit: Nikon has abandoned (most of) their legacy support for odd orientations. What the heck does that mean? Well, the Nikon exposure meter, for example. Plus has always been on the left, Minus on the right. Not any more: the default is the opposite. Why was Nikon's metering the way it was? Because of lenses. The metering bar was telling you which way to turn the top of the aperture ring. Of course, new Nikkors don't have aperture rings, so that isn't really necessary any more. Except, of course, for all of us that have legacy lenses and use them on our cameras. The good news is that you can still reverse the bar to its old state, but the old state is no longer the default. I suspect that this change is because Nikon got a lot of feedback from people switching to Nikon from Canon (which uses Plus to the right). (Of course, they could have just put a "Canon Shooter" option into the Custom Settings and modified the camera to the way a Canon shooter would want it.) Yes, this is a little thing, but these little things are going to add up.
Next, we have the zoom buttons. I suppose if we're going to rationalize the Plus on the right for exposure, the Plus should be on top for zooming, right? That's what Nikon has done. Again the opposite of what experienced Nikon shooters are used to.
The big control change that stops most loyal Nikon users in their tracks is the Focus Mode button/lever on the front, D7000 style. Now, I happen to like that, as it means I can change autofocus settings without taking my eye from the viewfinder. But a lot of you coming from earlier cameras are just going to find that another piece of cheese that isn't where you expect it. My problem with the Focus Mode control is that the button is too hard to find, even with my hand in the right position to find it. The button is small, and an inward dimple does absolutely nothing to help you find it (it should be a raised dimple).
Next up is the Mode button (controls exposure mode). To squeeze in the red "record video" button, Nikon has moved the relative positions of the Mode and Exposure Compensation buttons. You're going to find that you press the record video button way more often than you want to. Fortunately that button does nothing unless you're in Video Live View. But it also means that until you get used to this change (and I'm still not used to it almost six months later using cameras with it), you end up taking your eye from the viewfinder to change exposure mode. Heck, even looking at the camera I keep pressing the record video and Delete buttons when I want to format a card. This was probably the worst of the legacy changes: the two buttons behind the shutter release were the same on every camera since 1988 with very few exceptions (N80 was a big exception, swapping the Exposure Compensation position). Long-time Nikon shooters' decades of muscle memory have been violated.
The Live View control is going to get its share of frustration, too. Indeed, Live View has dramatically changed in nature from its earlier implementations (the D4 shares this change, as it does many of the other things I'm writing about, by the way). While the mirror-always-up nature (except when flash is used) is a highly welcome recent change to Nikon's Live View (shared by the D5100, D7000, D4), the other changes are frustrating a lot of pros. In particular, studio and macro shooters are complaining. What are they complaining about? How exposure is handled.
The new Live View system in the D800 respects apertures. I mostly like that change. What that means is that whatever you see on Live View is at the aperture you've set. Previously, Live View used the aperture you had set when you entered Live View (and a few had a maximum aperture they'd set for viewing, typically f/8). That's all gone: from a DOF perspective, what you see is what you now get. Change the aperture while Live View is active, and the aperture actually changes. Bravo. But it's the old Nikon giveth and taketh thing: you're going to need to learn when Live View is "faking exposure" and when it is showing "true exposure." It's not something you can set directly, it's a derivative of what you're doing and what exposure mode you've set, and what controls are active. It takes me a bit to describe it fully in my book, so I'm not sure I can do it justice in a short review (yes, this is short for me). But let me cut to the chase that's enraged a few studio shooters: they shoot in Manual exposure mode with Live View (so that they and their clients can see the image via monitor). There's only one problem: they use very bright studio flash for the actual shot and there's only dim ambient when composing. You guessed it: in this situation the camera is showing "true exposure" so what's on the color LCD during Live View is an underexposed image, sometimes so underexposed the display might as well be black.
The irony is this: Nikon made a big deal of the "dual white balance" abilities they added. You can have Live View show you one white balance (ambient studio) while the shot is taken with another (flash corrected). But if you can't see the Live View screen, it doesn't make much difference what the white balance is, does it? (By the way, the manner in which you bring up that alternative white balance isn't exactly intuitive: press the OK button, hold down the Thumbnail button, and turn the Rear Command dial. Moreover, since there's no indicator of "display WB set" it's easy to get confused as to why "changing WB" doesn't show any difference.)
In short, the new Live View is another of those "best of cameras, worst of cameras" things.
But we're not done with missing cheese.
Next up we have a doozy: you can take out of focus pictures when you've got the camera set to take only in-focus pictures! A setting of "Focus" doesn't actually mean what it used to. In fact, not only does it not mean what it used to, but it also removes a very useful capability that virtually all higher end Nikon models have had for 30 years: trap focus via the AF-On button. This change is so egregious that I made my complaint directly to Nikon: at least give us the option to restore the old functionality via a Custom Setting.
You think I'm done? Barely started, really. Apparently Nikon believes that the D80 matrix metering system was the world's best (it was in actuality the worst matrix metering system Nikon has produced). More so than any other camera, the D80 would rethink its matrix patterns based upon the gray value it found underneath the current autofocus sensor. Put the focus on a bride's white dress and you got underexposure. Put it on the groom's dark suit and you got overexposure. Well, it's back. Not nearly as bad as the D80 was, but the D800 has the same tendency to overemphasize correcting what's under your chosen autofocus sensor. Try this little test at home: set your camera up on a tripod pointed at ColorChecker chart. Make sure the chart is a large part of the frame and that you can put your autofocus sensor on the individual white to black patches on the bottom. Now try moving the autofocus sensor in matrix metering and see what you get. I just did it: 1/320 on white, 1/200 on black (all other settings and lighting not changed). That's two-thirds of a stop difference when the exposure shouldn't have changed at all. This is a step backward, and not the matrix metering most of you high-end Nikon DSLR users are used to.
Why isn't it being complained about? Because a lot of you are shooting people. The color metering information in the viewfinder actually does a good job of finding skin tones and doing the right thing with them. So if you focus on people, matrix meter not so bad. Focus at different parts of a high contrast landscape, matrix meter bad. (Okay, "bad" is probably too harsh a word: the meter is never truly bad, just wrong a lot more than it was on my D3 models.)
Sometimes the cheese stays the same when it should change: we have the same D3s type of bank control, including that Extended banks setting that most people have no idea what the heck it does. Surprisingly, Camera Control Pro can't set all of the settings on a D800 model, either. Banks are the same mess as they've been for some time, in other words. We don't even get the C1, C2 abilities (also not exactly well thought out, but better) that the D7000 has (and the upcoming D600 will have). The reason I mention this is: cheese we don't want is still there, cheese we wanted is moved or missing. Is that really the best Nikon can do?
Okay, we have some other handling issues to explore.
That 100% viewfinder: if you have glasses, you may have troubles seeing the LED display below the image area; the tops of the LED get cut off if your eye position isn't right snug up to the eyepiece and centered. This is going to bother those with transitional lens glasses a lot, I think, as they move their eye position to get the best correction from their glasses. A number of people have complained about the short red highlight when you move focus sensors (they should program their center button to show autofocus position; you get a longer highlight when the center button is pressed). We still get grid lines that divide the image area into fourths, not the thirds that every book on composition would suggest.
More so than any previous Nikon DSLR you need to be using the built-in eyepiece shutter if you're not looking through the viewfinder (that includes Live View use). I'm not sure why, but ambient light entering the eyepiece can and will be recorded on your images with the mirror up! Perhaps not a lot, but enough to mess you up, especially if the light behind the camera is intense.
A number of people have complained about the slight change of right hand position and location of the shutter release on the D800 models (and D4 for that matter). But I think that Nikon basically got this right. I prefer the D800 positions slightly over the D700 ones. That's especially true of the optional vertical grip, though its small Game-boy type Direction pad is a little problematic for some. I should note that the MB-D12 user is going to be button deprived, despite the AF-ON button being there: you can't reach the Fn and DOF Preview buttons and you have no AE-L/AF-L button. All that wonderful customizing ability that Nikon gave us on the three primary buttons is messed up by the vertical grip only having one accessible button.
Those of you thinking you want to record uncompressed video from the D800 models need to read my article on recording HDMI: the things you have to change and watch are scattered all over the menus and handling of the camera.
In the land of positive changes: automatic ISO now can be focal length aware (Auto setting for Minimum shutter speed), something we've all been asking for since automatic ISO first appeared. Not only has it appeared, but it is tunable (Slower to Faster).
The bottom line on handling is this: you don't just drop your D300 or D3 and pick up a D800 and be happy. You're more likely to be frustrated. At least until you change some defaults, figure out how you're going to deal with missing cheese, and get used to new controls and positions. A year from now I'll probably be complaining less about the changes in handling, but for now, three months in, I find that there's still enough cognitive dissonance in moving from my D3x to my D800 that it slows down my shooting considerably. One thing that was always true of Nikon (just pick up any body and shoot) is not so true any more. I consider that a loss.
Before we get into performance particulars, I need to get some more general things out of the way. Like quality control.
I wrote it in an earlier article on this site: I'm concerned about the quality control on the D800 models. Some of this has to do with bugs in the firmware (I found many, several of which I reported to Nikon, one or two of which have been fixed in firmware updates; eventually most if not all should be fixed). I don't remember a Nikon model in which I've encountered so many unexplained problems as I ran the camera through the paces necessary to write my Complete Guide. At least once I triggered an ERR condition on my D800E, even though I don't believe that there was an actual problem with my camera (hasn't recurred, and I was doing some abuse testing of Live View and menu interactions at the time).
Meanwhile, we have numerous reports of focus issues, in particular left AF sensor problems, which I've verified in at least two copies of D800 models I've examined. Nikon now has a redone test and correction procedure for dealing with cameras with this problem, but if you encounter it your camera will have to go back to Nikon for the fix.
Finally, on my D800 some AI-S lenses don't index the aperture correctly and report an FEE error, while my D800E has no problems with those same lenses.
In sum total, I've personally seen and heard from others of enough cases to come to the conclusion that this isn't the level of QC I expect from Nikon, especially for a US$3000 camera. I seriously hope that my sample is skewed and that there isn't a decline here, but I do note that real D800 complaints seem to be prevalent on pretty much all Nikon-related sites at the moment. So my recommendation is this: don't be afraid of those reports, but also don't rush off to a critical shoot with a brand new D800 model before you've seriously put it through its paces and you're sure there's nothing amiss. Trust, but verify.
I tested both a D800 and D800E model.
In further reviewing my notes, I see two things I need to look at more closely: (1) most of my EN-EL15 batteries are well into the middle of their life (and I've got one that is at "4"); this may be skewing my average shots number as lithium batteries tend to lose a bit of their capability over continued use; (2) shooting sessions that were just that (not much image review, no MUP or Live View, no use of gadgets like GPS) show higher numbers than what I report above. One reader wrote to me that he's seeing twice the number I report in "normal" shooting. Given Nikon's comments about the change in the D4 battery, I'm wondering if they're tuning the cameras differently than before, so again, I'll have to look at this further to make a more specific comment.
The recall on EN-EL15 batteries made them scarce for awhile, but I'd strongly suggest that you pick up a couple of extras, especially if you do anything other than "normal" shooting. That's doubly true because the supplied charger will take two-and-a-half hours to fully recharge an EN-EL15. This is something I'm going to have to watch when I'm in Africa. I'll likely chew through a couple of EN-EL15's in a typical day, but now I've got at least five hours of sequential charging ahead of me if I don't bring an extra charger. The possibility for slowly getting behind on having a full set of fully charged batteries is definitely there for those with minimum number of batteries and maximum amount of shooting.
The MB-D12 doesn't particularly help (unless you opt to use D4 batteries in it). True, you get two batteries to use before you have to change, but now you have to remove the grip to get to the in-camera battery to recharge it. It's wear and tear on the connection between grip and camera, and it's a pain to keep doing over and over. Not a solution I'd lean on myself.
Writing to Card
The caveats are these: (1) If you shoot Raw Primary JPEG Secondary or Backup for the second slot, the SD card is going to determine your camera's performance, and there the news ain't so hot. The fastest SD card I've got barely hits 40MBs, and many of my SD cards are far slower than that, often in the 20MBs range. Solution: buy fast cards, use only Overflow. (2) If you shoot raw you need UDMA-7 (CF) or UHS-1 (SD) cards. Really, you do. The real test comes when you fill the buffer: UDMA-7 and UHS-1 cards will return the camera to a shooting state faster, simple as that. Using some older cards, such as the SanDisk Extreme III, well, the performance was dramatically lower, enough so that I it frustrated me. (3) It appears that there are some older cards that simply won't work in a D800 model. I don't have enough stats yet to trust publishing a list of these, but it seems pretty clear that a few common pre-UDMA CompactFlash cards fall into this category. If you skimped by purchasing low-end, high-capacity cards a few years ago, you're going to want to check your card stock carefully, as I bet they won't all work in your new D800 model.
One thing that worried a lot of potential D800 buyers is how big the files are. If you shoot raw without compression you generate 76MB files, after all. But you don't have to do that. Lossless Compressed 14-bit gives you 41MB files, a big gain, and doesn't seem to change the speed dynamics of the camera. In general, unless you're constantly maxing out the buffer, even moderately fast cards (300x CF, 45MBs SD) may suffice (though I still say UDMA-7, UHS-1).
The real issue here is balance of card size versus file size. Really large cards do slow down the camera for image review. So that temptation to buy 64GB cards? Suppress it. 8GB is probably the right balance, and certainly no more than 16GB. That may mean you need to carry extra cards, but I'd be tempted to load up on state-of-the-art 8GB or 16GB cards for my D800. That also reduces your catastrophic loss risk should something happen to a card (you lose ~100 raw images on an 8GB card).
There really isn't a lot to say: in standard shooting the D800 models' autofocus performance is basically state of the art.
Of course, flip the mirror up for Live View and things change. Nikon still has a long way to go to master contrast AF. If you're shooting video with the D800 models, don't expect camcorder type autofocus performance. Heck, don't even expect m4/3 type autofocus performance. More like "older compact camera" performance, at best.
In particular, the green channel response on the D800 is a little better than the D700. The red and blue channel are a bit weaker, yet we still get the same results in our JPEGs. That's a really nice touch of EXPEED tuning by Nikon, and appreciated. Of course, not everyone likes Nikon's standard color (not as warm as Canon, not as yellow-biased as Fujifilm, etc.), but frankly I've liked EXPEED and Picture Controls since they first appeared in 2007. There's enough color control there for me to tune things the way I want, and the repeatability across cameras is great.
Whoa, Thom! What about those difference in channel response, doesn't that mean something? You know, I thought it would, but in practice it really doesn't. If you let more light through the green portions of Bayer and less light through the red/blue, you'd expect that you'd get better luminance data (less luminance noise) and worse performance in non-standard light (e.g. incandescent would produce more color noise). Either I have my expectations wrong or Nikon has pulled a rabbit out of their hat. I suspect a bit of both ;~). Let's move right to the noise comments:
Noise and Dynamic Range
We're now in an era where output sizes aren't really expanding but capture resolution is. For almost everyone, they've got a fairly fixed print size. Let's assume for a moment that this is the maximum size a desktop inkjet can print (13x19"). In that 19" a 12mp camera is going to have to fit 4000 pixels, a 36mp camera 7000. So we've got 210dpi for the 12mp image and 370dpi for the 36mp image. Pixels are buried in both cases, but more so in the 36mp image. You're simply not going to see low-level noise in visible amounts using either camera.
Meanwhile, at the other end, for those pictures we have the same amount of light hitting the same sensor area (FX, or 24x36mm). The net result is that: same amount of light, same sized print, you're going to get similar looking results from the 12mp and 36mp cameras. We can completely equalize things by first downsizing the 36mp image to 12mp, thus masking per-pixel noise, but I don't think we have to go that far for most people because they're simply not printing large enough to produce a visible difference in the first place.
Will you see noise in pixel values at higher ISO settings when you pixel peep at 100% on a D800 image? Absolutely. But we've been in a realm where pixel peeping isn't necessary for awhile now (unless you're tying to print as big as possible, where pixel-level issues become visible). Simply put, I treat my D800 cameras much like I treated my D3x: I'll use them without concern up to a particular ISO value, then switch to my D4 above that (previously D3s).
Dynamic range at base ISO is the best we've seen so far. The D800 doesn't have the read noise limitations that the D3, D3s, and D700 had. I use a slightly different way of measuring dynamic range than you'll see on other sites, and I've tried to keep that consistent for a decade (though it does involve a bit of subjectivity on my part, as I study what I can achieve through to a print). The D3x was a bit over 10 stops on my scale. Well, the D800 eclipses that by about a stop. Even more importantly, shadow detail is very, very good and recoverable on a D800 image. So for very high contrast landscapes, I shoot just like I did with the D3x (UniWB to get highlights positioned right), but I can generally pull up shadow detail better than I could with the D3x. I keep graduating to less and less strong graduated ND filters. These days I carry 1 and 2 stop soft and hard. The days of needing 3 stops (let alone 5 stops) are gone.
Each bump of ISO will lose close to a stop of dynamic range, though. 11, 10, 9, 8, 7 and we're as far as I want to go in terms of losing dynamic range (that would be ISO 1600). That said, we're still better off than we were with the best ISO 50 slide film by the time we get to ISO 3200 on the D800. If you're not pushing the image to huge print sizes, the higher ISO values are quite usable on the D800, though you need to watch out when in high contrast situations, as you'll be picking what tonal values you can keep.
In terms of per pixel basis, ISO 3200 is the point where per pixel (actual view) noise starts to be clearly evident. At ISO 6400 it increases substantially, but is still well enough under control I wouldn't avoid it (note that camera will be applying noise reduction on JPEGs even if you set that to Off once you get above ISO 1600). Some of you may find HI1 (12800) usable, but I find it very blotchy and too low in dynamic range. HI2 is unusable, in my view: things clearly fall apart at this value. If you push up noise reduction as high as it goes, you get crudely defined edges and color shifts at HI2, too.
Yeah, I know you want pictures. Here's my standard basketball test (underexposed, ISO 3200, High ISO NR Off, otherwise defaults). First, a JPEG straight out of camera (actual pixels):
Here's the same image as a NEF version cleaned up in Adobe:
Now a little game action so you can see how that plays:
Not bad for a camera that's not supposed to do indoor sports. By the way, there's something else being shown here: the camera was set for AFC (continuous AF) 3D (camera picks sensor), as I was also testing autofocus performance at the time. This is from a sequence of seven shots (almost two seconds) where the action moved from the key to the corner of the court near me; all seven shots are in within acceptable focus (most are dead on).
Resolution, Diffraction, and To E or Not to E
Here's the thing: certainly when we were at 12mp and lower we were living in a sort of Disneyesque world where everything was slightly sharper than reality. What do I mean by that? Diffraction wasn't getting fully recorded or seen in most cases. A D3 at f/16 was just starting to show visible differences on edges at 100% view for most people (though diffraction was already present, it wasn't clearly destroying edges enough for people to get upset). Some of this has to do with the way Bayer sensors record data. I've been saying for a long time that diffraction really only starts to be fully recorded by a Bayer camera when the Airy disc becomes about twice the size of an individual photosite. It's not a perfect predictor, since there's an optical system that sits above the photosite (AA/IR filter, which may have a waveplate in it, microlenses on the sensor itself). But it's been a "good enough" predictor for some time now.
So what do we see on the D800 and D800E? At and above f/8 diffraction is being fully recorded (at f/8 the Airy disc diameter is 10.7 microns, while the D800 sensor photosite implied diameter is a bit less than 5 microns). Even at f/5.6 the Airy disc is big enough to be producing clearly visible diffraction.
Okay, so what about the AA (D800) or lack of an AA (D800E)? Does that make a difference. At f/8 and above, not really. Diffraction is a worse sin than anti-aliasing, at least when you use the "visible" criteria. Technically, I measure a bit more resolution on the E than the non-E in these mid-range diffracted apertures. But looking at pixel views of images, the diffraction kills the edge acuity that the E normally provides you. I'm not convinced there's enough gain to warrant the difference if you're shooting at f/5.6 or above all the time. That shouldn't surprise anyone, considering that I concluded the same thing with a D3x (24mp) with and without an AA filter. In that case, the "without" was optical glass, with no fuzz/defuzz system like the D800E has.
Below f/5.6, things are completely different, and surprisingly so. From f/1.4 to f/4 the D800E has crisp, clean edges and is clearly gaining something from the lack of an anti-aliasing filter. But the D800 is different. While the D800 has what I would characterize as a weak AA filter (I've got plenty of moire examples from it, and can produce color fringing with it, too), it exhibits a different pattern than the D800E. From f/1.4 to f/4 there's a small but steady degradation of edges, almost like some form of weak diffraction were in play. That's actually entirely possible, as Nikon claims that there is a waveplate involved in the AA filter, and it may be the culprit.
So basically the conventional guess about who would want an E and who would want a non-E are exactly backwards. If you shoot wide open or near wide open with your lenses all the time (portraits, wildlife, sports, etc.) there's something to be said for having the D800E. If you shoot landscapes and are going for depth of field, diffraction will be your real enemy, not the AA filter.
But how does 36mp resolve, Thom?
Incredibly well, thank you. Even with diffraction in play, I can measure clearly better resolution scores from my D800 than my D3x, which is exactly as expected. As I've written before: all else equal, I'll always take more pixels.
But this will be an eye-opener for some of you. Resolution not only reveals more detail, it can also reveal more about how your lens performs. A lot was written about how the D800 would out-resolve lenses. Get that out of your mind, because that's not what's happening. Your lenses are capable of resolving even more than the D800 models will manage. But along with that extra resolution comes the ability to actually resolve what the lenses are doing. Poor corners become very obviously poor. Edge to edge sharpness differences (miscentered elements, etc.) become more obvious, especially on a D800E at or near maximum aperture. Chromatic aberrations now encompass more pixels on edges, so often become more visible at pixel level, too. Be prepared to see how your lens actually performs, at least if you're a pixel peeper or printing big.
As it turns out, that list that Nikon had in their Technical Guide for the D800 turns out to be basically right: the modern zooms (f/2.8, f/4 max apertures), most of the recent fast primes (f/1.4, f/1.8), the Micro-Nikkors, the PC-E lenses, and the exotic telephotos (200mm f/2, 300mm f/2.8, etc.) are all very good on a D800 model. Used with discipline and given a bit of post processing clean-up, they can be stunning.
Drop down to the next level of Nikkors--the 28-300mm f/3.5-5.6 comes to mind--and the results aren't quite so stunning. Heck, that lens is diffraction impacted above about 135mm, even shooting wide open, so this really shouldn't surprise anyone. That doesn't mean I wouldn't use that lens on a D800 model, only that my expectations would be that I won't get as much out of the camera as I would with even the 24-120mm f/4 in the overlapping focal lengths. Lenses that you can stop down to hit f/4 tend to do even better, such as the f/2.8 zoom trio.
Lest you get the wrong idea, I'm quite happy with the results out of the D800 (not E) model at f/8 with most of my lenses. Edge acuity is masked by the AA filter and diffraction, and lessor lenses definitely are struggling to perform, but the results can still blow away where we were with 6mp cameras a few years back. What I'm struggling to come to grips with still is where f/11 stands on the D800 models.
To put that in perspective, f/11 was as far as Galen Rowell would typically go with his Nikon 35mm film bodies. His careful analysis told him that the DOF he gained beyond that would be destroyed by the diffraction and lens under performance past that point (most lenses have peak performance one, two, maybe three stops below their maximum aperture). Not all film shooters agreed with him, but I came round to his position the better I got with my own shooting: I wasn't gaining anything and was starting to lose important things at f/16.
I suspect we're fully back where we were with 35mm film now. Indeed, those that say you can get a good 5000 dpi out of slide film with a drum scanner should note that the D800 sensor is within a few pixels of 5000 dpi, but it won't be resolving grain ;~). (I don't really agree with that 5000 dpi statement; I came to believe that 4000 dpi was about the limit where you got usable information out of 35mm slide film.)
So bottom line about resolution is this: 35mm film shooters that were happy with that should be happier with a D800 model. If that's not enough for you, then you're a candidate for medium format and always have been. But all that resolution (and edge clarity on the D800E up through f/4) doesn't come without a price: you simply won't be happy without the best possible lenses and technique if you're worried about resolution to this degree. You're a candidate for some of the better Zeiss manual focus lenses, maybe the PC-Es if you know how to use them, and only a handful of other lenses.
The rest of the world? They'll be happier than they were, even with their not-quite-so-great lenses. I was shooting with the new 24-85mm VR today on the D800E, and while I'd still prefer many of my other lenses, I can live with the results from this low cost new entrant, especially if you're not overly concerned about the absolute corners.
Uncompressed video out the HDMI port is as good as I've seen from any DSLR, despite the sub-sampling issues. If you can keep your focus field tight on a subject that isn't producing artifacts, the video can be stunning, even though it's really only 8-bit 4.2.2 (we really want 10-bit 4:4:4 to get video up to the levels of "raw" that we get from stills). Clean, clear, sharp (sometimes too much so), and if you're paying attention, even broadcast approved (make sure HDMI/Advanced/Output range is set to Limited range and the video stream will stay within broadcast IRE limits for white and black).
I'm not sure why we don't have a 720P/24 mode (we have 1080P/24). That seems like something that should have been there to be complete. PAL users will like the 25/50 options.
Audio is another story. The inability to adjust sound while recording coupled with the poor internal amp means that I'm happier recording audio externally and combining it with the video in post (e.g. FCP X sync clip, PluralEyes, etc.). If you really have to record audio in camera, get a BeachTek or other outboard mixer, and do some trial and error to see what in-camera recording level works best for you (if you're monitoring and changing audio levels correctly on the external mixer you don't want Auto set on the camera's audio).
The good news is that I've now shot several times with the D800 in Live View dumping to external HDMI video recorders. Even in three-hour long sessions I've yet to have the camera overheat and stop. Of course, this was indoors. Not sure what will happen when I go outdoors in a heat wave and try the same thing.
There, I said it: the D800 and D800E are demanding cameras. If you're buying them for image quality (and I'm not sure why you'd be buying them otherwise), the possible output can be spectacular, but getting there takes work. I suspect the Big Croppers will poo-poo that. They'll think that they can just treat the D800 like a point and shoot and then crop to get the picture they want. They'll be disappointed, as diffraction and sloppy camera handling can make those 12mp crops look worse than if you had just used a 12mp camera with the right lens in the first place. As I noted earlier, if you're an aggressive cropper, the D7000 is a perfectly fine choice for you, as it's basically the same thing at 1.5x crop, much less expensive and thus you're more likely to be able to afford the right lens in the first place.
Personally, the D800 and D800E are near perfect for me and the types of things I shoot (nature, wildlife, landscapes). It's the first time I've felt 100% back to where I was with an F100 (film body). Indeed, I'm ahead of the game.
Should You Get a D800?
Another way to look at it would be this way:
To a large degree, it's true. I'm looking forward to returning to Africa and Galapagos and Patagonia and Alaska and more with my D800 cameras. But it's not a perfect statement, as there are things I shot with the D3 and D3s that I would want to reshoot with a D4, maybe, but wouldn't want to reshoot with a D800 model. In that category we have a lot of sports work, for example. The 4 fps limitation, as much as I don't generally use burst shooting, is still a limitation, after all. Still, the headline of the ad is an accurate statement for most shooters, I think.