Are you ready for some pixels?
Small updates: 2/8/2012
Support this site by pre-ordering the D800 from the following vendor:
If Nikon wanted to get some press, 36mp for US$3000 sure did it. It wasn't that long ago that we were paying US$20,000 for medium format backs that did that, now suddenly we have a bombshell from usually staid Nikon. Remember, it also wasn't that long ago that people were complaining about "only 4mp? or only 12mp?). I guess Nikon got tired of hearing those (naive) complaints. Now they're going to hear "why 36mp?" ;~)
Let's back up for a moment. Last year was supposed to be a big one for Nikon. R&D had prepared it this way: D5100, Nikon 1, D4, D400, D800. Coupled with the late 2010 turnovers of the D3100 and D7000, this was supposed to produce a complete lineup change in a year. Eight interchangeable lens cameras in 12 months. Moreover, the strategy was very calculated: faster focus than competitors in the mirrorless category, more pixels than Canon at the prosumer level, better low light performance at the top of the pro line. In other words, lots of different forward thrusts to give Nikon bragging rights of some kind virtually across the lineup. By moving D800 production to Thailand Nikon thought they'd be able to finally get past their reputation for slowish deliveries on new products, too.
The goal of all that was to pass Canon as the leading interchangeable lens camera maker, or at least catch them.
Unfortunately, Mother Nature got in Nikon's way, and with a vengeance. Quake, tsunami, floods. Other than the Nikon 1 initiative, pretty much everything got derailed. Instead of having a brand-spanking new eight camera lineup for Christmas, they had two new low-end cameras (J1 and V1), not enough D5100 and D7000 bodies to go around, and three unannounced cameras.
That said, they still look they're managing the turnover whole thing in about 18 months (only one camera left unannounced at this point). As I wrote on the front page of my site in early February, Nikon did a remarkable thing: basically replacing their entire DSLR production capacity from the concrete floor of the plants up while still managing to increase their overall DSLR sales. One has to imagine what they could have accomplished on their original plan.
So here we have a D800 to contemplate. 36mp. 4 fps. 100% viewfinder. The D4's new metering and focus improvements. The D4's video capabilities. The D4's 3.2" display. EXPEED 3. Even a relative of the D4's shutter (tested to 200k cycles). All of the D700's abilities, too, including the pop-up flash, and a (new) extended hand/battery grip.
Some of the little things that you might want to know (yes, yes, I'll get to the sensor eventually):
- Body: The D800 follows the D700 style, but with features lots of small tweaks.
- On the front we've got the D7000-style autofocus control (which I think is a plus, as you can now change all critical AF settings without taking your eye from the viewfinder), but not much more. The curves are different and the prism profile different.
- On the back, we've got one primary change, and it's a doozy: the zoom in and zoom out buttons have been flipped. Yeah, that's not going to cause a few people upgrading any frustration, is it? The autofocus method selector is no longer necessary (due to the change on the front), and is replaced by the Still/Video/Live View control. Other than that, everything is much as you'd expect, though the LCD has been slightly upgraded.
- Up top we get a bracketing button and a Record Movie button (though it's location seems a little suspect).
- Overall, the body drops over 4 ounces (95g) of weight and is sealed about the same as the D700 was.
- The battery is the EN-EL15, same as used in the D7000 and V1. While some may lament the passing of the EN-EL3e, this is a reasonable choice by Nikon, and for me, at least, means I'll be using the same battery on pretty much everything I shoot with that has a Nikon label on it, except for the D4. That's a nice change. Now if we could only get a multiple battery charger.
- Speed: We get 4 fps (6 fps with the MB-D12 grip and at DX sizes with the right batteries). Like the D700, the D800 isn't going to be a speed camera, especially with those large files to write. Still, that's a reasonable speed I can live with. The Compact Flash slot supports UDMA-7 compliant cards, which should net noticeable improvement over the original UDMA cards that the D700 supported. Nikon claims that the camera can hit 100MBs+ transfer speeds, though such claims usually turn out to be maximum speed, not throughput. As a second slot, we get an SD slot that is UHS-I capable (95MBs). Buffer capacity for FX shooting is 16 NEF at worst compression settings (25 at best), 16 for TIFF, and 56 for JPEG fine large (100 for other JPEG settings).
- Subsystems: The 51-point autofocus module should operate better in low light (EV -2 rated), and you can now focus with maximum aperture of f/8 instead of f/5.6 (with reduced number of points). The metering is now performed by a 91k pixel sensor and 3D Color matrix metering is improved. The extra pixels in the meter give the D800 the ability to add face detection in phase detect and contrast autofocus. The shutter and mirror mechanisms are revised and rated to 200k cycles (D4 is 400k). The sensor gets an auto-cleaning feature. Almost everything in the subsystem realm is (welcome) tweaking, with the added information from the metering sensor providing much of the improvements.
- Video: The D800 is an impressive video camera. It needs to be in order to compete with the 5DIII. 1080P/30 and 720P/60 maximums may be a bit of a disappointment to some, but the rest of the changes are all basically pro-grade: headphone and audio level monitoring, HDMI outputs clean uncompressed video (and rear LCD still active during HDMI output), better interframe compression, ability to trigger video via the 10-pin remote, sophisticated noise reduction in real time during video recording, full manual control of shutter speed/aperture/ISO, smooth aperture changes (when using HDMI out), built-in time lapse movie making, and index marking. The D800 allows FX or DX crop video, but not the 2.7x crop that the D4 does. You can capture 1920x1080 stills while shooting video.
- Connectivity: Strangely, the D800 gets USB 3.0 while the D4 got USB 2.0 and Ethernet. That seems like a missed call on someone's part. USB 3.0, of course, doesn't currently give Mac users much to get excited about (though I suppose if you've got a Thunderbolt-equipped Mac and have ponied up for a fancy Thunderbolt port extender you might be happy).
- New stuff: The D800 gets the D5100's two-frame JPEG HDR capability. The Protect button gets overloaded some more: if nothing is displayed on the color LCD it becomes a Picture Control button. Auto ISO now can be set to "automatic inverse of the focal length." IPTC data fields can be entered in camera. The mirror delay is programmable, out to 3 seconds. In Live View, the mirror stays up in Tripod Mode. (As much as all this is good, it barely makes a dent in my User Suggestions list.) The new vertical grip is the MB-D12, and it can take either EN-EL15 (D7000, V1, D800) or EN-EL18 (D4) batteries. You need AA or D4 batteries in the grip to get 6 fps. Eye-Fi cards are officially supported. IPTC info can be entered. The Exposure Delay can be 1, 2, or 3 seconds.
Which brings us to the change everyone wants to know most about: the sensor. At 36mp, the pixel pitch is a bit less than 5 microns, which puts it in the relatively "small photosite" range. 36mp FX gives us about 15mp in the DX crop mode, making the D800 a D7000 equivalent, as well. That, more than anything else might tell you something: Nikon is hoping to entice D7000 as well as D700 owners to the D800. After all, the D7000 user shouldn't lose anything significant in the DX realm in doing so, but they gain an FX camera in the process. The D800 is thus a natural DX-to-FX migration path for a large number of users. Something that it looks like Nikon wants to encourage.
The sensor itself is a Sony EXMOR-derived one, which is mostly good news. Since the sensor has the ADC circuitry on-board and this results in low read noise and good linearity, we should get the strong shadow capabilities we've seen in cameras like the D7000 and NEX-7. On the other hand, adding all those power and data paths to service 36mp has to have taken some toll on light collection (fill factor). Microlenses can only deal with some of that loss.
So here's the question everyone has been asking me forever: how will the D800 perform in low light compared to a D700? My answer, if you recall, was to tell everyone to shoot with the D700 in DX mode, then shoot the same thing with a D7000 and downsize to 5.4mp. Did everyone do their homework? No one? Bueller?
There's good news in that test: the D7000 holds its own far higher into the upper ISO ranges than you might think. It's a good ISO 1600 camera to start with, but if you let me downsample from 16 to 5.4mp size, it's even better. So I'm expecting the D800 to be as good when downsampled to 12mp. Is that as good as a D700? In some ways yes, in some ways no. The pixel peepers are almost certainly going to notice noise at highest ISO values (just as they do on the D7000).
Also, the thing that is a bit unique about the D3/D3s/D700 sensor tech is that it is relatively friendly to highlights. More so than any other Nikon camera, I can recover more highlight detail, even if my exposure is a little hot (but not too hot). What's happening with the new sensors (J1/V1, D7000, D800) is that we're getting really good detail definition in the shadows (partly due to the on-board ADC and very low read noise). We're all finding an amazing level of detail down in the lower bits with these latest sensors. To a small degree, more so than with the D700. But at the highlights, not the same. So one thing everyone is going to have to adjust slightly is their exposure practices--highlight recovery is tighter on the D800 than on the D700. Thus, maximizing a D700 is a bit different than maximizing a D800 (okay, I'm theorizing here, as I don't have a D800 in my hands yet to test that assumption; but still, I'm pretty sure from everything I've heard and seen so far and the side evidence of other recent sensors that this will be the case).
The other thing that seems clear is that Nikon changed filtration from the D3/D3s/D700 sensors. Either that or the base silicon under the filters had something that goosed color channel performance (not likely, but I suppose physically possible). This one's going to have to wait for real testing, but one of the things I've noticed in the few side-by-side D3s/D4 image pairs I've been able to obtain is that the D4 doesn't have the same RB channel response as the D3s. I suspect this is making for the few contradictory comments about which is better in low light (low light can be blue shifted in outdoor dawn/dusk, red shifted in indoors incandescent lighting). But that's something I really have to test in a wide range of lighting conditions to make a meaningful comment on. So, I expect different RB channel response than the D700 from the D800, but not sure what that means yet.
The D800 throws a few other wrinkles into the pudding, too. (Wait, what kind of metaphor was that? The pudding I just ate was smooth ;~). The AA filter, for instance.
Leica doesn't use an AA filter on their 18mp full frame camera; Nikon uses one on their 24mp camera but will make it optional on their 36mp offering (the D800E, for an extra US$300). Obviously, the two companies have a differing opinion about where the point is that the AA filter really isn't adding anything useful to the image. In other words, at what frequency of information are you likely to trigger visible moire? You can trigger moire at any level, it's the word "visible" that's the key here. Frankly, I think we've long been past the point where it's a real issue other than for those that shoot fashion or industrial/architecture. My assistant shoots with a D3x without the AA, I shoot with a D3x with the AA. I haven't seen an image yet that has meaningful, visible moire out of my assistant's camera. But we're also not printing at over 24" and we also don't shoot things with absolutely repetitive patterns. Bird feathers don't really count here, for example, as there is a bit of randomness to their size and weave, so you don't get the huge areas of nasty moire that you get from, say, certain fabrics shot with certain lenses at a certain distance (a fashion shooter can always move and change the moire production, but that changes perspective, and they'd have to notice it being produced in the first place).
So which should you choose, AA or non-AA? Again, it's going to take testing to absolutely determine that answer. However, I don't think it makes a huge difference. You won't resolve more detail, you'll see faux detail beyond the Nyquist frequency which might be perceived in some situations as more detail. You'll get better edge acuity, which might help you with all the diffraction you're resolving if you're using smaller apertures. (Can you add an AA to the lens? Yes, you can. See www.caprockdev.com/antimoire.htm.)
My initial advice? If you're buying a D800 to print big (36" or larger) and you're shooting something that generally doesn't have large areas of exact repetitive patterns (e.g. landscapes), you'll probably prefer the D800E. Otherwise, I suspect you already have more pixels than you need with a D800 and downsizing will generally take care of the anti-aliasing.
The funny thing is that there seems to be a lot of brewing anger over the 36mp decision. I'm not sure why that is. Okay, sure, you might need to buy a few terabytes more drive space and a computer with more CPU cores and lots more memory (largest NEF size is about 75MBs and TIFFs are 108MB), but you're going to do that anyway ;~). If not this year, maybe next or the year after. So I don't see that as a huge issue. If you think you need to move forward with new gear in one part of tech (cameras), I don't see why you're averse to moving forward in another part of tech to support it (computers). It's a bit like asking for a Ferrari but not wanting to pay taxes to build a road that you can actually use it on.
Personally, I find a lot to like about the D800. It's got quite a bit of the D4's bones to it, and that's all good. I never expected blinding frames per second speed; 4 fps is just fine. In fact, if I could crank out 11 fps at 36mp for ten seconds at a time I think I'd need to have my head examined, as that would be almost a gigabyte a second). I didn't need 36mp, but I never turn down pixels, all else equal. So that's really the only outstanding issue for me: get a D800 in my hands so I can make sure that everything else is still nearly equal. Once I've proven that, I'll be very happy.
But this starts to bring up a question: whither the future? While I can imagine more (pixels, DR, speed, features, etc.), I can't imagine needing more. Assuming my assumption that downsampling to 12mp gives me a reasonable ISO 3200, what exactly would be missing from my DSLR? Oh, yeah: communicating, programmable, modular (CPM). Okay. Yeah, I'd upgrade for that. But what if those things aren't really coming and the D800 replacement is just a D900 that iterates the prosumer DSLR one more time? No, I don't need that.
As I've been writing for some time now, we've peaked with DSLRs as we currently know them. Couple that with slowing sales and in order to get most of us to move on from a D800 Nikon is going to have to pull something impressive out of their hat. 3-layer sensors. CPM. Something. Still for the next year or two, Nikon has an impressive lineup to sell the camera user, and the D800 is one of the most impressive of those.
The D800 will be available in late March for US$3000 (US$3300 without AA, which will ship in mid-April). You can pre-order yours and support this site simultaneously by using links at the top of this page.
Nikon's Microsite for the D800