With three days of shooting with a production-level D2H under our belts, it's too soon to make final pronouncements about the quality of photos produced by Nikon's latest news and sports digital SLR.
We have, however, spent about as much time poring over files from a production D2H, making prints from its 4MP files and exploring the range of Capture 4's NEF image adjustments as we have actually making pictures in the field.
This page contains only low-resolution photos; subsequent pages are comprised of full and reduced resolution photos from a production-level D2H body with release version firmware, with some comparison photos from the Canon EOS-1D, Nikon D1H and Nikon D1X as well. On this page you'll find some initial observations based on our examination of these and other production camera files.
Image quality generally matches or exceeds the D1X and D1H Saturation is up, detail is up, noise across the range of ISOs looks to be slightly worse while skin tones seem to be both up and down, depending on the subject and illumination.
The most noticeable change is the increase in saturation, evident particularly when comparing photos shot on the combo of Less Contrast Tone Compensation and Color Mode II. The boost is welcome. Files also show more detail per pixel than either the D1H or D1X and, like similar-resolution photos from the EOS-1D, they enlarge, sharpen and print well.
A lot of noise has been made about the noise from the D2H. Much of the hand-wringing has been prompted by the posting of camera store and trade show happy snaps, which have really demonstrated only that the D2H will faithfully record certain scenes in all their poorly-shot glory. Add that to what's discussed in the point a few paragraphs ahead, and what we've seen in photos captured in venues with which we're familiar, and we're left thinking that D2H noise is a wash: not as good as we'd hoped, but not terrible either.
While ISO 200 photos look pretty clean, thanks to Nikon's ballyhooed new noise suppression technologies, higher ISO photos shot in available darkness can get pretty noisy. Which makes it not unlike the D1H, the camera it replaces, and the Canon EOS-1D, its most direct competitor (though the EOS-1D's higher-ISO noise suppression appears superior and, ultimately, its files are cleaner).
Update, December 9, 2003: With a few weeks of D2H shooting to look back on, and carefully testing of three production D2H bodies in that time, our opinion of D2H noise levels has changed. While higher ISO files continue to look noisy but usable to us, D2H files shot in the ISO midrange of about 400-640 do show an unacceptable level of noise. Unacceptable, that is, for a sensitivity range that is well short of the camera's maximum. This noise is most visible in midtone-to-shadow transitions, dark suits and other similar areas. An upcoming article will look at this apparent problem more closely. For now, this image quality characteristic seems to be the one giving working news and sports photographers the most difficulty.
Colour: Our information from Nikon suggested that the colour appearance of D2H photos was going to be a slightly-tweaked, slightly-improved version of what had come before. This isn't so. While it's true that, given identical shooting and image processing settings, some scenes will look similar overall when shot with the D2H, D1X or D1H, this has been the exception rather than the rule.
Saturation is up all round, as mentioned, which is great. But the look of some scenes as rendered through Nikon's 3 different Color Mode settings, however, varies slightly to wildly in our testing, when compared to previous generation Nikon cameras. We had become accustomed to being able to predict the sort of hue, saturation and brightness differences that would occur as we flipped from Mode II to Mode I to Mode III. Not anymore - D2H files in some or even many cases respond differently. This isn't necessarily a bad thing, but it is different. What hasn't changed is our preference for Color Mode II, which is also the setting in which Nikon has bumped up the saturation the most.
Skin tone colour has been a mixed bag. It has been surprisingly good under some forms of artificial light, pretty good under strobes but outside under daylight, with and without fill-flash, we've seen a hint of the sort of ruddy skin tones and too-pink lips, tongues and noses that were typical of the original D1 in some situations. This unpleasant characteristic is completely absent in properly white balanced D1H and D1X photos, so we weren't expecting to see it again in a Nikon camera. We doubt this is going to evolve into Son of Magenta Cast - The Sequel, but it is a colour trait to be wary of in D2H photos.
You'll find examples of all three skin tone characteristics on subsequent pages. And no, this possible colour weakness isn't responsible for the colour shift in the hair shown below.
Production cameras are less noisy than preproduction cameras Okay, at least the preproduction cameras we had occasion to shoot with. The experience of some other preproduction camera testers we compared notes with suggests that this wasn't universal.
That said, the skateboarding photos below demonstrate what we believe was a significant contributing factor in the perception early on that the D2H made noisy pictures. At a glance, you should be able to see that the D2H frame on the left is about the same overall brightness and density as the D1H photo on the right. This is what you would expect from the same scene shot on the same settings with the same lens from the same position and so on. Except they weren't - the D2H frame is shot at ISO 800, the D1H frame at ISO 1250, but at the same shutter speed and aperture.
When we first noticed this we borrowed another preproduction D2H from Nikon. Same thing. To make a long story short, we compared a D1H and D1X to two D2H preproduction models, under three types of illumination, and got the same outcome: the two non-final D2H cameras were 1/2 to 2/3 stop more sensitive than their ISO ratings suggested. Put another way, ISO 800 on the early D2H's we tried was a lot closer to ISO 1250 in sensitivity and noise, with other ISO settings also showing the same differential. As a result, we're of the opinion that the preproduction D2H earned a bad noise rap in the early days because at least some of the bodies floating around were seriously out of whack in the way we're describing.
Our production D2H does not behave this way - it's only a hair brighter than the D1H - and is noticeably less noisy across the ISO range than preproduction models too.
Score one for Auto WB The D2H is the first digital SLR we've ever shot with to tame Calgary's McMahon Stadium. Balancing off a gray card at night football games has never been an option, because the colour of the light changes noticeably across the length of the field. Balancing for one location only makes the cast worse in another. Same goes for picking the closest manual WB setting. With most cameras we've simply thrown up our hands, cranked white balance to Auto and hoped for the best.
With varying results. The EOS-1D, for example, does a reasonable job, but tosses in sequences of blue-casted frames at semi-regular intervals. The D1H jumps around fairly continually, though it's often close to the mark. The D2H, by comparison, has delivered neutral colour across hundreds of frames, at least during the course of two night football games. This is great.
The flicker-detection capability, however, has not yielded better frame-to-frame consistency in either a local hockey arena or basketball gymnasium. These are venues that have given us fits in the past because of the colour shift brought about by their pulsating lights. Even when going so far as to hold the camera above my head, with its Ambience Light Sensor positioned under one of the flickering lights, the shift from frame to frame was about as evident as when the D2H was set to Preset WB under the same conditions.
Perhaps this isn't how Nikon envisioned the flicker-fixer function of the D2H's Auto WB would be used, but it is how we hoped it would help fight the colour battle. We've got our fingers crossed that other locations will yield better results.
Auto WB appears to not attempt to correct for the warm cast of bright tungsten pot lights or standard incandescent household illumination. We've even filled the frame with a gray card, to see if Auto WB would be moved to respond. Nothing. Part of the explanation probably lies in a specification for the D2H that has changed since the camera was first unveiled. Back in July 2003, Auto WB's range was announced as 2500K-10000K. Now, it has been revised to 3500K to 8000K. That puts typical tungsten and incandescent lamps outside its working range.
Note: For those keeping score, another image quality specification has also been changed. Long exposure noise reduction, when enabled, kicks in at shutter speeds of 1/2 second or longer. Originally, the threshold was 1 second.
Auto WB has long been a setting of last resort around here, regardless of camera model, simply because we've been able to get better and more predictable colour on another setting. With the D2H, we've found at least one situation - night football at McMahon - where it appears to be the ideal WB choice. Only time will tell if the considerable engineering effort Nikon has put into the D2H's Auto WB will make it the ideal WB choice in a variety of environments.
We're already sure that Auto Tone Compensation (ATC) is a hit. In D1-series cameras, ATC would simply choose between the Normal, Less Contrast and More Contrast curves based on its evaluation of the scene. The D2H version is essentially an automatic set-black function, done on a per-picture basis, even at 8 fps. We've shot everything to date with ATC, and then compared the result to a manual selection of Tone Compensation curve. In the vast majority of cases the effective contrast landed somewhere between Less Contrast and Normal, and in the vast majority of cases we liked the contrast decision the camera made.
Update, December 9, 2003: We've settled into a practice of using ATC when shooting in uncontrolled light - outdoors during the day, for example, as the light changes from soft to harsh. For controlled light shooting - covering indoor sports or shooting a portrait - a manual selection of Tone Compensation setting seems ideal, though ATC still comes awfully close to the mark in many cases.
Auto Sharpening - the jury is still out. We have noticed that Auto Sharpening applied in the camera to JPEGs is different than what's applied through Nikon Capture 4 to NEFs. The result is that NEFs processed through Capture 4 set to honour the Auto Sharpening setting will come out noticeably crisper than a JPEG out of the camera processed with Auto Sharpening on. Comparing the two is easy, thanks to the NEF+JPEG setting of the D2H. In fact, NEFs seem to be crisper across the range of sharpening settings, though this doesn't appear to point to an image quality advantage of NEF but rather a discrepancy in the sharpening applied in the camera vs Capture 4.0.