Nikon D3 Image Quality

Saturation & Hue Accuracy
Slight oversaturation of strong red and blue tones and some greens, but better than average accuracy and pleasing color overall.

Saturation. The Nikon D3 oversaturates reds, blues and some greens a little. It also undersaturates some green-cyan colors very slightly, but results are still pleasing. Some may find its default saturation a little high for a professional SLR (10.3% oversaturation, compared to only 5.2% for the Nikon D2Xs), but this seems to be the trend for Nikon these days, away from the technically more accurate but slightly dull looking saturation levels of the past. A firmware upgrade is available to mimic the D2Xs' color response, and as you'll see below, the D3's saturation adjustment gives good control, if you'd like a bit more subdued or even brighter color. Most consumer digital cameras produce color that's more highly saturated (more intense) than found in the original subjects. This is simply because most people like their color a bit brighter than life. Professional cameras tend more toward technically accurate color rendering.

Skin tones. In this case, the D3 did render skin tones slightly on the pink side in most cases. Still, results are quite reasonable, well within an acceptable range. (And here again, many people prefer a slightly ruddy rendering of Caucasian skin tones, often reacting to technically neutral renderings as being "washed out." Here again also, the Nikon D3's saturation adjustment can knock down this mild over-saturation quite effectively.) Where oversaturation is most problematic is on Caucasian skin tones, as it's very easy for these "memory colors" to be seen as too bright, too pink, too yellow, etc.

Hue. Overall results here are pretty accurate, though some reds are pushed slightly toward orange, oranges towards yellow and cyan towards blue. Still, overall color is quite good, with an average hue error after correction for saturation variation of only 5.2 delta-C units. This is closer to technically accurate than many DSLRs on the market. The other important part of color rendition is hue accuracy. Hue is "what color" the color is.

Saturation Adjustment
The Nikon D3 lets you adjust the image saturation, contrast, and sharpness in seven steps each. As can be seen below, the saturation adjustment worked very well, providing a reasonably fine-grained adjustment over a useful range of control. (Although, personally, we'd like to see slightly smaller steps: For our own tastes, we find only the first couple of steps below the default and perhaps only the first step above default saturation photographically useful. It'd be nice to have more steps of adjustment covering that range, to provide finer control.) The saturation adjustment also has almost no impact on contrast. That's how it should work, but we've often found interactions between saturation and contrast (and vice versa) on the cameras we test.

Exposure and White Balance

Indoors, incandescent lighting
Auto and Incandescent white balance settings were quite warm under our household incandescent bulbs, but Manual white balance setting produced very good color; slightly less than average positive exposure compensation required.

Color balance indoors under incandescent lighting was very warm with the Auto white balance setting, with a strong yellow cast. - A little disappointing that a $5,000 camera can't do a better job with auto white balance, but it's unfortunately fairly common. The Incandescent setting was a bit better, but still quite warm, which is to be expected on a professional SLR in this setting: Pro studio lighting is color-balanced to 3,200K, while the household incandescents used here are much warmer-hued, typically falling in the range of 2,600 - 2,800K. The 2,700 Kelvin setting produced a very slight magenta cast, which some might actually prefer as being more evocative of the mood of the original lighting. I was a little torn between the 2,700K and Manual results, but ultimately found the Manual setting a bit more pleasing overall. The Nikon D3 required slightly less than the average amount of positive exposure compensation here, at +0.7 EV. Overall color is quite good, Marti's white shirt is almost perfectly neutral, although her face is a little pink, and the blue flowers looked purplish as they often do with this shot. Our test lighting for this shot is a mixture of 60 and 100 watt household incandescent bulbs, a very yellow light source, but a very common one in typical home settings here in the U.S.

Outdoors, daylight
Good color overall, though a tendency toward a slight warm cast. Slightly high contrast under harsh lighting, but really excellent highlight/shadow detail preservation. Slightly better than average exposure accuracy outdoors.

In the "Sunlit" test shot above, the Nikon D3 required less than the normal amount of exposure compensation, and did a really excellent job of preserving detail in both the strong highlights of Marti's shirt and the deep shadows under the bouquet. The image is a little contrasty with the default settings, making the shadows very dark, but the D3 did better with this shot than the vast majority of cameras we've tested.

The outdoor house shot is just slightly overexposed at default exposure, but with natural looking color. The harsh glare from the white paint here, combined with the low sun angle (this shot was captured in mid-winter) are a bit too much for it to handle, but see the "Extremes" section of this report below, for more detail on highlight and shadow detail handling, including the D3's very effective Contrast adjustment, and its Active D-Lighting feature.

See full set of test images with explanations
See thumbnails of all test and gallery images

Resolution
Very high resolution, 1,700 lines of strong detail.

Our laboratory resolution chart revealed sharp, distinct line patterns down to about 1,700 lines per picture height in both the horizontal and vertical directions, with extinction past 2,000.

Important note: We've recently begun shooting all our DSLR resolution and laboratory test target shots with Sigma 70mm f/2.8 macro lenses. These lenses come in flavors for all major lens mounts except the FourThirds system, and are arguably the sharpest lenses we've tested to date over on SLRgear. In the case of the D3 however, the Sigma 70mm f/2.8 Macro inexplicably produced incorrect exposures, leaving us uncertain what aperture the lens was actually using on any given shot. This was very strange, as we've never found any exposure issues with this (or any other Sigma lens) before, and the same lens performed flawlessly in our testing of the D300. To avoid these exposure problems, we used our Nikon 105mm f/2.8 Macro lens for the D3's test shots. The 105mm f/2.8 Nikkor is also an exceptionally sharp, low-distortion lens, so its results here should be all but indistinguishable from those we might have obtained with the Sigma 70mm. (Particularly shooting at the f/8 aperture we're using in the shots above.) We call attention to the lens used only because it differs from our current standard.

Use these numbers to compare with other cameras of similar resolution, or use them to see just what higher resolution can mean in terms of potential detail. Our interpretation of this standard is somewhat conservative. We watch for artifacts and color fringing then move back to the nearest pure part of the scale. In our opinion, detail with artifacts shouldn't be considered detail. You may see other numbers quoted elsewhere, but across the site, our reviews judge this parameter by the same conservative standard.

See full set of test images with explanations
See thumbnails of all test and gallery images

Sharpness & Detail
Sharp images overall, with excellent detail. Minimal noise suppression visible in the deep shadows at low ISOs.

Sharpness. The Nikon D3 captured fairly sharp images, though the very conservative default in-camera sharpening leaves details in camera-rendered JPEGs shot with the default settings looking just slightly soft overall. The good part of this is that there's very little in the way of sharpening artifacts left to interfere with subsequent sharpening on the computer. In the high contrast shot above, there's only a hint of edge enhancement artifact visible along the edges of the white house trim and roof, as well as some of the larger branches. There are some minor artifacts from the in-camera sharpening though: Sharpening on the computer does tend to bring out the very slight halos that are present in the original images, but generally not noticeable unless additional sharpening is applied. (Edge enhancement creates the illusion of sharpness by enhancing colors and tones right at the edge of a rapid transition in color or tone.)

Detail & Noise suppression. The crop above right shows very little evidence of noise suppression in the shadows, with quite a bit of fine detail in the strands of hair visible. Noise-suppression systems in digital cameras tend to flatten-out detail in areas of subtle contrast. The effects can often be seen in shots of human hair, where the individual strands are lost and an almost "watercolor" look appears.

JPEG vs RAW

Mouse over the links in the box above to compare the difference in sharpness and detail from camera JPEG versus a 14-bit RAW file processed with Nikon's Capture NX version 1.3.0 and Adobe Camera Raw 4.3. Camera settings for the JPEG settings were the defaults.

In our testing of many DSLRs, we're finding that there is often more detail locked up a the camera's RAW files than makes it out in the camera-produced JPEGs. This once again seems to be the case for the D3. It's in-camera JPEG processing is actually pretty competent, but you can nonetheless produce a much sharper, more finely-rendered image by manipulating the RAW files in a good third-party RAW converter.

The camera's own JPEG processing and that of Nikon's Capture NX software produce very similar-looking images (no surprise there), although Capture NX gives you more control over image-sharpening parameters. The size of the sharpening operator in Capture NX seems to be somewhat constrained though, preventing you from using as tight an operator as you might like. As a result, finer detail can be revealed by processing in Photoshop, importing via Camera Raw with its sharpening turned off and then applying ~250% unsharp masking with a radius of 0.3 pixels once the image has been imported. (We've generally found that careful unsharp masking delivers better results than we can achieve using the sharpening built into Camera Raw.)

ISO & Noise Performance
Very low noise up to ISO 1,600 (!) sensitivity, moderate noise at ISOs 3,200 and 6,400, high (but fine-grained) noise and strong blurring at ISOs 12,800 and 25,600.

Noise levels are remarkably low at the Nikon D3's lower sensitivity settings, with buttery smooth images up to ISO 800, where there is only a slight hint of "grain" visible in darker shadows. At ISO 1600, noise is still very low with very little loss in fine detail. We start to see some minor loss of detail at ISO 3,200 where high ISO noise reduction is smoothing away fine detail along with noisy pixels. Noise and the effects of noise reduction becomes obvious at 6,400, with stronger blurring, and increased "grain", but chroma noise is kept in check and overall results are still excellent for a 12 megapixel sensor at such a high sensitivity setting. (The D3's big pixels really help with light sensitivity and low noise, and Nikon's sensor technology and noise reduction appear to be top notch as well.) ISOs 12,800 and especially 25,600 are quite noisy and soft looking, with blotches of chroma noise and little fine detail remaining after default high ISO noise reduction is applied. That said though, even ISO 12,800 images look surprisingly usable when printed as large as 8x10 inches, and ISO 25,600 ones looked surprisingly good at 5x7.

Bottom line, the D3 has by far the best high-ISO noise performance of any DSLR we've tested to date.

See the High ISO Noise Reduction page though, to see how the D3's various high ISO NR settings impact image quality.

Single-Pixel Faults?

Eagle-eyed IR staffer Zig Weidelich sniffed these out: There appear to be at least a few single-pixel defects in our D3's sensor. They seem most apparent in higher-ISO shots, but some can be found even at relatively low ISO settings. Here's an example of what one looks like, cropped from Marti's hand in the Indoor Portrait ISO 400 shot, enlarged 400%. It looks like it's probably a single hot pixel, affecting the interpolated data of its four nearest-neighbor pixels. This particular artifact showed up at all ISO settings above 100.

Having spotted it in the Indoor Portrait shot, it was easy to find it in many of our other shots as well, at pixel coordinate x=2,108, y=2,726. For whatever reason though, it didn't always appear as the cross-shaped artifact seen above: In some of our lowlight shots, it appears as only a short horizontal line. Whatever the cause of this sort of defect, they can be hard to find (which is to say that they don't seem to affect images too much), as it seems to take just the right combination of subject and exposure to reveal them. (As witness how much less evident the defect above is when seen against the background of the DaveBox in the low-light shots.)

We only found a couple of these in our shots, but there could be more we haven't found yet: As noted, they're hard to find, and for whatever reason don't seem to correlate well with the typical "hot pixel" errors in long exposures: These artifacts do show up as hot pixels in very long exposures (a couple of minutes, say), but other pixels that appear "hot" under long-exposure conditions don't show up in shorter exposure they way these do. Having brought this to light, I'm sure the online community will sniff out more examples of them, so we'll perhaps be able to characterize better how and when they appear.

High-ISO comparison: Nikon D3 vs Canon 1Ds Mark III
Based on reader emails and various forum posts, this seems to be a comparison many people are interested in making. It strikes us as a little bit of comparing apples and oranges though: The two cameras are made to very different design specifications, the D3 being optimized for high-ISO work, the 1Ds Mark III for high resolution at lower ISOs. Nonetheless, there seems to be great interest in comparing images from the two cameras, so we'll accommodate that interest here.

In the crops below, we've chosen to not resample the images, despite the very different resolution levels of the two cameras. Resampling either the 1Ds Mark III image down or the D3 image up introduces another variable, namely the type of resampling done, and what type of sharpening is applied to either image after the resampling. (Note that some sharpening is required in that situation, just to keep the comparison as fair as possible: Even downsampled images will require sharpening, as the downsampling will partially or fully negate the effect of any sharpening that has been done in-camera.)

So, we present the results below with the caveat that there's inherently a large component of apples vs oranges present here. That said, the results aren't too surprising: The D3's larger pixels show lower noise but also lower resolution than the smaller sensor elements of the 1Ds Mark III.

The comparison table above shows crops from the Canon EOS 1Ds Mark III next to some from the D3, comparing the Off and On settings of the 1Ds Mark III to the NR Off and Normal settings on the D3. Apart from the obvious differences in resolution, saturation and tone curves, we can see the D3 exhibits less noise even with NR off than the 1Ds Mark III has with its high ISO NR on (which does a good job at removing chroma noise, but does very little to remove luminance noise). No real surprise here, since the Mark III's sensor has nearly twice as many pixels as the D3's.

High-ISO comparison: Nikon D3 vs Nikon D2Xs
This is perhaps a bit more fair comparison, in that both cameras at least have the same pixel count, and the D3 outperforms the D2Xs in terms of shooting speed. (So, other than the different crop factors, the D3 could easily substitute for most any application the D2Xs could have been used in.)

The comparison table above shows crops from the Nikon D2Xs next to some from the D3, comparing the Normal high ISO NR setting of the D2Xs (sorry, we didn't shoot any with high ISO NR off), to the Off and Normal settings on the D3. Here we see again, apart from the differences in color and saturation, just how much cleaner the D3 is at higher ISOs. Keep in mind the D3 has a full-frame 36 x 24mm sensor, while the D2Xs packs a similar number of pixels into a DX format (23.7 x 15.7mm) sensor, which is less than half of the area of the D3's. As noted above though, the very fast shooting speeds of the D3 make it an even better choice than the D2Xs for sports and other fast-paced applications, the only downside being that you won't get quite the reach out of your long teles as you would with the D2Xs.

Extremes: Sunlit and low light tests
High resolution with excellent detail in both the highlights and shadows.

This camera sees better in the dark than you do! - Excellent low-light performance, capable of capturing bright images under average city street lighting and much darker conditions.

Sunlight. The Nikon D3 performed well under the deliberately harsh lighting in the test above, and captured bright midtones without sacrificing too much detail in the highlights or shadows. Though Marti's face still looks a little dark at +0.3 EV, I preferred it to the image at +0.7 EV, which had too many blown highlights for my preference. That said though, even the +0.7 EV example did an excellent job of preserving highlight detail, in that only a few areas were completely blown out. Shadow detail was also really excellent, there's very good detail there, surprisingly far into the deep shadows. (In "real life" though, be sure to use fill flash in situations like the one shown above; it's better to shoot in the shade when possible.)

The series of shots above show the results of the different contrast settings, all shots captured at an exposure setting of +0.7 EV. While it can be difficult to evaluate small differences in contrast on small thumbnails like these, it's pretty easy to see the impact of the Contrast adjustment in the images above. At its lower settings, the D3 did a really excellent job of handling the deliberately horrific lighting of this shot.

With the contrast dialed down to its lowest setting, even the +0.7EV base exposure used above results in almost no loss of highlight detail (only a spot on Marti's collar and the red channel in a few spots on the red flowers are blown out), and the shadows are unusually open. Color saturation is also almost entirely unaffected by the contrast adjustments, as it should be.

The series of shots above show the results of the available Active D-Lighting settings (once again, using a nominal exposure compensation of +0.7 EV), used to preserve highlight and shadow detail in high contrast images. This is different than the touch-up menu's D-Lighting, as it is performed during image capture instead of after. (It does affect only JPEG images though, Nikon very properly leaves RAW file data strictly as it comes from the sensor.) The camera's default setting is to have Active D-Lighting applied at the Normal level.

The results here are very interesting: At the highest D-Lighting setting, highlight and shadow detail is as about as well preserved as with the lowest contrast setting, but midtone values are richer, not as flattened-out looking as they are when the overall contrast is dialed down. See the images below for a closer comparison between the effect of low contrast and Active D-Lighting. When you really look at the photos here, the difference in highlight detail in Marti's shirt here, from Active D-Lighting off to the High setting is pretty dramatic, it's a feature that seems to work very well.

We had some discussion around the office about the advisability of leaving a feature like Active D-Lighting enabled by default. This is something that's going to be very much a matter of personal preference: Some will love it, others may just dismiss it out of hand. I (Dave) confess to having had some uneasiness about a camera mucking with exposure characteristics to this extent by default. That was before I shot with the D3 (and D300, which has the same feature) though. In practice, I always preferred the shots with Active D-Lighting enabled to those with it off. Unless I were deliberately trying for a high-contrast image (a silhouette, say), I don't think I'd ever have cause to disable it. I wouldn't have said this about Nikons first implementations of D-Lighting in its earlier SLRs, but I'm very impressed with how well it seems to work in this latest incarnation. (And of course, the RAW files are always untouched, so as long as I shoot RAW+JPEG, I always have the option of going back and extracting a JPEG from the RAW with D-Lighting turned off.)

Here's a couple of side by side examples showing the effects of Active D-Lighting vs using the contrast adjustment to handle high dynamic range situations. The lowest contrast setting typically pulls the highlights and shadows further in that does Active D-Lighting, but the result is a much flatter-looking image overall. Active D-Lighting manages to concentrate its action in just the strongest highlight and shadow areas, leaving the midtones more or less alone. The result is generally a more vibrant, natural-looking image.

Note: Because digital cameras are more like slide film than negative film (in that they tend to have a more limited tonal range), we test them in the harshest situations to see how they handle scenes with bright highlights and dark shadows, as well as what kind of sensitivity they have in low light. The shot above is designed to mimic the very harsh, contrasty effect of direct noonday sunlight, a very tough challenge for most digital cameras. (You can read details of this test here.)

Low light. The Nikon D3 performed extremely well here, able to capture usable images down to the 1/16 foot-candle light level (about 1/16 as bright as average city street lighting at night), at all ISO settings. As seen above, color was very good with the D3's custom white balance setting, but its auto white balance setting unaccountably struggled with this shot, producing pronounced blue tints.

Noise is almost nonexistent up to ISO 800, and remains very low at ISO 1,600, and 3,200. Noise levels are moderate at ISO 6,400, but there's still plenty of detail to work with, especially when high ISO NR is set to "Off" (which still applies some minimal filtering to JPEGs at ISOs over 2,000 -- RAW files are always left untouched). As you would expect, noise is much higher at ISOs 12,800 and 25,600.

The D3 gives you 4 options for high ISO noise reduction: Off, Low, Normal and High, so you have some flexibility in deciding how much noise to trade for detail. Except for the "No NR" shots, these were all shot using the Normal NR setting, and Long Exposure NR was enabled, so was applied to exposures longer than one second. There are a few hot pixels visible at ISOs 12,800, but hundreds visible at ISO 25,600, and Long Exposure NR did not come into play, because the shutter speed was still relatively high. (Only 1/10 second at the lowest light level we test at!) With these ultra-high ISOs, Nikon should perhaps consider offering a way to adjust the Long Exposure NR shutter speed threshold in a future firmware revision.

The D3's autofocus system also performed well here, able to focus on the subject down to below the 1/16 foot-candle light level with its AF assist light turned off, and in total darkness with it enabled.

Bottom line, the Nikon D3 is clearly the best low-light performer we've seen to date, and by a significant margin. If you need to maintain high shutter speeds under dim lighting, the Nikon D3 is head and shoulders above the rest of the field in this respect.

(Note: Focusing -- These shots were taken with the Nikkor 24-70mm f/2.8 lens at 70mm and f/2.8. The depth of field was thus very shallow, so not all parts of the test box will be in focus at the same time.)

How bright is a foot-candle? The one foot-candle light level that this test begins at roughly corresponds to the brightness of typical city street-lighting at night. Cameras performing well at that should be able to snap good-looking photos of street-lit scenes.

NOTE: This low light test is conducted with a stationary subject, and the camera mounted on a sturdy tripod. Most digital cameras will fail miserably when faced with a moving subject in dim lighting.

Like most professional SLRs, the D3 does not have a built-in flash, so this section left blank intentionally.

Print Quality
Excellent print quality, great color, good 20x30 inch prints, excellent 13x19 inch ones. ISO 6,400 images are surprisingly good at 8x10, even better at 5x7.

Output from the Nikon D3 was good enough to produce good-looking 20x30 inch prints, and sharp 13x19 inch prints. At 20x30, its images were a little softer looking, but would be more than OK when seen at the viewing distances typical for such large prints. As noted elsewhere in this review, the D3 rewards RAW shooters with really excellent detail when its NEF files are processed through a good RAW converter. The difference between camera JPEGs and those from RAW isn't as dramatic as with some cameras we've tested lately, but the results are well worth the effort if you care about extracting every last bit of information from your images.

High ISO images were phenomenal, the D3 clearly leads the field in high-ISO performance, thanks to its large pixels, CMOS sensor technology, and Nikon's excellent noise-reduction processing. D3 images shot under incandescent lighting (always the tougher test) looked great when printed at 8x10 inches, all the way up to ISO 6400. At ISO 6400 and that size, there was just a little noise present, but we had to look close to see it (closer than you'd normally view a print of that size), and it was very fine-grained. - There's also almost no chroma component to the D3's noise at ISO 6400, making it even less apparent than similarly-grained noise patterns from many other cameras. Shot under daylight-balanced lighting and printed at 13x19 inches, the D3's ISO 6,400 shots were softer and somewhat noisier, but the results were still little short of amazing.

As you might expect, the Nikon D3's noise processing at high ISOs varies quite a bit, depending on the setting you're using. At the low setting (our personal preference), a little fine-grained noise creeps in at ISO 3,200, but it's pretty minimal, and fine subject detail is preserved very well. The Normal noise reduction setting almost entirely eliminates the noise (even at ISO 3,200), but loses a little fine detail in the process. The High setting leaves images very clean, but also very soft, with much more detail lost. That said though, the underlying images are so clean to begin with that the amount of detail loss from any given level of noise reduction is much less than we're accustomed to seeing. (In practice, for print sizes 8x10 or smaller, leaving high-ISO NR turned entirely off produces prints with excellent detail and surprisingly little noise up to ISO 6,400.) As noted, the Low setting was our favorite, it produced surprisingly clean images that still contained loads of fine detail.

Color-wise, the Nikon D3 did very well. At its default settings, its color (particularly reds) was a bit more saturated looking than that of the last generation of Nikon DSLRs, but we think it'd be in an acceptable range for most shooters. Some orange, and some blue/purple shades in particular were brighter than the D2Xs' handling of those colors, but the overall effect was pretty pleasing. (If you like less-saturated color, just take the saturation adjustment down a notch.) Hue accuracy was good, almost identical to that of the D2Xs. A very good performance overall. (Note that since the color error plots for the D2Xs were generated, Imatest has incorporated new, more accurate reference values for the color coordinates of the Gretag MacBeth ColorChecker. You'll thus see a lower mean color error in the Imatest plot for the D2Xs than for the D3, but in actuality, the two cameras saturation-corrected hue error is virtually identical.)

Testing hundreds of digital cameras, we've found that you can only tell just so much about a camera's image quality by viewing its images on-screen. Ultimately, there's no substitute for printing a lot of images and examining them closely. For this reason, we now routinely print sample images from the cameras we test on our Canon Pro9000 studio printer, and on the Canon iP5200 here in the office. (See the Canon Pro9000 review for details on that model.)