Back to the HomePage
This page, due to its heavy photographic content, is optimized for a screen size of at least 1000pixels in x-direction - please adjust your viewing screen accordingly!
NIKON D3 - NIKON D2Xs
How much better is Nikon's new flagship?
I have waited months for the opportunity to get both cameras not only for a couple of hours or one day but for at least one week to compare them to find an answer to the question - how much better is the new flagship, the Nikon D3 compared to the previous one, the D2Xs? Is it really better and if so, what is better? Is it the highly acclaimed new sensor, the new color management algorithms, the larger viewing screen, the overall performance, the wider range of ISO settings which is the main and most important update and key performance enhancement?
Well, for all those, who do not like to read more than a few lines of text - here my conclusion: Yes, the D3 is really better, much better than the Nikon D2Xs - not only because of the much larger high-resolution backside viewing screen or the wider range of ISO settings, it is the entire camera, which got even more attractive. I have reviewed the D2Xs some time ago and was rather enthusiastic and since I had the opportunity to work with the D3, I am sure that this DSLR is certainly the best one which Nikon every made.
Let me start with a bit more coordination - first the overall impression - weight, size and handling are pretty identical for both cameras - both are not lightweight and small but both are well balanced and you can hold them firmly and well with average sized (male) hands (- both cameras are pretty large and heavy for smaller hands) and most functions are there where an experienced Nikon user would want them to be found. Some functions which might be interesting for the advanced user, like sharpening, color rendition setting, color saturation etc. are buried under several layers of commands inside the user selectable functions - not too easy to find, but once you know where they are, you can find them again - but consider reading the manual (which is huge....) first before trying to find them without guidance.
One of the key features - at least according to the
published literature is the larger range of ISO settings - at the D3 from ISO
100 to 25600 (approx). The lowest setting ISO 100 is a one stop low setting to
the base setting of ISO 200 - nothing to complain about, but worth mentioning -
so no direct dialing for ISO 100. This is not a disadvantage, it is a different
approach to ISO settings. The main base setting of speed is an ISO 200
equivalent and if you need a one-stop lower sensitivity for a certain reason
(too much light??), then you can do so, but there is no visible advantage of ISO
100 over 200.
ISO settings at digital cameras are anyway not "correct" ISO settings, because by definition, ISO settings are referring to a certain amount of exposure density above the base fog of a film - they were not designed for electronic imaging, that is a reason why certain manufacturers call the settings ISO-equivalent and not ISO settings.
The ISO values above 200 can be dialed in directly until you reach 6400, which is the highest value which can be chosen. The two settings above 6400 are called Hi-1 and Hi-2 which means to me, that they are amplifier enhanced settings. And that is also how the images look like - yes those settings are producing images, but you get a lot of "electronic grain", noise at those settings. ISO 12800 is still useable under certain conditions but ISO 25600, well - OK; what do you expect at that speed? Try to push a film up to ISO 25600 and then see if you get any image which you can use - at least with the D3, there is still an image which is existing above a rather well visible background of noise. I must admit, despite those rather a bit negative comments, shooting in darkness with an exposure time of 1/500 sec is quite an experience....
Another highly acclaimed improvement is the TFT viewing screen - and it is really impressive. Large, detailed and bright - you can work with and on it, no problem. Not only large, but also much more details revealing - these are for me the main improvements of this new screen on the back side of the camera - it is approximately of the quality of a VGA screen with 920.000 dots on 3inch size with 100% frame coverage.
An improvement, which is not visible from the outside but which is very attractive and often also important, can be found by opening the cover which protects the digital storage media slot. Yes, there are two slots for two compact flash cards! Finally, we have got it - this is so important for many applications - not only to get more storage room but also to get some additional safety for shooting. These two slots can be used in several ways - independent from each other, consecutive and as back-up as well. A very attractive mode to use these two slots was for me to use one slot with a larger storage volume card for raw files and the other in simultaneous use for the JPEG files.
What else did I like from the very beginning, before I even took the first shots - well, the viewfinder. It is clean, bright, nice and large and free from unnecessary information dots and frames. One advantage of the full frame sensor - finally again a nice and large viewfinder! I might sound a bit conservative, but I prefer bright and clean viewfinders with a reasonably large viewing angle - exactly what the D3 has to offer.
There are several more features or enhancements (for me gimmicks), which are now incorporated into the D3, like the life-view (with and without tripod use) or the virtual horizon tool. The first one is of not much use for me - but this is my way of shooting only, for others it might be of significance, the latter one is nice but needs some improvements to be of real value. This virtual horizon tool is basically an electronic horizon (similar to the one on airplanes) to show if the camera is held horizontally or not or to adjust the tilting angle to a certain number of degrees. Well, fine but for me it would be more beneficial if it would not only work horizontally left-right but also forward-backwards - with that additional function, it would be easier to avoid, or at least reduce, perspective distortion caused by slight vertical front-backside tilting of the camera.
I have read in another pre-review that the power consumption is rather high compared to the available battery power - I cannot confirm this, but I am not shooting several hundreds of photos within one or two hours as many photojournalists do. I am also not commenting on the 9 frames/second shooting speed as this is of no importance for my work and also not of interest for me. What I am interested in are color rendition, resolving power, signal to noise performance and other image quality related aspects.
Let me start with exposure, ISO speed and noise - as always, I shoot in raw format (NEF) and only where it is absolutely necessary, I later convert those large files either into TIFF or JPEG. Especially the photographic comparison of the two cameras was performed with NEF files only.
First exposure - it was amazing for me to see that the very same scene taken with the same lens and with identical camera settings produce very different results - the D3 exposure algorithms expose the shots consistently more than those inside the D2Xs. Both shots below are unmanipulated NEF shots, converted and compressed for website publishing - no adjustment of exposure, brightness, contrast or color temperature.
Left side: D3 Right side: D2Xs - Absolutely same settings
It is pretty consistent what I observed - the D3 shots are brighter than the D2 shots. I need to point out that they are not overexposed raw files, I would say, these files reveal the fact, that the D3 has got a larger and wider tonal range - a wider range of exposure which can be used for imaging. Once you get used to it, you start liking it very much, I can assure you.
The following comparison was done to see if I expose as good as possible to make the dark parts identically dark, what happens to the bright parts of the image - you can see, again, the D3 shot is brighter in the bright part of the image. On top of this observation, the color balance was adjusted for the D2Xs shot because it was revealing a more yellowish tint (both cameras set to Auto-Whitebalance). Both shots had been 5stops underexposed to cope with the bright background - full sunshine on the Museum of Natural History in Vienna.
This difference in color rendition despite identical Whitebalance setting can be seen on the following two shots - again full sunshine, WB to Auto, same lens, shots taken within minutes (NEF) and identically converted by PS3 into JPEG and later compressed with identical settings - left the D3 and right the D2Xs. No further filtering, adjustment or corrections made.
The more yellowish tint in the D2Xs shot is clearly visible, especially the change of the color of the large vase.
Left: D3 Right: D2Xs
Noise was the next very interesting topic for me - it is well known that in general, the smaller the pixel size, the more noise at high ISO settings - this is due to limitations of the ability of small sensor areas to collect enough photons to get a well balanced statistics and due to the decreasing signal/electronic noise ratio. The new CMOS sensor of the D3 has pretty well sized pixels with not much space in between them - but still, the camera offers "only" about 12 Megapixel on a full frame size - Canon full frame sensors are much denser and the individual cell is smaller. It was expected to get an outstandingly low noise level with the D3 and it did not disappoint at all. The noise level is impressively low, even at speeds which are as high as ISO 6400. I need to point it out again - at ISO 12800, the noise level is clearly visible - what else, of course and at ISO 25600, well, you get an image, even in color but the noise is pretty impressive - so, use ISO 25600 only if you really need it.
Here some samples - one and the same shot, taken at all major ISO settings from ISO 800 up to ISO 25600 - not the entire image ( it is shown at the beginning for information), but a small crop out of those same images. I had decided to start as high as ISO 800 for this comparison, because the shots with ISO settings under 800 do not show noise enough that it can be recognized on an image after compression for publishing purposes. What you see are crops of shots at ISO 800-1600-3200-6400-12800 and 25600:
Abbey in the Monastery of St.Gallen/Switzerland
What I have used is the center this photograph - crops at ISO 800 - 1600 - 3200 on the upper part and those at ISO 6400 - 12800 - 25600 in the lower part of the two adjacent images below - I had to leave them rather large to enable the reader to see the differences:
ISO 800 - ISO 1600 - ISO 3200
ISO 6400 - ISO 12800 - ISO 25600
To avoid any loss of details, I have turned off the noise reduction algorithm - this algorithm is certainly good for certain applications, but here I wanted to show the unaltered effect of the ISO setting on the output (unaltered as much as possible - as usual, all those images had to be converted to TIFF, processed by cropping and then compressed for display on a website - each of the original shots is about 20MB large...)
To give you impression how a full shot at ISO 25600 inside that Abbey looks like - here it is, taken with AutoProgram at f:13, 1/640sec !
What happens in a much darker environment - would "color" pixel noise be visible? To answer this question, I shot a similar series of shots in the underground railway station of the airport in Zurich/Switzerland, on my way to the district general assembly of the Swiss Professional Photographers association in St.Gallen.
Same procedure, this time just the composed set of crops with the ISO settings indicated inside the image - the expected color noise at ISO 25600 is clearly visible, but do not forget, that is ISO 25600 - exposure details: f:9, 1/320sec, 1 stop underexposed
How does the D2 compare with the D2Xs - I did not bring both bodies to this one day meeting in Switzerland, so I took a series of shots in Vienna at night. What you see is on the left side a column of shots taken with the D3 - from ISO 200 to ISO 25600 and on the right side the column of crops from the D2Xs from ISO 200 to ISO 3200, the highest possible ISO settings. Not much more to explain, the visual comparison speaks for itself - the D3 is clearly by far superior to the D2Xs
The crops represent again a small part of the entire image - which can bee seen below here:
And here the crops - the left column the D3 shots and the right column shows the comparable D2xs shots
Even after the compression to match the requirement for websites, one can clearly see that the D3 shots are more detailed, less grainy, clearer, better defined and overall just much better than the ones from the D2Xs.
Color and color rendition is another topic of significant interest for me - and I guess, for most photographers. Digital cameras reveal a different behavior towards color compared to film based cameras because the digital sensor works totally different compared to film. Skin tones, highly saturated colors, very light or very dark colors with a very low degree of saturation - these are the problem zones in photography. There are many ways to approach this, I usually evaluate the results obtained with the Gretag Macbeth Color Checker and the Color Checker SG, which had been specifically designed for use with digital cameras.
First the results obtained with the standard Color Checker and the Nikon D3 in comparison to the D2xs - you can see the D3 results below (file DSC5429..) but when you place your mouse of that image, the results of the D2xs are displayed - note the different file numer!
What one can see on this graph is the aberration of the color patches, which are located on the Color Checker as photographed with the D3 and the D2xs in comparison with their theoretical location - the data are standard date supplied by Gretag Macbeth (Xrite) for this chart.
Mean camera chroma - the saturation - over 100% means that the colors are reproduced more vividly than the patches are theoretically - something which is very common to all digital cameras, even when the setting for color rendition is "neutral". The D2xs shows a tiny bit different result, but nothing worth mentioning.
Much more of importance are the color errors - these data reveal the deviation of the color as shot with a camera from the theoretical location in the color map - the larger the values, the more deviation in color. As with most digital sensors, color errors are normal (they of course do also exist when shooting with film - more about this later) - a mean chroma error of about 4 for the D3 is very low, and the delta E in the CIE-Lab color system for the D3 as shown here about 8, means that yes, one can see that the colors displayed are somehow different from their theoretical location but if you examine the chart, you can see that this deviation varies a lot - also a very normal and common behavior. The results obtained from and with the D2xs are showing that both date a slightly larger - about 5 for the delta C and about 11 for the delta E. This shows, that in general, the color rendition of the D3 is slightly better than the D2xs - not a huge difference, but noticeable.
Camera manufacturer usually tend to minimize color error for those colors which are very well known to humans - like skin color tones and by doing this rather sacrifice those color regions, which are less important - odd colors or colors, where a larger deviation does not make too much a difference for the visual impression. Again - if you think that film is better, do the same test with you most favorite film and you will find out, that most likely, these deviations are at least as large as they shown here.
I used Adobe RGD as color setting, which is wider than sRGB and D65 which is a standard illumination, more or less like standard daylight.
Next set of results are White Balance Error - again, the primary display shows the D3 results and when you put your mouse over the chart, the results for the D2Xs will be displayed - and here one can see a difference!
This chart shows, as mentioned before, the actual shot of the Color Checker with its 24 patches and compares the actual colors with the theoretical colors - visually to show how different a color was rendered - with and without luminance correction. Interesting part here is the lower, grey range of patches.
In an ideal situation with an ideal camera and ideal illumination, these grey patches will reveal no difference from their theoretical values - no color shift, no density deviation. But in real life, things are different - cameras tend towards either a more bluish or a more yellowish reproduction of neutral gray tones. As one can see, the brightest grey patches are a bit too warm and the dark ones a bit too cold - but their deviation is very small, which means that the white balancing of the camera under these lighting conditions are excellent.
The D3 reveals again the interesting fact, that all exposures are a bit brighter than the ones with the D2Xs - when you move the mouse over and out, you can see that this is a significant difference. Next - the WB error - in both cases (same illumination) quite good, actually very good. The D2Xs is slightly behind the D3 and also got a slightly larger exposure error - 0.6 f-stops compared to less than 0.3 f-stops with the D3. But overall a rather impressive result for both cameras.
Keep in mind - the illumination with D65, which means basically simple daylight.
The next two sets shows the results for the gray steps, some information on noise, exposure error and noise spectrum. As before - same setup - the displayed results are the calculations for the D3 and when you move your mouse over the graph, the results for the D2Xs will come up.
Interestingly, the D2Xs results claim that the illumination was somehow uneven but the D3 did not show this result - probably because the new sensor of the D3 can cope with it much better due to its larger exposure latitude, its ability to digest a wider tonal range. Other data can be directly read from the two charts - again, the D3 in general better than the D2Xs - not by miles, but visibly better.
Look on the left side at the difference in pixel noise and especially at the lower right corner, the noise spectrum - quite a difference between the D3 and the D2Xs.
I am aware that this comparison is not to100% accurate because the exposure errors are different, but I am sure, that even with absolutely identical exposure, the noise power of the D2Xs would be higher than that of the D3.
For those who are interested to see how a rather good 35mm film behaves compared to the D3 - here is a link, which leads you to a new page, showing the same D3 results in comparison - film (Fuji Superia 100). (FujiSuperia/NikonD3)
Can one be surprised, how "bad" film is ? - donīt be surprised, that is 35mm reality! 35mm film behaves like that, and as all recording media, it has certain characteristics - Fuji as example, pronounces the green tones, which is very well known - and that film shows it and its grain, no wonder, that is how film was designed....
A more detailed information can be obtained by shooting the newer Color Checker SG, which was specifically designed for the evaluation of digital cameras, but is also very usefull for evaluation of film based cameras because it contains many more color patches - not only the 24 standard colors, which are maintained as centerpiece of this color checker - as before - what you see are the results for the Nikon D3 and again, when you place your mouse over the graphs, then the files and charts for the D2Xs are displayed.
There is quite a significant difference again, especially in the exposure itself - however, I have exposed the chart with both cameras under the very same lighting conditions, same settings and even the very same lens, which was the newly released 2,8/24-70 AF-S zoom, which is suitable for use with the D3. To compensate for the different sensor sizes, I have changed the focal length to get the same field of view.
Digital Color Checker SG
The Imatest program displays the color differences in various modes - here is the display set to a color scheme - with white as no difference between theoretical values and actual image and black the maximum. As you can see, most of the areas are quite different, but more in brightness than in the actual color location
This way of displaying shows the actual colors in direct split-image comparison to the theoretical colors. With the exception of some, most colors as taken by the D3 are just brighter than their theoretical values, but some are quite different!
In this display, one can see the 2D a*b* differences - in the La*b* system, which is a mathematical derivative of the CIE diagram, the DeltaE values (color differences) are more comparable between the colors than in the CIE diagram
Finally here the CIE xy diagram, which is very well known. As one can see, especially the very saturated colors, like patch 49,51, 35, 47 or 34 had been moved inwards towards less "colorful" locations, most of them by quite a large amount. This is not uncommon for digital media, because the most saturated colors represent usually significant problems for color rendition (as do the very light, almost non-colored) colors. Overall I have the impression that the D2Xs in its normal setting (color-neutral) exposes this chart much darker but more accurate in certain areas. It would be interesting to see if one can adjust the D3 such, that these partially large differences become much smaller - I guess, it should be possible but I did not have time enough to find it out as time for that loaner camera was rather limited and it needed to back with Nikon for another exhibition and more photographers who are potential buyers.
How does all this translate into "normal" photography? How do these differences show? I have taken a series of shots of a subject, which readers know from various of my previous review, an art nouveau vase, silver and colorful glass, which I can access for photography if I want to. To show a direct comparison, I have this time placed the two results parallel and underneath the shot taken with the D2Xs, by placing the mouse of it, one can see the unaltered NEF exposure (only converted for publishing) - to show how the D2Xs and D3 raw files differ from the optimized files after conversion it with Nikon Raw 4.3.1 and processing in PS3.
The raw file of the D2xs is colder, despite the very same lighting and the same setting for Whitebalance - Automatic (I wanted to see how WB Auto copes with this kind of an object in front of a neutral gray background) and after processing has changed the color rendition more than the one from the D3, which to me is just darker, not different. Those, who have a large enough screen with higher resolution can use this link here (Parallel view) to see both images parallel and simultaneously.
Another topic, which is always of interest to me is to see how good digital cameras resolve small details - both are in a range of well above 10Mpx, which means, that they are real high resolution sensor cameras, but does the new D3 sensor also show a better resolving power than the D2Xs sensor? Another one of my standard target is the interior of the Church in Mariahilf/Vienna, which is one of the nicest district churches in Vienna - same procedure as always - both cameras, both equipped with the same lens and same settings. Here the results - first of all one can see that in identical lighting situations, the D2Xs tends to produce a different color tint - more towards green/blue compared to the D3, which to me appears to be much more correct.
Left side: D3, Right side: D2xs
Now the details - I used a crop out of the center of the images, above the altar - you know this crop from previous reviews - it contains a lot of details, some inscriptions and many different color and brightness tones. For both cameras ISO 200, neutral color rendition, images taken as NEF files and then converted with Nikon raw 4.3.1 into TIF files before processing/compressing for display on a website.
Left side: D3, Right side: D2Xs
For those readers, who have got a larger monitor with higher resolution, I added a link (HighResD2D3) to a separate page where one can look at those details at higher resolution and less compressed - however, the caveat is, that these files are larger and take longer to load, but the direct comparison is interesting.
As it can be seen clearly, the images taken with the D3 are of higher resolution and overall also more appealing in color. That outcome was expected and I hoped to see it myself, which I did.
Finally, I wanted to see if this subjective impression of a better image and a better definition can be seen also on the MTF - I need to remind the reader, that comparing MTF graphs is tricky - MTF depends on a large amount of subtle details and as no two models can be set to absolutely the identical settings, any comparison is not perfect. What I did was to set the D2Xs and the D3 to as many as possible identical settings and values to get them as close as possible identical in their setup. For evaluation, I used IMATEST again, which has more or less become the industry standard for optical performance evaluation with digital camera systems.
The "slanted edge" is a very simple yet very powerful method to evaluate imaging systems in terms of resolution, MTF, noise, Shannon capacity and noise statistics. For those, who want to read more about these values, please look at the Imatest home page and browse until you get to the manual - everything is well explained there in detail. Same procedure with the following charts and graphs again - first, the D3 results are displayed and by placing the mouse over the respective chart/graph, the D2Xs results are shown
Would you like to see how film compares under same conditions with the D3 - I took the same shots on Superia 100 with my Nikon F100 and if you like to see the comparison, use this link here:SuperiaD3Edge.htm Superia/D3Edge
I mentioned in the subpage, showing the comparison between film and digital, that sharpening influences the outcome of these measurements dramatically - for information, I have used again the results of the D3 without sharpening (the first graph from above) and when you place your mouse over the graph, the results obtained with a sharpening setting of 4/9 on the D3 are displayer - very different results. Maybe I find some time one day to write a whole tutorial about resolution and all the factors which influence the outcome.
It is important to find the right amount of sharpening not to get to useless evaluation data - for the D3, the be approximately right, which results in more or less neither undersharpening nor oversharpening (see the text inside the graphs), the value in custom settings for sharpening have been 2/9 - a value which delivers rather soft images for most scenes, but the SW tells me, that this value is the best for this target. Keep in mind, that this value depends also on the contrast of the target - a less contrasted target can be sharpened more that a high-contrast target like a black-white edge. With a sharpening value set to 2 on the D3, the results looks like this:
With the mouse over the graph, the Shannon capacity and noise graph is displayed - data look very attractive and confirm the visual results.
What is my conclusion - well, first of all, the D3 is a significant improvement over the D2Xs, looking at all details and the results, obtained from photography and evaluation.
It is not only the nice and large backside screen and some additional "features" of the camera (artificial horizon as example), which makes it very attractive and interesting, it is its performance, which is clearly superior to the D2Xs. The full frame size sensor did not compromise the resolving power at all, despite the fact, that this sensor is, regarding pixel count, not much better than the D2Xs sensor, the resulting images are clearly better. Noise is visibly significant less compared to the previous model, color rendition is different - for many scenes better (what does better mean - this maybe would be another longish article) or at least better suitable, for some objects and lighting situations, one needs to get used to the different characteristics of this sensor and its new algorithms and processor.
I did like the much larger viewfinder viewing angle very much, the dual-slot for storage media and the ability to customize the way of taking, recording and processing images - any user can adjust by means of these many settings (some of them are not too easy to find) all important parameters according to liking, flavor and task. I have concentrated in this review on what I regard as important for me, probably other photographers would concentrate on totally different topics (like speed of shooting, frames/second etc) but as I always want to make clear - a review is never complete nor comprehensive nor really neutral. I tried to be as unbiased as I can be. If any reader has got questions, please contact me, I shall try to answer them without too much of a delay.
There was not too much time to do some nice shots, some of them are in this review, but maybe in the future, I have the opportunity again to get my hands on the D3 and have it for some time to do some nice shots....
Did you really like what you read and have seen here in this review - well, the author would be very happy to receive a small donation via PayPal, because this review, like all others by this author, was not compiled, written and published because the author got paid for it, - the author did it, because of his interest in photography, imaging technologies and color.
Back to the HomePage