D600: here comes the DXO Mark rating...

[anuragagnihotri]

<p>Second highest overall score after D800. High ISO score is better than D800, especially true for a non 'normalized for print' shot.</p>

<p>http://www.DXOMARK.com/index.php/Publications/DxOMark-Reviews/Nikon-D600-sets-high-bar-for-sensor-image-quality/Measurement</p>

[jose_angel]

<p>If we compare the signal-to-noise ratio performance charts, it`s surprising (or maybe not?) that the linearity between the D4, D800 and D400 is almost identical, with the <em>only</em> exception at 25600ISO (top of the D400&D800), where the D800 is slightly behind the D4, and the D400 behind the D800...</p>

[fedelucero]

<p>Somebody is in denial and was expecting a D400 instead of a D600 on the latest nikon's announcement =)</p>

[elliot1]

<p>The differences in the ratings are too small to be significant and would be impossible to see in a print or even pixel peeping. Bravo Nikon!</p>

[borgis_karl_johan1]

<p>I very much reserve my judgement, I suggest everyone should read what Brad Hill has to say in his admirable nature photography and Nikon blog:<br>

http://www.naturalart.ca/voice/blog.html<br>

<br />Intuitively I think there´s something fishy with the way they do these comparisons, reducing the resolution to around 8mP the way they do. I´m no technician, I grant them that though.</p>

[Kent Shafer]

<p>I don't understand what Mr. Hill means by, "<em>Many</em> of us want to use our images at full resolution …."</p>

<p>It seems to me that our images ultimately get used in only two ways. Either they're displayed on a computer screen, downsized to A x B pixels, or they're printed on paper at some given size in inches or centimeters. When an image is printed, it's upsized or downsized—either by the printer software or beforehand by other software—to match the printer's resolution. For either use, it would seem that the noise reduction benefit of starting with a higher resolution image would actually matter and that, therefore, it would be correct to compare the performance of different sensors only after resizing the images to some standard dimensions in pixels.</p>

<p>How does one use an image "at full resolution"?</p>

[m_tt_donuts]

<p>This applies mostly to Borgis Post. To compare the lowest common denominator, comparing the D600 to a D4 you'd have to lower the D600 image to 16.2MP. Now throw in the D80 and all have to be lowered to 10.2MP. They found 8.4MP to be the lowest demonimator apparantly of all. The thing here is that <em>"when you scale down to a lower resolution, the resulting noise and Dynamic range are equivalent to what you would have gotten if you had started off with a sensor which had the required target resolution"</em> - <em>luminous landscape. </em>SoNoise and dynamic range scale meaning comparing 8.4MP to 8.4MP is one to one. If a D600 did worse than a D4 at 16.2MP it's not possible for it to do better when both scaled to 8.4MP.</p>

<p>On the other side, one has to realize with Low-ISO they are testing what's the highest ISO that can be achieved while maintaining a SNR of 30db, dynamic range of 9EV, and 18 bit color depth any one of those 3 conditions fail it drops out of the race... I'm looking at the DxOMark scores of the D800, D600, and Mark III and it's clear to me higher than 12800 ISO images from the Mark III pull ahead... but it's long after it has failed the 18 bit color depth (it's a full stop below the Nikons). If the race continued, you'd see at around 12800 ISO the Mark III surpasses the D600 and D800 but the race already closed so if you use Canon at 16000 ISO the Mark III would have better images. Then changing it to just Nikons D4, D600, D800 they are all neck & neck and lose at about the same time. However, looking at the charts if the race continued wow does the D4 pull ahead (even compared to the Mark III at high ISO). Higher than 12800 ISO the D600 performance sinks like a lead balloon, the D800 performance drops a lot at that moment as well (not as much as D600), the D4 is just wow, DxoMark shows the D4 images taken at 36,000 ISO will look like the D600 images taken at 19,000. That's crazy, but again the race already over. Looking at the blog link, he says the D4 has better image DxOMark does actually agree... and yes the D4 does beat the D600 and D800 quite handily over 12800 ISO. </p>

[anuragagnihotri]

<p>what it means:<br>

D600 will be better at low light at 100%<br>

It will be better at pixel level sharpness and noise. <br>

D800 will be better at low light when printed large.</p>

[Andrew Garrard]

<blockquote>

<p> If the race continued, you'd see at around 12800 ISO the Mark III surpasses the D600 and D800 but the race already closed</p>

</blockquote>

<p>Speaking as an unbiased someone with a D800E and absolutely no interest in buying a 5D3, I'd like to point out that "the race has closed" argument still makes quite an assumption. I've taken quite a lot of shots with my D700 at ISO 6400 and above - no, they're not perfect full size, but for a small print or on-line image that may not matter. If there's no light, a noisy image may be better than a very noisy image. So the slight advantage of the 5D3 and the D4 at very high ISO aren't completely irrelevant. I'd rather have the low-ISO performance of my D800E as a compromise, but some people might make the opposite argument (especially for the press, where you need images in bad conditions rather than wall-size prints, which explains a lot about the D4's behaviour).</p>

<blockquote>

<p>D600 will be better at low light at 100%<br />It will be better at pixel level sharpness and noise. <br />D800 will be better at low light when printed large.</p>

</blockquote>

<p>Actually, if the lens keeps up, at very low light I suspect the lack of filter on the D800E trumps other considerations. It's true that - per pixel - the D600 will be better at a per-pixel level at intermediate ISOs (in that it'll have less noise); a plain D800 may well appear less sharp at lower ISOs since the D600 will be sampling a larger portion of the lens output. I suggest this argument is relevant only if you're trying to decide whether a D800 will really allow you to shoot sqrt(1.5) times larger than the D600. At a fixed output size, per-pixel response from two different resolutions are generally not useful to compare</p>

[keith reeder]

<p>And yet, Mätt, none of this supposed superior IQ shows up in images <em>unless </em>we're being treated to tortured base ISO shadows...</p>

<p>As Brad says (with my emphasis):</p>

<blockquote>

<p>In my role as moderator of the Wildlife Gallery on the Nature Photographer's Network I see thousands of images annually shot by both Canon and Nikon cameras. And... can the Canon sensors be as bad as dxomark.com's tests show <em>if the images I see shot by Canons are so darned good</em>??</p>

</blockquote>

<p>And he's <em>absolutely</em> right.</p>

[Andrew Garrard]

<p>Keith - quite. Generally, we can agree that for 95% of shots (depending on what you're doing) the camera doesn't make much difference. But we're presumably on this forum because, all else being equal, we'd like to spend our money on picking the best option for the last 5% - or, equally, deciding that spending money for the last 5% to be better is not worthwhile to us. So long as we keep sight of that, I'm perfectly happy to discuss whether one camera is very slightly better than another under particular circumstances. I have (by the standards of most people, but probably not around here) lots of cameras, and each has its own merits. If it didn't, I'd have got rid of it. I can take an equally incompetent photo with most of them, and it'll pass muster.</p>

[rodeo_joe1]

<p>Scaling only works correctly to cancel noise if the pixel count is reduced by integer divisions of 4. Anything else results in some kind of weird fudge. So the D800 would have to be scaled down to 9 Mp, and the D600 to around 6 Mp. You can't get true noise cancellation by "subtracting" 0.8 of a pixel from a whole pixel or suchlike, that's just crazy talk. To arrive at a true common denominator across a variety of cameras would result in being left with a really tiny number of pixels.</p>

<p>Here's an idea. Why not take a standard subject and make a standard sized <strong>print</strong> from the respective cameras, then compare the image quality by <strong>eye</strong>? Or is that just too radical?</p>

[Andrew Garrard]

<p>RJ: It depends whether you're trying to reconstruct a signal at the frequency of the lower-resolution sensor. Making a higher-resolution print works because you're producing a finely-sampled representation of the reconstructed waveform once it's been generated by whichever underlying frequency the sensor holds. But it's possible I shouldn't post to this forum after spending the evening at a bar in a graphics conference.</p>

<p>(Upshot: Printing is a good way to compare, non-integer downsampling may or may not be valid depending on what you're looking at.)</p>

[elliot1]

<p>The Imaging Resource site has posted samples shots. Their site makes it easy to compare IQ from one camera to another.</p>

<p><a href="http://www.imaging-resource.com/IMCOMP/COMPS01.HTM">http://www.imaging-resource.com/IMCOMP/COMPS01.HTM</a></p>

<p>The differences in picture quality between the D600 and D800 are slight at best, and would difficult to see after post processing RAW images (sample images displayed on that site are from in-camera JPGs I believe).</p>

[dan_south]

The D800 is very clean up to about ISO 800. However, at ISO 3200 it is about is noisy as my old D200 was at ISO 1600. It's pretty

disappointing by today's standards. Maybe DxO likes photos that look as though they were taken in a sandstorm, but after using a D700

and the 5D Mark 2 and 3, I'm admittedly rather spoiled. I love my D800, but regardless of how it's portrayed on forums, it's not a high ISO

camera. Other bodies are far superior.

[Andrew Garrard]

<p>Dan - you're right <em>at the pixel leve</em><em>l</em>. If you consider the D800 as being equivalent to an older camera except that it produces three times as many pixels, then yes, it's not as much better as you'd think. If you think of it in terms of capturing an FX frame, the noise is better than the D700 - more so than I expected. (The 5D3, modulo its smoothing function, may or may not be better at very high ISOs, as I've argued elsewhere; the same can be said of the 1Dx and D4.)</p>

<p>Put another way, a D7000 vs D800 is like using 35mm vs medium format film. Produce a 10x enlargement, and the sensors look just as noisy as each other. Produce the same print size and the D800 (and D600) look less grainy.</p>

<p>Now, if we want to compare with the D700, we could say that the D700 is like 35mm ISO100 film and the D800 is like medium format ISO200 film. (That doesn't quite work because grain size affects resolution, but bear with me). At 100% (looking at a fixed area of film, or looking at pixels) the D700 is smoother. Consider the whole capture, and not so much.</p>

[anuragagnihotri]

<p>So what's the deal?<br>

When you downsample a D800 image to DXO standard, 8mp, it looks as clean as a similarly downsampled D600 image...<br>

What happens when you downsample a D800 image to D600 size...which looks cleaner?</p>

 

[Andrew Garrard]

<p>Anurag - they should look extremely similar. The downsampled D800 image will look slightly softer if you've got detail that approaches the resolution of the D800 sensor, because there's no 1:1 mapping from each D800 pixel and each D600 pixel - but the same is true (and more so) if you scale a D600 image up to D800 size. Scale to a significantly higher resolution than either (as in a print) and this distinction goes away. But, for noise, they'll be similar - and better than a D700.</p>

[rodeo_joe1]

<p>It might be salient and instructive to think back to a previous and almost forgotten imaging technology that had to dither a very limited number of colours to produce the appearance of continuous tone. What was it called? Errr, f-i-l-m or something like that. Very noisy stuff that could only represent a full-on tone or absence of tone. However, its image was comprised of so many tiny splodges of a single tone that they visually merged into the appearance of smoothness. And that's basically the difference between the D800's "fine-grain" approach and a lesser camera's more coarse-grained image.</p>

<p>At a similar reproduction size, a high pixel density will <em>give the appearance of </em>a much smoother tone than a lesser number of pixels with the same pixel-level noise. Ergo and conversely, the pixel-level noise can increase considerably while the visual impression remains of a similar or lower noise than that given by a lesser number of pixels/unit area.</p>

<p>The reason for this is that any noise, being at the pixel pitch, is of a higher spatial frequency and is more easily visually merged, or is physically obscured by either printer dithering or pixel interpolation. It's only at pixel peeping magnifications that such high frequency noise becomes viewable and obvious.</p>

[allan_crookes]

<p>All I can say is that my D600 takes some nice shots. I took it out for its first shoot this morning, and with the 24-120mm it shoots very impressively. I'm not talking "technically" here (what's on paper or the internet), I'm talking real world stuff. Conditions were a little overcast at times but otherwise sunny and although the 24-120mm may not be the perfect all-rounder, it sure performed well for me today. I left the 70-300mm lens behind, simply forgot to take it, and I really wanted to give that a go, as I have shot some fine pics with it on the D300, even though it's not highly rated by some. I had an earlier 24-120mm DX on the D300 and I thought it under-performed so was a little reluctant to go for the new FX 24-120mm but I was assured by the local Nikon specialist at the time that it is a well matched lens for the D600 and would really perform...</p>

<p>I'm going to have to get some more lenses come the new year but for now these two lenses will have to do... I'm sure they will get my feet off the ground for now...</p>

[dan_south]

<p>The D600 looks like an extremely capable body. It has a lot of great features, and the Chasing The Light promotional video was very enjoyable.</p>

<p>Regarding noise and downsampling, however, I don't recall anyone having to down sample 12 MP D3 files to make them more noise free than 12 MP D2X files. The high ISO performance was just plain better, and obviously so. So, I get a chuckle when someone suggests reducing 24 or 36 MP to the D700's resolution for noise comparisons. If the noise performance were comparable in the first place, there would be no need to level the playing field.</p>

 

[ShunCheung]

<p>Dan, since both the D3 and D2X are 12MP, you can't down sample a D3 image to match the D2X. However, it makes perfect sense to down sample a 36MP D800 image or a 24MP D600 image to 12MP to compare against a D700 image; otherwise, it is not a level playing field.</p>

<p>Since the D800 has 3 times as many pixels as the D700 and both are FX, each D800 pixel only occupies roughly 1/3 of the area as each D700 pixel. It should surprise no one that the D700 is better per pixel. However, if you are making the same 8x10 print from both, you need to down sample the D800 image anyway, and only then you have a fair comparison and the D800 is far better than the D700 in terms of high-ISO performance.</p>

[keirst]

<p>If you compare cameras at DXOmark.com you can always look at the measurements (other than ISO) as either Print, which is normalized to 8MP or as Screen which is based on 100% resolution if I understand that correctly. If you do that the Signal to Noise ratio of the D600 falls between that of the D800E and D4. I do wonder if there is much significance to that, however, for most users. Would a 1-2 dB better SN be visible in a final print or other end use? I plan to do large prints, so I opted for a D800E, because I will want the high resolution.</p>