Oh dear....5D3 vs D800

<p>Well, DxOMark has spoken.</p>

 

<p>The Nikon D800 scores a 95 overall with a 14.4 EV dynamic range at ISO 100</p>

 

<p>The Canon 5D MkIII scores and 81 overall with an 11.7 EV dynamic range at ISO 100</p>

 

<p>The 5D MkIII shows a very slight noise advantage (less than a stop) when the images are viewed at 100% on a monitor screen.</p>

 

<p>The 5D mkIII has a slight tonal range advantage, offset by the slight color sensitivity advantage of the D800</p>

 

<p>The D800 obviously has higher resolution.</p>

 

<p>The 5D MkIII is the clear winner at ISO 51200 and 10400 since the D800 is limited to ISO 25600, though at 25600 the noise is verysimilar to that of the 5D MkIII, so it makes you wonder why they didn't just go to 104,000 like Canon did.</p>

 

<p>Speaking as a Canon shooter, quite honestly it does look like the D800 is  the better camera (at least on paper as far as performance specs go).  I haven't seen comparison tests of AF performance yet, or anything definitive on video results.</p>

 

<p>Let the wailing and gnashing of teeth by Canon shooters begin...</p>

<p>Oh, the humanity...</p>

<p>The overwhelming majority of Canon and Nikon shooters will be unaffected. My T1i and 20D still work the same as yesterday.</p>

<p>Read 'em and weep</p>

<p>http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/795%7C0/(brand)/Canon/(appareil2)/792%7C0/(brand2)/Nikon/(appareil3)/767%7C0/(brand3)/Nikon</p>

<blockquote>

<p>My T1i and 20D still work the same as yesterday.</p>

</blockquote>

<p>So does my 1978 Chevy Chevette, my 8-track tape player and my Betamax video recorder.<br>

<br /> I'm unaffected too because I don't need and can't afford a 5d MkIII or a D800 and I certainly couldn't afford to add Nikon gear to my Canon gear.</p>

<p>For all I know these numbers may mean something, but I've never been able to figure out anything at that site.</p>

<p>From their Overall Score explanation,</p>

 

<blockquote>

<p>The Sensor Overall Score is an average of the Portrait Score based on Color Depth, the Landscape Score based on Dynamic Range and the Sports Score based on Low-Light ISO.</p>

</blockquote>

<p>How do you calculate the average of 24 bits, 11.7 Evs, and 2293 ISO?</p>

<p>Note I'm just commenting on how bewildering it all is. I actually do not care how they arrive at their numbers. At all.</p>

<p>"Feeling good about your equipment is the most important aspect of photography"</p>

<p>H. Cartier Bresson (NOT)</p>

<p>If you shoot Raw, the difference in DR is real. Just experiment with different sliders for highlight and shadows and you will see the difference. The reason you have to use the sliders is because the DR of all output mediums (LCD, Paper, probably even CRT) is much less than that of Camera.</p>

<p>The extra pixels definitely bring out additional details, and with skilled shooting and processing even the scaled down image can display more clarity. The real kicker here will be D800E with merciless pixel level clarity showing all the gory details.</p>

<p>If you don't shoot Raw, 5D Mark III will provide a superior JPEG by a mile. </p>

<p>For mere mortals like me, I will take 5D classic any time.</p>

 

<p>Overall scores for anything are bewildering. It's an attempt to consolidate too many different factors. It's like rating a car as a 73/100 based on top speed, luggage capacity, number of cupholders and mpg.</p>

<p>However the DxOMark individual scores and their plots of things like SNR, Dynamic range etc. are quite meaningful (assuming they are correct). They do describe their testing methods which seem to be pretty valid.</p>

<p>The factor that's boosting the overall score of the d800 appears to be the unusually high dynamic range at low ISO settings. I'll admit I'm not 100% sure how they are computing that since the difference between the 5d3 and D800 SNRs at low ISO settings are small and DR is often an indirect measure of shadow noise.</p>

<p>I guess I'm getting hung up on the word "average", which has a precise meaning in math and statistics. You can't "average" 24 bits, 11.7 Evs, and ISO 2293, any more than you can average 24 meters, 11.7 seconds, and 2293 grams.</p>

<p>It wouldn't bother me if they said "Overall Score is an artificial heuristic metric based on the bit depth, Ev range, and ISO".</p>

<p>As a Canon still shooter I do wonder whether Canon focused its attention on professional video and left the amateur still shooters in the cold as far as full frame is concerned. The Nikon D800 does look better on paper and is cheaper to boot. New Canon lenses for full frame are astronomically expensive... $850 for Canon 28 f/2.8, compared to $700 for the just announced Nikon 28 f/1.8 (although without IS). The new 24-70 jumped from being the least expensive such lens (compared to Nikon and Sony/Zeiss) to being the most expensive. I get a distinct feeling that Canon aims the latest full frame products at professionals for whom the extra few hundred or thousand dollars on gear is small change compared to other expenses and/or income.</p>

<p>I don't know whether there indeed are over 3 stops difference in dynamic range/tonal capacity between these camers in real life but if there indeed are, that would indicate a quantum leap in the sensor technology. 3 stops is nothing to snicker at.<br>

<em></em> <br>

<em>> HCB or not</em><br>

However, HCB always used the best of the best of the best equipment, so he did care. Otherwise he'd use Kodak Pathé Browné :-)</p>

<p>The DR range is odd. It only appears at low ISO settings. However at both low and high ISO settings the SNR of the 5D3 and D800 are similar. That would suggest shadow noise should be similar, which would normally make DR similar.</p>

<p>If it was some odd effect of pixel count or size that comes out of their algorithm, then it should apply at all ISO settings.</p>

<p>It looks like you may be able to dig more highlight detail out of RAW D800 files than RAW 5D3 files, but only at low ISO settings.</p>

<p>DxOMark don't seem to explain exactly where the extra DR at low ISO is coming from, so I guess we'll have to wait for more explicit and transparent tests to figure that one out.</p>

<p>Nikon seems to be doing a lot of in-camera processing on their RAW files (akin to post-processing highlight and shadow recovery) which may account for some increase in DR/tonal capacity but might make the files less responsive in post processing to shadow/highlight recovery and other tortures. I know from my own experience that at ISO 200 ir so, I can use much more drastic post processing techniques on files from Canon 1Ds3 than on files from Nikon D3s or Canon 5D2. Nikon D3s files, although marginally "better" than the output from 1Ds3, tend to fall apart in terms of artifacts sooner than the files from my benchmark camera 1Ds Mk. III. I don't have any experience with D800 in this respect but something tells me that it still might be true, barring a true technological miracle in sensor tech.</p>

<p>DxoMark say:</p>

 

<blockquote>

<p>Certain manufacturers embed a small part of the processing directly in the sensor, which means that some degree of processing occurs before the RAW image is sent to the RAW converter. In this case, measurements for these "pre-cooked" RAW images can be biased by this processing.<br>

To avoid any potential impact on our measurements, DxOMark always tests all cameras to detect any pre-processing of RAW images. A processed or pre-cooked RAW image has different characteristics from a genuinely unprocessed image. To some extent, these characteristics enable us to walk back the processing and reconstruct the original image to perform unbiased measurements, and we always inform the user about models with embedded pre-processing.</p>

</blockquote>

<p>I don't know if Nikon do much in-camera processing of image data before the RAW file is written, but I didn't see any mention of it in the DxOMark report.</p>

<p>Why does anyone even go to their web site? DxOMark that is. I have never read anything good about what they do! But, they are there. A lot like a pimple on a gnats butt! Poor Gnat! Seriously, is there any reason to visit their web site? I did it once...by mistake! And I am with many I have 2 canon's and am not about to think about Nikon. I am photography poor as it is!</p>

<p>Because they are are only website that makes (or at least claims to make) actual scientifically based tests on camera sensors. They don't carry camera advertising so one might assume that they are unbiased. They certainly have no reason to falsify results and they (DxO) are a compnay with enough resources and business reasons to do accurate and unbiased testing. DxO - <a href="http://www.dxo.com/us/corporate/home">http://www.dxo.com/us/corporate/home</a> - is a company based on scientific methodology for optical testing. Not some guy like me shooting test targets in his basement!</p>

<p>If you don't want to know anything about camera sensors (and many people don't), then there would be no reason to visit them, just as if you have no interest in cars there's no point reading test reports in Road & Track.</p>

<p>Regarding the claimed 14.4 EV dynamic range of the D800 - since the raw files are at best 14 bit "deep", to achieve 14.4 stops the highlights must undergo some kind of compression. This probably disconnects SNR from dynamic range, so SNR ratio could be similar to the 5D Mark 3 yet the DR could be different.</p>

<p>Bob and others:</p>

<p>I'm watching and waiting, but hardly gnashing my teeth. I've said it before and I'll probably say it many more times in the future: Nikon is doing a really fine job right now and producing some excellent gear at excellent price points. If I were a Nikon shooter doing the sort of work I do, I would almost certainly have ordered a D800 by now. Canon photographers who care about such things, have gotten used to being able to argue strongly that their brand was better overall - but that is much less of a sure bet now. There are clearly two very strong competitors in Canon and Nikon, and that is good for photographers and good for photography.</p>

<p>Am I worried? No. Am I considering a brand switch? You must be kidding! :-)</p>

<p>But a bit of perspective is important. Yes, the new camera from Nikon measures better than the new camera from Canon in several ways. But keep a few things in mind:</p>

<ul>

<li>Better measurements - unless it is ones goal to own things that measure better - is important only to the extent that it makes for better photographs. The 5D2 and the 5D3 and other Canon cameras produce excellent photographic image quality. For a few people doing certain kinds of photography in certain ways and producing certain types of output, there will be some situations in which the improvements from the D800 might be visible, but they would generally be in the category of "even better than the excellent quality we already have" rather than "finally, a camera that works!"</li>

<li>In many industries, the "lead" will often seesaw back and forth between two or more competitors. Company A produces some really cool new widget with a special feature and is momentarily "ahead" of Company B. Company B already has something equivalent under development that will be released soon and/or has some other super duper widget that has its own form of specialness. A few months later, Company B makes its product announcement and suddenly the tables are turned. But... a little later Company A (or C or D) responds and the landscape changes again. (And the teeth gnashing begins...) Over time, users of products from any of these companies will have excellent and competent equipment regardless of short term comparisons.</li>

</ul>

<p>Once we start down the road of "spec fever" things become very strange. Incremental differences are inflated into "good and bad," and end up always being associated with the selection of "winners" and "losers," "best" and "worse," and the rest - not based on what one can do with the gear, but based purely on the specifications. (My car can go 165.7 mph and yours can only go 165.3. I win! :-)</p>

<p>As a San Francisco Giants fan I have a wonderful case in point. Matt Cain is one of the outstanding pitchers on the team. I don't track statistics, but someone on the radio this morning was lamenting that Cain's stats don't look all that great. For example, his win/loss record might not strike people as all that stupendous in some ways. The stats tell us that yesterday, for example, he did not get a win. But <em>he pitched nine innings of no-hit baseball</em> before the game went into extra innings and he was finally taken out for a reliever. In the 11th inning, some crazy shenanigans occurred and the Giants got a run and another Giants pitcher got the win. So the stats tell us that Cain is a decent but unremarkable pitcher - but what he actually does tells us a very different story.</p>

<p>Statistics and specifications are interesting and can be loads of fun to argue about. But relationship between them and these things called photographs is very, very tenuous.</p>

<p>Besides, does anyone really doubt that Canon will come out with a high MP full frame camera? ;-)</p>

<p>Dan</p>

<p>These guys give a nod to D800 RAW IQ and but say 5D3 AF is a bit better. Oddly, 5D3 JPEG output gets the nob over the D800.</p>

<p>http://www.luminous-landscape.com/videos/d800_vs_5dmkiii_video_1.shtml</p>

<blockquote>

<p> But <em>he pitched nine innings of no-hit baseball</em> before the game went into extra innings and he was finally taken out for a reliever</p>

</blockquote>

<p><br /><br>

Nope! He gave up "two whole hits" in the nine innings he pitched. Cain is worth every $$ as you already know. He gave up zero runs, not zero hits. Cliff Lee pitched ten shutout innings over the Giants. Both are stupendously GREAT pitchers.<br>

<br /><br>

The 5D3 and D800 are *stupendous* cameras.</p>

<blockquote>

<p>Let the wailing and gnashing of teeth by Canon shooters begin...</p>

</blockquote>

<p><em>What? A Nikon camera performs better than a Canon Camera? Oh my. Now I'm sure the world will come to an end at the end of 2012. Canon will simply not be able to stand the shame.</em></p>

<p>How did I do?</p>

<p>Happy shooting,<br>

Yakim.</p>

<blockquote>

<p><em>Nope! He gave up "two whole hits" in the nine innings he pitched. Cain is worth every $$ as you already know. He gave up zero runs, not zero hits. Cliff Lee pitched ten shutout innings over the Giants. Both are stupendously GREAT pitchers.</em></p>

</blockquote>

<p>Ken is obviously even a bigger Giants fan than we are at my house! You got me Ken - it was "no runs" but not no hits! He's still one heck of a pitcher. </p>

<p>Sounds like we agree about baseball and about cameras in this case... :-)</p>

<p>Dan</p>

<p>...and now they have the D3200! Come on Canon let's get off those laurels!</p>

<p><em>Regarding the claimed 14.4 EV dynamic range of the D800 - since the raw files are at best 14 bit "deep", to achieve 14.4 stops the highlights must undergo some kind of compression. </em><br>

<em><br /></em>No. The DXOmark "print" data (from which all their scores and rankings are calculated) are calculated from images that have been resampled to 8MP to facilitate the comparison between different sensors. At the pixel level, the D800 DR at ISO 100 is 13.23EV so this explains the apparent discrepancy you're seeing.<br>

<em><br /></em><br>

<em>This probably disconnects SNR from dynamic range, so SNR ratio could be similar to the 5D Mark 3 yet the DR could be different.</em><br>

<em> </em><br>

The DXOMark SNR figure is evaluated for a 18% gray target. SNR depends on how many photons you record; it is different for each luminosity value. Dynamic range is how many stops below saturation does the SNR gets reduced to 1. Thus it is a measure of how dark the extremely dark areas of the image are while still containing (some) information. Now, why do they show different behaviour for the two cameras? SNR for a medium gray target is mostly determined on how efficiently the sensor collects and detects photons (quantum efficiency or QE). Dynamic range is dependent on QE, but it also depends on the ability of the electronics to read small signals. This latter thing is what the D800 excels in, while also having a very good QE.</p>

<p>In practice obviously, both cameras are just fine for making images. I wouldn't get too hung up on the quality of an individual camera's sensor. There are different lens lineups for Nikon and Canon and other areas of advantages and disadvantages. Look at the whole, instead of just one component. I'm sure that Canon will catch up soon enough (before Nikon has some lightweight long glass with high quality, or a 17mm PC-E). What's nice is that sensors are fast approaching theoretical ideal sensors so we can all put these things to rest. Or we could just decide to do that and get on with our photography.</p>