Methodological Validity of Comparisons of Canon 5D II and Nikon D700

[elliot1]

<p>Additional advanced processing of the above images would produce even better results.</p>

<p>While I find the various sites showing comparison photos offer valuable information that can assist in decision making, the results can often be misinterpreted or may require some consideration as to what the true differences between the various test shots mean. Landrum's example illustrates this point well. Often a small amount of PP can easily adjust for the differences in images and negates any differences. <em>I may do a D3/D700 vs 5D Mark II side-by-side comparison in the future.</em></p>

<p>At this point in time, a tripod and low ISO is still a photographer's best choice when photography of a low light stationery subject is required. Dynamic and color range drop quite a bit as the ISO increases above ISO 1600. The DXOmark site illustrates this well.</p>

[Landrum Kelly]

<p>Thanks, Elliott. I was hoping that you would weigh in again. I went to the DXO site last night and was favorably impressed by the results and the comparatively low cost. I shall have to look into that further.</p>

<p>Back to existing comparisons, I also went browsing and found this on Photography Bay:</p>

<p>http://www.photographybay.com/2008/12/27/canon-5d-mark-ii-vs-nikon-d700-in-depth-iso-comparison/</p>

<p>There are also some Ken Rockwell comparisons, but I haven't rechecked those.</p>

<p>Thanks to everyone for their contributions. </p>

<p>Elliott, what you are telling us is that "it ain't over till it's over"--until the post-processing is done. I'll buy that.</p>

<p>--Lannie</p>

[fjp]

<p>Yes, with some postprocessing these H2 images can be made usable.<br>

For instance if you want black and white photos, you can get better results<br>

since most of the 5D2 noise seems to be chroma noise.</p>

<p> </p><div></div>

[Landrum Kelly]

<p>That's an awesome shot, Fernando. Coincidentally, while I was comparing some Nikon v. Canon high-ISO shots last night, I was also struck by the high chroma noise in the Canon.</p>

<p>Not that we were not warned by dpreview.com. . . .</p>

<p>Since the 5D II has variable noise reduction, what happens when one reduces it or turns it off? Does the chroma noise go crazy, or does the lumina noise hit you in the face? I am presuming that it is the latter, since Canon's bludgeon-like noise reduction must be being directed toward the suppression of that kind of noise.</p>

<p>There has got to be a better way.</p>

<p>--Lannie</p>

[elliot1]

<p>Landrum, your thinking about turning noise reduction off is on track with my thinking. There are many exceptional noise reduction programs available that offer full control to the user, which, in the end, will likely give the photographer the best possible photograph.</p>

<p>The comments that appear after the photo comparisons (referring to the last link you provided) give a lot of insight into the difficulties in testing images from different cameras and lenses.</p>

<p>Post processing levels the playing field of most cameras in the same class and can greatly affect/impact test results to the point that some may not even be valid. For example, in the sample test crops above where Landrum 'fixed/equalized' the Mark II photo, the initial observations made are no longer valid.</p>

<p>Is the D700 and excellent camera at high ISO? Yes. If the 5D Mark II also an excellent camera at high ISO? Yes. Which is better? After PP, I don't know that one can <em>clearly </em> win out over the other.</p>

[Landrum Kelly]

<p>Thanks, Elliot. Thanks also for keeping us on topic. You've pulled us all the way around to the original question in a way that I never could have anticipated.</p>

<p>Gotta get that DXO software. . . after I get the camera!</p>

<p>--Lannie</p>

[lalon_karim]

From imaging-resource.com:

 

About 5DMkII: http://www.imaging-resource.com/PRODS/E5D2/E5D2IMAGING.HTM

"At higher ISOs, the 5D Mark II's images held together surprisingly well. For 13x19 inch prints, ISO 6,400(!) was about the limit, as the noise in shadow areas began to be noticeable at that size, and detail was obliterated in areas of subtle contrast. This is really a remarkable performance from a camera with this much resolution: Comparing their images side by side, prints from the Canon 5D Mark II at ISO 6,400 were quite comparable (similar or slightly higher luminance noise, lower chroma noise, a bit softer overall) than those of the EOS-1Ds Mark III at ISO 3,200"

 

About D700: http://www.imaging-resource.com/PRODS/D700/D700A5.HTM

"High ISO images were really extraordinary: Together with the D3, the Nikon D700 clearly leads the field in high-ISO performance, thanks to its large pixels, CMOS sensor technology, and Nikon's excellent noise-reduction processing. D700 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 at that size, there was 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 D700's noise at ISO 6400, making it even less apparent than noise patterns from many other cameras with similar "grain" size. Shot under daylight-balanced lighting and printed at 13x19 inches, the D3's ISO 6,400 shots were softer and somewhat noisier than those at lower ISOs, but the results were still pretty amazing."

[lalon_karim]

Here is an interesting link, though it may not look like directly related to the current topic, it actually is:

http://www.dxomark.com/index.php/eng/Insights/More-pixels-offsets-noise!

[jamie_robertson2]

<p>I've been saying this for ages. Yes, when viewed on screen at 100% the D3 and D700 are superior to the 5D2 when it comes to noise control. But, the images from the Nikon cameras are a lot smaller.</p>

<p>So, <strong>when making prints</strong> the images from the 5D2 almost always turn out better <strong>AT ANY ISO. </strong>If you make a large print eg 20x16, the Nikon images have to be enlarged a hell of a lot more than the Canon ones and therefore lose resolution and the noise at high ISO becomes more noticeable. Likewise, if you make a smaller print eg 12x10, the Canon images have to be downsized considerably which makes any difference in high ISO capability vanish due to the shrinking of the much larger file size.</p>

<p>I think in practice, all things considered, <strong>when making prints</strong> the 5D2 and the Nikon D3 and D700 are fairly evenly matched when it comes to noise control. I am not biased towards any manufacturer but I would honestly rather have the 5D2 than ANY Nikon camera (D3x included) for the image quality alone. Yes, the Nikons have better weather sealing, better frame rate and superior AF but when it comes to IQ the 5D2 is superior.</p>

<p>Look at the D3x. Its resolution is only marginally higher than the 5D2 but it tops out at ISO 1600. What does that tell you? It tells me that Canon currently have the high ISO advantage. If Canon had been just a little more sensible and made the 5D2 with a 15MP resolution it would have set a completely new standard for low light photography. Perhaps they will consider this with the 1D mk 4.</p>

[lalon_karim]

<p>Errr... Jamie, thank you for your response. but I really dont think that for the same sized print 5D Mk II has better high ISO performance than D3X. There is a good chance that it is actually other way round. (I dont have any proof, but you may see this link: <a href="http://www.kenrockwell.com/nikon/d3x/iso-6400-5d-mark-ii-d3.htm">http://www.kenrockwell.com/nikon/d3x/iso-6400-5d-mark-ii-d3.htm</a>). D3X has recommended ISO upto 1600 because it is designed as a studio and landscape camera - not necessarily because it has worse high ISO performance.<br>

Apart from this, I agree with what you wrote.</p>

[jamie_robertson2]

<p>Lalon,</p>

<p>I stand corrected. I have never used the D3x and I wasn't aware that it had an ISO 6400 option. You are indeed correct.</p>

[Landrum Kelly]

<p>Actually, Jamie, the Nikon D3X can be boosted to ISO 3200 and 6400--and, if the dpreview.com pictures are to be believed, it does very well at both:</p>

<p>http://www.dpreview.com/reviews/nikond3x/page31.asp</p>

<p>(I always assumed that Nikon did not go for 12,800 and 25,600 on the D3X because shooting in near darkness is really not a strong marketing point for a studio/landscape camera--and who wants to go stumbling around in the dark taking flashless snaps with an $8,000 camera?)</p>

<p>Interestingly, on this comparison dpreview.com did what we have been asking for: resizing files before trying to make final comparisons, although in this case doing so with the D3X and the D3/D700.</p>

<p>As for the rest, Jamie, you may well have it right, but I think that Elliot is on target when he says that IQ of both brands at high ISO comes down to post-processing--and, let's face it, Canon's in-camera processing in the 5D II is arguably too strong on luminance noise control, but not strong enough on chroma noise control. If one shoots RAW, however, and uses third party software (<em>e.g</em> ., DOX), perhaps the noise control problem can be remedied.</p>

<p>Since I have yet to hold any of these cameras in my hand, however, I am operating in a personal empirical vacuum. My equipment is strictly "last generation" (5D and 1Ds II). My personal opinion carries no weight at this point.</p>

<p>I do want to have the 5D II in my hands, however, and when I do I intend to put it through the wringer at high ISO on my Epson 3800 (17" wide). I doubt that I will ever try to have the files printed any larger--unless I go Fernando's route and shoot it at high ISO with black and white post-processing to take care of the chroma noise (pretty darned effectively, I dare say!). Taking and printing the Nikon files I will have to leave to someone else.</p>

<p>I've been on this "low-light kick" since the winter of 2008, when I went out shooting at night with a mid-sixties vintage Canon 7 rangefinder--with the 50mm f/0.95 lens. The light-gathering ability of the lens was awe-inspiring, but the image quality just was not that good when the lens was shot wide open--and, if one did not shoot it wide open, what was the point of having a 50mm f/0.95 lens in the first place?</p>

<p>Regardless of which brand one shoots, when it comes to low-light photography, we have come a long way--and digital has made it all possible. As a life-long resolution freak (in my case, splitting double stars with telescopes when I was thirteen) and now a low-light freak, I can only say that these are great times to be alive, and I am glad to have lived long enough to see it all happen.</p>

<p>What some of you will live to see I can only imagine.</p>

<p>--Lannie</p>

[Landrum Kelly]

<p>Sorry, guys, but on my very long post I cross-threaded both of you. Sorry for the redundancy.</p>

<p>Lalon, I haven't been trying to ignore your posts, but you have said it all so well that you left me nothing to challenge or add.</p>

<p>--Lannie</p>

[lalon_karim]

<p>This is just a personal opinion, but I do not like post processing that much. Every camera should have a "standard" setting for color, contrast and noise reduction that should produce "perfect" result for printing straight out of the camera. For advanced users, who like post processing, there sholuld be other settings available (muted/ neutral and high/low noise reduction etc) or they can shoot raw. Just like we do not frame carelessly thinking that we will crop later- "perfect" outputs straight out of camera can give a lot of pleasure. </p>

[Landrum Kelly]

<p>Lalon, by "perfect" I presume that you mean "lifelike" or "realistic." I won't even get into the problems with that, except to say that, to the extent that photography is art, it surely has to do more than give a realistic portrayal of reality.</p>

<p>As for the present thread, what would that mean? It would mean that shots made in near darkness would be underexposed, not at all what most of us who have addressed these issues are concerned with .</p>

<p>--Lannie</p>

[jamie_robertson2]

<p>Landrum</p>

<p>I own the 5D2 and it is every bit as good as people say it is. A lot of the high ISO performance is due to extremely good in camera processing (and also good processing in DPP for example). But, whatever the reason... the fact remains that it is very good indeed in low light especially when compared to the last generation of DSLRs.</p>

[Landrum Kelly]

<p>That's good to hear, Jamie, 'cause I've now got one on order!</p>

<p>As for the original topic of the thread, I found this on dpreview.com dating back to last fall:</p>

<p>http://blog.dpreview.com/editorial/2008/11/downsampling-to.html</p>

<p>--Lannie</p>

[lalon_karim]

<p>Landrum, by "perfect", I didn't mean lifelike. I donno the proper words to define it, but it actually means something like - "can give excellent print in terms of color, sharpness and noise without further post processing". And it's true that photography is an art. A photographer does not try to show what we actually see. Instead he tries to show what he wants us to see. That is what art is about . And I myself is very fond of high ISO photography. Thank you for your opinion, Landrum.</p>

[lalon_karim]

<p>Just to clarify it further, I want my camera to give perfectly usable (for print) high ISO result without any post processing. This is what I tried to mean.</p>

[Landrum Kelly]

<p>I can't disagree with you there, Lalon. The issue of what the default settings ought to be always comes up, but I understand your point.</p>

<p>--Lannie</p>

[angel_bocanegra]

<p>5dmkII is the best, PERIOD!</p>

[philippartridge]

<p>The 'diglloyd' guy has a pretty bulletproof method for a rational comparison of the characteristics of DSLRs with differing MP counts. He also cites DxO in his analysis. Both parties are likely to be more trustworthy than DPReview's methodology, which is already under the gun from observers over at Fred Miranda.</p>

<p>Lloyd was looking into low ISO noise performance of the 5D2 and Nikon's D3x; his site has some pretty compelling evidence of the ugly noise characteristics of the Canon at ISO100, together with mention of the measurement of an additional 2 stops of DR delivered by the Nikon. It is not uncommon for landscape shooters to work up the shadows by two stops or more. Many lenses lose that much to optical vignetting in the extreme corners in any case, and the highlights' LV determine your shadows in any case. So it is far from an outlandish thing to do...And the 5D2 has some unsettling noise patterns, very 'digital', whereas the Nikon was super smooth, even under duress.</p>

<p>Being in the market for a big print-making DSLR for landscape work in extreme light conditions (Himalayan region) I have decided to await Nikon developments re a 5D2 competitor. 2 stops of dynamic range together with ultra clean low ISO noise RAW files is a quite persuasive, even compelling factor in this decision. Too bad, as Canon produce the 70-200/4IS, several other functional if not spectacular f4 zoom lenses, and can accommodate a host of fabulous low cost alternative lenses via adaptors.</p>

<p>Like a lot of shooters, I am sick of the merry-go-round and want a body for 2-3 years at least...and with the demands these high Mp DSLRs place on lens quality, lens apertures and shooting technique, we are closing in on an decent threshold of quality, a watershed in DSLR development, speaking as a 6x7 photographer.</p>

<p> </p>

[hawkman]

<p>These crops are from in-camera JPEGs thus the comparison is worthless to start with. Also to judge <strong>sensor performance</strong> you have to look at photos at 100% (i.e. one pixel and sensor = on pixel on screen), otherwise it is meaningless.<br />When making a print at certain size both photos have to be resized for print, but that is not how you compare sensor performance (which is what dpreview.com does).For example, you can down-sample the photos to 6 mpixels and all of the current cameras will be perfect with no visible difference. So dpreview.com are not comparing prints but native sensor noise performance.<br />Also from about 5 thousand shots I have snapped with my MKII and D700, high ISO is excellent on both with D700 having an edge over the MKII at native size, by the time you down-sample MKII to D700 native size they are very close but 5DII still has some residual color noise blotches left which you need to eliminate with NR sw. 5DMKII has more chroma noise in shadow areas in lower ISOs as well, but the sharpness and detail makes up for it.<br />At any case, the weakest link for 5MKII is not its sensor by any means but it is the lens, I did a test with D700 and 14-24 f/2.8 AF-S vs 5DMKII 16-35 f/2.8 L II, at 18mm f/8 and then printed the output at 18X12 and to my surprise D700 print was much better thanks to the 14-24. MKII print had very soft edges with color fringes that were quite visible to trained eyes, so I kept the 700 print and gave the MKII print to one of my friends who also bought my 16-35 ;) In order to make use of all 5DII's pixels you need an L prime or telephoto, I use it with my 400 f/5.6L, 70-200 f/4IS and 50 f/1.4 and you can make excellent large prints.<br />print from D700:</p><div></div>

[lalon_karim]

<p>I think nobody here will deny that D700 has a sensor that has lower noise in higher ISO than 5D Mark II. What we are trying here is to figure out what it translates into (lower noise 12MP vs higher noise 21MP sensor) in terms of output. Here is a quote from dxomark.com article: ".....Moreover, the measuremnt above corresponds to 1:1 viewing conditions on a screen. Except for image retouching and examining details, this is probably not useful way to view the image. A fit to screen viewing or a print on a given-size paper format is a more likely use case." And the noise comparisons in high ISO that are done in dpreview use JPEG, not RAW (personally I do not have any objection though).</p>