< ! --Digital window verification 001 -->

Another Nikon D700 vs. Nikon D800 high ISO comparison

This is another high ISO comparison between the Nikon D700 and a pre-production D800 (see also this related post). Here are details from the reader who shot the images (click on samples for larger view):

Here are some noise comparisons I have made in all ISO levels between D700 & D800 from ISO1600 to ISO 25600 with Nikkor 24-70mm f/2.8 lens at f/4. The crop area is in the almost extreme left of the frame and does not include the focus point, so it is not for resolution comparison, only for noise. The D800 files are downsized at 57% in PS to the D700 size, then cropped to 1:1 for comparison as same size pictures. I used RAWTherapee because ViewNX uses the noise reduction settings of the cameras. As ICC profile for D800 I have selected D7000 profile.

Another set of Nikon D800 samples at different ISO can be found on flickr.

This entry was posted in Nikon D700, Nikon D800. Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • cb

    I was wondering how some “old” glass would perform on a D800. I guess, the performance on a D7k would give an indication…
    Does anyone have an opinion on the
    * Nikkor 24mm f/2.8D
    * Nikkor 50mm f/1.8D
    * Nikkor 85 mm f/1.8D
    I am satisfied with these lenses on my D300. I am thinking about ordering a D800 and would be happy if I could use that glass…

    Chris

    • Nutt

      If those work on D300, they will work on D800. However, you might notice some imperfection in the corner of images taken with those lens because DX camera only use the ‘sweet spot’ of FX lenses so most people wouldn’t notice those little imperfections in the corner. My suggestion, just put those on D700 or any FX camera that you can get you hands on, took some test shots and take a look on your computer :)

    • Sly Larive

      There are not the type of lenses you’ll be using at very small apertures, except maybe the 24. If you’ve used your lenses at f/8 and so forth, then worst that will happen is you’ll run into diffraction problems roughly one stop sooner.

      Think about it, your D300 at f/8 uses a much smaller portion of your lens than the D800 ever will at f/6.3 and below. Its a gross generalization and not entirely true but as a rule of thumb it makes sense. IF one ever runs into diffraction problems, them downsampling will most likely fix the problem. At that point you’re losing resolution you can’t get anyway due to diffraction. Its a technique similar to what Nokia is proposing with their new “41 MP sensor”…

      • Patrik

        All lenses will work the same on a D800 as any other FX camera. The only difference will be that you will resolve more detail with the higher resolution. This also includes motion blur, lack of sharpness, and lens aberations. The sensor will not limit the detection of these imperfections any more. If you want the most from the D800, then technique will become more important, but your image will not become blurier just because you shot with a D800 as oposed to a D700! (though it will become more obvious if you pixel-peep)
        On the subject of lenses: I have had all 3 of those. The only one that I still own is the 85/1.8. Stopped down to about f/2.2 it transforms into a precision tool! The 24 is OK in the centre, but not in FX corners, even stopped down. The 50/1.8 is very soft until about 2.8-4 on a D700, you can guess how it will be on a D800. Better off getting a 60 micro (or my favorite: the 45/2.8PCE). At least at f/2.8 it is usable!

    • Swiler

      I’ve heard that lenses are sharper on FX because the sensor isn’t cropped and therefore not a blowup of the resolution of the lens. The higher resolution of the sensor might resolve more defects too, but the size of the pic should make them even better than on DX I think.

    • Andy Cam

      Hi Chris – I own and use all three of these lenses on my D700, though only one would I recommend: The 85mm. Pin sharp, fast focus – great even wide open (though I needed to offset the AF finetune to stop backfocus).
      50mm is AWFUL wide open, but fine closed down a few stops.
      24mm is disappointingly soft for a fixed lens, on digital or film, at ‘fullframe’. I’ve tried 2 and they’ve both been far from the quality of the older Ai lenses.
      So 1 out of 3! Nikon: must try harder!

  • Charles

    I am sure that all but one of my Nikon lenses will work great with the D800. It will be interesting to see how the Nikkor 80-400 VR and Sigma 150-500 work out. I currently use them primarily on my D300′s.

  • Natan Lorenzi

    Does anyone know when D800 will be avaliable regularly? I’ve heard this camera will not be regularly on the market until 2013. I need know that, cause I plan go to USA, just to buy one, in my vacation (August). But if this camera will not be available B&H, Adorama, Ritz, Best Buy, Amazon, I don’t go to USA.

    • Sly Larive

      There was a post a week or so ago where it was estimated that Nikon could produce 30,000 D800 per months. Not to say the market will be flooded, but I think it should be available by then.

      • Natan Lorenzi

        Oh, great! But a friend calls to B&H and a seller tells him that the B&H will recive only 20 cameras in March.

        So I need one. Here were I live, D800 will cost only $ 6000.

        • hlritter

          Natan, I placed an order through Amazon last week and they tell me it will ship March 20. Do you have a friend in the US who you can visit? Buy it through Amazon (or whomever) and have it shipped to the friend. Of course, if the ship date is correct, the camera should be easily available when you visit in August.

  • http://www.seanmolin.com Sean Molin

    Most situations I need high ISOs I can afford to down sample. In most portrait/event/wedding situations it won’t hurt at all to go down to 18-24mpx. The ones where I want them big, like landscapes, I tend to be prepared to shoot long exposures at base or close-to-base ISOs anyway.

    Ask yourself “what type of subjects do I regularly need to print above 20?”

    • burgerman

      Of course in those situations where you need to downsample you COULD just shoot in 9mp or 20mp instead and save yourself the bother as the camera does it for you. That QUAL button on top of this camera has much more use than on previous nikons..

  • KnightPhoto

    @James

    I’d still get the D4 and one D800. Sure you lose the identical camera benefit, but you gain flexibility in extreme light (D4) or high-res (D800). Anyhow I’m getting the D4 first and scheming how I can sell other cameras or gear to add the D800(E) sometime as soon as I can – looks GREAT!

  • Hubbel

    What I absolutely don’t like is the fine grain visible even at ISO 100 in smooth middle and darker areas. I remebers me to my D300 which was great, but not so clear and smooth like my D700 at lower ISOs.

    • burgerman

      Yes, if you printed it 70 plus inches wide, as per your 100 percent view on screen, you would see that…

      Not that I can see it anyway mind!

  • pc

    Were these shots made on a tripod or hand held?

    • Diol

      I have already answer “hand held” cause I had no time. But the speed was from 1/800 to 1/8000, so it dosent matter. Is a noise comparison, not resolution.

  • noal

    Well, I don’t know about you, but to me the D700 images looks suspect. Yeah D800 samples are really better, but I have seen and processed some high iso raws from D700 and they don’t look like this. Even without noise reduction. And I’ve never seen this kind of pattern artifacts on them, it looks like they are oversharpened or something. What do you think?

    • Hom Thogan

      I shoot with multiple D3s and D700′s and to tell you the truth their results aren’t on par with what I get everyday shooting with these cameras: I don’t use CaptureNX either but I don’t get that level of chroma noise he’s getting (neither with the D3s nor with the D700).

      Something smells like BS on that test… Besides if you are downsampling noise gets hidden

      • Diol

        What is BS?

        BTW, I also notice that the D700 samples were bad. But this may be because I have not apply any noise reduction other that 10% default that RAWtherapee applies which has no visible effect at all. As I told, I wad not alter any controls in RAWtherapee, I have leave all in default except the ICC profile. I know that D700 will became better with some NR, but I just wanted the same conditions with no NR.

        Another thing that I have already declare is that I have do also the comparison with D700 upscale x1.73 and D800 original, but the conclusion was the same, the difference between cameras was the same. Next time I will do a similar comparison I will chose to both upscale and downscale to the geometric middle. This will be useful if the differences are small, but in this case, the difference in Hi ISO was big. Anyway in the future I will improve the procedure.

  • Sly Larive

    I agree, but my point is, you’re not actually LOSING resolution, just not getting the most out of the D800. I feel this blur /diffraction issue of the D800 is overstated. Sure, the D800 will be more sensitive to camera shake, IF and only IF you want that 36MP out of it. It is more sensible to camera shake than a D700 would, if you downsample to a lower resolution, absolutely not. Very few people are going to use the native 36MP as the “end result” either via prints or some ultra expensive display. When viewed on a computer monitor, or a 8×10 print, the D800 won’t be any more sensitive to blur than a d700…

    Unless you’re a pixel peeper, or a heavy, heavy cropper.

    • Hom Thogan

      Why do you think Nikon released the D800E? because the D800 is affected by diffraction, I have had the chance to play with a preproduction D800 and Diffraction by f/11 is really noticeable in 8×10 prints, by f/13 to f/16 you notice it even more…

      It was the same with the D7000

    • Diol

      Absolutely right Sly. I will say that “D800 under any circumstance will give from at least similar to much better resolution than D700”

  • Dig

    For information purposes.

    Power Retouche has an ‘Anti-aliasing Filter’ plugin which works with a multitude of OS and programs.

  • http://Www.lake-district-wedding-photography.com Adam Naylor-Whalley

    I dont need 36 mp, but I am surprised by the high iso comparison. My D700 seems to handle noise better than that. Maybe it is different in working practice.

    • Diol

      The samples above are RAW without noise compression. Also this is a dark area of the frame where noise is more visible. The important think is that the post process is the same for both D700 & D800, in fact no process except for the selection of D7000 ICC profile for D800 as there was not D800 ICC profile in RAWtherapee

  • burgerman

    My D700 doesent get close to this:

    http://www.wheelchairdriver.com/D800_noise.htm

  • Gino

    ISO 6400 video D4 v. D800

  • Michael S

    Not sure how this test should be done but it seems to me you should be comparing the same number of pixels not the same sized picture for a noise test no ?

    Also can someone make a black picture part of noise tests. Black ought to be easy to compute but my Sony NEX 5n which according to all the reviews has amazing low light performance does horribly noisy blacks. I think they have probably over boosted the sensors to get the higher ISO.

  • seb

    a few examples from one of the polish website http://www.fotopolis.pl/index.php?n=14565&p=0

  • F64Photo

    The way I would test cameras is to make the same size prints with both cameras with the same lens side by side, as you would do in a real life situation. I do not care about the noise per pixel, I DO care about the noise per print, however. At any rate, why compare the D800 with the D700; they are for two different audiences. Why not compare the D700 with the D4?

    • Diol

      Because I wanted an answer to the question “I have the D700. If I change to D800 for better resolution, do I have to sacrifice the low light performance of my D700?” I believe that this is a question of many D700 users, it was my question too. I use my D700 mostly for weddings but I do also landscape. I need the extra resolution for landscape and Hi ISO performance in events because I do not use flash. I have test D4 to, but not in direct comparison. Of course D4 is even better in Hi ISO, but the budget is out of question for me.

  • Dmitry

    Thanks for the comparison. Noted comments that 100% to 57% should be ok, because of the print size etc.
    Anyway please make 100% crop to 100% comparison.

    • Diol

      100% to 100% crop is meaningless between different resolution sensors. Absolutely meaningless and dangerous cause it will lead to wrong conclusions. The 100% crop is usefull, not for comparison but for evaluation of the qualities of noise. Not side to side, but alone and only from very experienced and mature person.

      All along in the film times, the comparison between two films for resolution and noise are made between two identical size prints with a loupe lens. Can you imagine if someone tries to compare two prints in different print size this way? It will be ridiculous. I will not do that.

  • F64 Photo

    Diol.

    I TOTALLY DISAGREE WITH YOUR STATEMENT.

    When you compare say, Tri-x grain with Plus-x grain (two different resolution sensors), you make two identical photos with the same lens and camera and compare same-size prints.

    The same holds true for digital images. After all, that is what you are looking for is better PRINTS. You really do not care what the individually grains, or pixels, look like! So 100%, side-by-side comparisons are the way to go!

    The only dangerous conclusion you will come to is that Plus-x produces finer grain!

    I fail to see what you want to prove.

    • Diol

      You didn’t understand at all what I have said, or my poor english makes you misunderstood.

      100% or 1:1 magnification in digital means that the 36 Mpix file when print is x1.73 the dimensions of the 12 Mpix file. For ex like to compare a print from TriX at 34,6 cm width with a print from HP5 at 20 cm width. It will be silly. The same is with digital, you can not compare two prints of different size. Comparing 1:1 pixel to pixel (that is 100% magnification), is the same like to compare two different size prints.

      Maybe you want to say the same thing but didn’t understand me. Maybe you didn’t realize what I mean by “1:1” & “100%”

  • burgerman

    I will try and put it simpler for those with little grasp of physics.

    A 12MP image on your computer screen at 100% is 40 inches wide.
    A 36MP image at 100 percent is HUGE at 70+ wide and about 40 deep!

    To make it the same sized “photo” to compare the result you have to shrink it back down to 1/3rd the area, so it is the SAME size as the 12MP one.

    Or would you view a finer grain film by enlarging it more??? Er no…

  • F64photo

    Let’s start out with several conditions related to pixels, resolution, and noise:
    1. Larger pixels produce less noise, because there is a greater area to accept light. Since the D700 has larger pixels you would expect less noise from its prints.
    2. The noise in a sensor is fairly constant over different ISO speeds.
    3. As you increase the ISO speed the signal gets lower due to the decreased light, and since the noise remains roughly the same, the signal to noise ratio decreases, and therefore the noise is more visible. Therefore, you would expect more noise from higher ISO speeds in both the D700 or the D800 camera.
    4. The noise comes from randomly-exposed pixels whose signal comes from random electrical currents that “expose the pixels with electricity, and not light that comes from the image. Since this noise comes affects pixels individually, you can expect the noise to be more noticeable if the pixels are larger. Therefore, the D700 with a same-size print would have larger pixels and larger evidence of noise.
    In any scientific experiment, you make comparisons by changing just ONE variable at a time. In this case, it is the resolution of the print that we are after, to show what effect the pixel size has on image noise, or the number of Pixels Per Picture. So, let me explain:
    To follow your logic, using the film analogy you would enlarge the Plus-X film enough so that the grain of the Plux-X film was just as bad as that of the Tri-X and you would arrive at the absurd conclusion that they both had the same grain properties. Obviously, this is not true. With this situation, of course the Plus-X negative would make a larger print. So what to do??
    In the film realm, keep the enlarger at the same height and make enlargements at identical magnifications, with the same lens and paper, to compare the grain. Whether you compare RMS granularity under a high-powered microscope of the psychophysical effect of graininess using an enlarger, the physics and optics of the test require that you use the same aperture and the same magnification, or else you arrive meaningless figures. You have to compare apples with apples.
    So, in digital terms you take the 36 mp full-frame image and make a 10 x 15-inch print, and also make a 10 x 15-inch print for the 12.2 mp full-frame image and you compare them. So, naturally, you have roughly three times the number of pixels in the 36 mp image, or the square root of 3 = 1.732 Pixels Per Inch. Of course, the 36 mp image will look better because it has more information in the print, or it has finer grain, or it has better resolution.
    Now, remember that pixels are square and dots are round, and that it takes many printer dots to make a pixel.
    My Epson 3800 printer has a DOTS per Inch (DPI) value is 5,760 DPI in the Super Photo mode. At 300 PPI this gives you 5760 ÷ 300 = 19.2 dots per linear pixel dimension or 19.2 x 19.2 = 368.6 dots per pixel (DPP).
    So, for easy figuring, lets use the one-inch side of the sensor and make a 10X enlargement (10”x 15”).
    A D700 camera has:
    2,832 pixels on the one-inch side and 4,256 on the one and one half inch side. For a ten-inch enlargement, you would have 2832 ÷ 10 = 283.2 Pixels on the 10-inch side of the print.

    A D800 camera has:
    4912 pixels on the one-inch side and 7360 on the one and one half inch side. For a ten-inch enlargement, you would have 4912 ÷ 10 = 491.2 Pixels on the 10-inch side of the print.

    D700 camera:
    5760 Printer Dots/Inch ÷ 283 Camera Pixels/Inch = 20 Printer Dots/Pixel side in the print, or 20 x 20 = 400 dots per pixel in the print (“grain size”).
    D800 camera:
    5760 Printer Dots/Inch ÷ 491 Pixels/Inch = 11.7 Printer Dots/Pixel side in the print, or 11.7 x 11.7 = 138.9 (“grain size”)
    Therefore, since the pixels “grains” in the D700 print are larger (400 dots in area) than those in the D800 camera print (139 dots in area), the D700 has higher resolution because it has finer “grain” just like the Plus-X print. All this means that you would have to enlarge the D800 print 1.73X in order to make the “graininess” of both prints similar, if you chose to do this for some reason. Now, this means that if you have a noisy pixel, it would be larger in the D700 print than it would be in the D800 print, so the grain would be more visible in the D700 print even if the noise in both cameras were equal, and my guess is that they do not have the same noise characteristics. So, for the D700 to have less noise in the 10 x 15” print, it would have to have even less noise than the D800 camera.
    So, you can see that the “grain size is 20 ÷ 11.7, or 1.7X finer grain or detail from the D800 camera, which is what we would expect from the pixel count of each camera, i.e.:
    D700 camera 12.2 mp
    D800 Camera 36 mp
    36 mp/12.2 mp = 2.95 more pixels in the D800 camea.
    To convert it to a linear mp value √2.95 = 1.71 or fairly close to our 1.732 figure above.
    To quote you, “Can you imagine if someone tries to compare two prints in different print size this way? It will be ridiculous. I will not do that.” I agree, but you should not shrink the size at all, just leave it alone and let the printer do its thing. Since the printer is the same for both images, we avoid one more variable. BTW, you would not shrink the SIZE by 3X, but by the square root of 3 (1.732) to get the liner magnification and not the area! The measurement is in pixels per linear inch, not area. The only scientific way is to compare prints of the same size. The D800 will have more pixels in the 10 x 15” print, and the D700, or other camera, will have fewer pixels in the 10 x 15” print, but that is jus the physics of the problem that we a re stuck with.
    If you took a wedding photo of a bride in a white gown, and a groom in a black tuxedo, with both cameras and made a 10 x 15” print, and viewed them both to see which one you liked better, that would be a good test, because that is the way you do it in practice.

    • burgerman

      When you do this, you see exactly the same result as you do by resizing the image to match the D700.

      Because the printer (or its driver) resize the images to the SAME OUTPUT SIZE which is exactly the same as resizing the D800 down to 12MP so it matches the D700 output size. You could do the opposite and upsize the D700 image if you want. Or make them both meet in the middle at 24 MP…

      Net result is always exactly the same, and the same as a print (or printer driver resize) in that the output size matches, the D700 shows less detail or more noise, or both. By a very noticible margin. The bigger the print the clearly more this difference shows up to the eye.

  • compuinfoto

    And what about my lenses I use now on my Nikon d300,

    105 mm sigma macrolens
    50 mm 1.4 nikkor and my
    18-200mm nikon

    Will they be ok for the Nikon D800

  • burgerman

    Viewed at 70+ inches across on screen, at 100 percent, is like looking through a microscope. These lenses will not be as good as some under this condition…

    But viewed at the same size as you look at them now they will look exactly the same, or slightly batter than they do already on your existing camera… They dont suddenly get worse just because you have more pixels!

    They will work at least as well as they are doing now, printed the same size they will likely look slightly better. But a 36MP FX cameras ultimate output is wasted with these lenses.

  • F64photo

    What you say about print size seems to be true until you actually do it. The D800 print looks NOTICEABLY sharper! True the printer makes the prints both to the same overall 10 x 15” size OF THE PRINT NOT THE PIXELS, but the D800 crams more pixels into that size, therefore, it is sharper. See my comment on dots vs. pixels.

    Also, I thought that my D7000 was not very good until I started using my Micro Nikor. And since the D800 is essentially two D7000 sensors glued together, I expect to see the same difference between so-so lenses and god high-rez lenses.

  • F64photo

    P.S. when I change the print size in CS5, I DO NOT click on RESIZE, maybe that is the difference!

  • babeliak

    There is set of studio test photos from D800 and comparision to D700 at
    http://www.imaging-resource.com/PRODS/nikon-d800/nikon-d800A7.HTM

  • babeliak

    high-rez lenses?!
    is that a JOKE?
    Lens is sharp or not,
    it has NO rez … AT ALL !!!
    the more resolution has a camera,
    the sharper lenses it needs, but HIGH-REZ lenses? … c’mon :D

    • Roy Shi

      Unfortunately, a lens does have a resolution limitation. Simple as that.

  • http://kennethsolfjeld.blogspot.com/ Kenneth$

    New pics from their test setup. Compare for yourself. I think it looks good for the D800: http://www.dpreview.com/previews/nikonD800/7

  • Back to top