[NR] exclusive: Nikon D4s high ISO/low light comparison with D4, D3s, D800E and D600 cameras

Nikon D4s DSLR camera 7
d4_01Nikon-D3sNikon-D800-6Nikon D600
Terry Hansen did a very extensive and detailed high ISO/low light comparisons between the Nikon D4s (including RAW and small RAW files) and the D4, D3s, D600, D800E cameras. In this post I have included only the samples from ISO 6,400 and above, since this is the range where most readers will be interested to see the results. The rest of the test images are available on flickr (direct links: D4s (RAW normal)D4s RAW small | D4 | D3s | D600 | D800E). Here are the details:

With the release of the new D4s flagship DSLR, Nikon is touting substantial image improvements, especially in high ISO low light shooting conditions. They attribute these advances to the combination of their new EXPEED 4 engine along with an all-new sensor design.

Such claims are all well and good for the photographer who shoots in-camera JPEGs; these shooters rely on high quality internal processing for fast delivery. However, compared to previous Nikon DSLRs, what are the image improvements for the photographer who shoots exclusively in RAW? In largely bypassing the JPEG processing advances of the EXPEED 4 engine, does the new D4s sensor, in and of itself, provide significant high ISO improvements or better dynamic range compared to other Nikon sensors? And with the choice of the new RAW L (Large, 16.2MP) and RAW S (Small, 4MP) file formats, does either format provide an inherent advantage over the other for cleaner images?

These are the questions I decided to tackle. For this purpose, I ran a comprehensive series of test shots of a still life under controlled conditions. Here are the testing parameters:

  • Test multiple FX bodies: D4s, D4, D3s, D800e, D600
  • Shoot exclusively RAW (including RAW L and RAW S for the D4s)
  • For online viewing, convert externally to JPEG without applying any image adjustments or noise reduction processing
  • Use the same lens (a Nikon 24-70 f/2.8) across all bodies, set at a constant focal length (38mm) and aperture (f/8), shooting in Aperture priority mode
  • Assemble a still life featuring a wide dynamic range, including both bright specular and inky matte surfaces, along with areas of flat grey contrasted with highly detailed textures and deep shadows
  • Use a dim and challenging lighting design
  • Other than shutter speed, be sure that all relevant camera settings are identical across all bodies, including metering modes, and all internal processing (if such would even affect a RAW file anyway) options turned off or minimized when possible
  • Focus was always set on the head of the grey mannequin
  • As neither Lightroom nor Aperture can yet process the new RAW S format, I used Nikon’s own Capture NX-D Beta software for all JPEG conversions.

The resulting images are grouped and labelled accordingly. (They say “RAW” in the name, but that is a reference to the source; you are seeing JPEGs.) Compare shots across the several bodies used, and feel free to draw your own conclusions, adding your thoughts to the comments area. No laboratory measuring devices here, let your eyeballs be the judge—who do you think owns the crown “Low Light Champion”?

By the way, this is not necessarily intended as a test of resolving power. From this series of photos, while one can certainly get an idea of a sensor’s ability to capture fine detail across a range of ISOs, this test was put together primarily with high ISO and dynamic range evaluations in mind; another test, another day, with all those lovely charts and test patterns, is for someone else to take on. 🙂

And finally, many, many thanks to Nikon Rumors for allowing/encouraging me to assemble these photos to be shared! I’ve been a fan and follower of NR for years now, and have long benefitted enormously from this site. This is my humble attempt at paying it forward.

—Terry Hansen

ISO 6,400

Nikon D800E:

Nikon D600:

Nikon D3s:

Nikon D4:

Nikon D4s RAW S:

Nikon D4s RAW L:

ISO 12,800

Nikon D800E:

Nikon D600:

Nikon D3s:

Nikon D4:

Nikon D4s RAW S:

Nikon D4s RAW L:

ISO 25,600

Nikon D800E:

Nikon D600:

Nikon D3s:

Nikon D4:

Nikon D4s RAW S:

Nikon D4s RAW L:

ISO 51,200

Nikon D3s:

Nikon D4:

Nikon D4s RAW S:

Nikon D4s RAW L:

ISO 102,400

Nikon D3s:

Nikon D4:

Nikon D4s RAW S:

Nikon D4s RAW L:

ISO 204,800

Nikon D4:

Nikon D4S RAW S:

Nikon D4S RAW L:

ISO 409,600

Nikon D4s RAW S:

Nikon D4s RAW L:

This entry was posted in Nikon D3s, Nikon D4, Nikon D4s, Nikon D600, Nikon D800, [NR] Reviews and tagged . Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • Mikkoy

    I like the D3s output.

    • d800e_Shooter

      I concur. The D3S images look the best.

    • metsatsu

      Oh god, why did I sell my D3s for a Nikon vacuum cleaner? 🙁

    • Df Fanboy

      But, the Df mops the floor with anything else. DxO Mark said so.

      • French Fries

        The Frankenstein camera?

    • lera

      i like LordEels moron

    • Fiod

      yep, D3s have details, D4, D4S blurry

    • Reality Check

      I didn’t check all ISO’s because anything over 25600 is unsuitable for commercial use, but just looking at the chest of the mannequin at ISO 25600 demonstrably shows that the D3S has much less chroma noise than the D3S. If this test suggests anything to me it is that the low light D3S image quality is still unbeaten within the commercially “useable” ISO range.

      • freddy

        “demonstrably shows that the D3S has much less chroma noise than the” …D4S.

      • Well back in the film days anything over 800 iso was considered ‘unsuitable’ for commercial use….fast forward 10-15 years….

  • Joseph

    Thank you Admin

  • judge

    Yes, big daddy of comparing!

  • Rameses the 2nd

    To me D800E and D3S images look more contrasty with rich black. D600 and D4/D4S images look similar.

    It still amazes me how well D800/E holds up against these pussy mega pixel cameras.

    • inyourbase

      By it’s nature, more contrast is less dynamic range. Why would you want that in a Raw?? Add it in later.

      • Rameses the 2nd

        If you go to Flickr and take a look at large versions of D800E and D600 IS0 6400 images side by side, you will clearly see that D800 has more dynamic range than D600 (look at the details on the earten pot). However, D800 image definitely looks more vivid to me. It could be how Capture NX processed raw files from these camera.

        But the differnces are minor and all images in this test look good to me. I’d be happy with any of these camera in my bag.

    • MyrddinWilt

      I predict that the D800 will gain a RAW-S mode in the near future. And that might just finally get the D700 successor crowd to STFU.

      That would be a feature that would make some folk buy a D800 who have been sitting on the fence. And does not take a major redesign.

      A D4x with 48mp would have a RAW-S mode of 12MP. Which is more than enough resolution for most applications and probably won’t kill FPS performance too badly either.

      Conspicuously absent from the Nikon DSLRs right now is support for WiFi. The huge overpriced dongles are not remotely interesting to me and the small ones aren’t that good either.

      • Reality Check

        “I predict that the D800 will gain a RAW-S mode in the near future. And that might just finally get the D700 successor crowd to STFU.”
        sRAW only solves two D800 problems… huge files for post processing and fewer storable images on the CF Card. It wouldn’t resolve the low FPS problem because the camera still has to process the overkill sensor data with the additional step of binning it into a smaller sRAW file.
        The biggest problem for the D800, however, is not file size… it’s poor PDAF focus accuracy. To this day it still doesn’t focus as well as the 2007 era D700/D3/D3s with fast lenses shot wide open in poor light.

        • D800-fan

          As a D800 and a D700 owner, I have to say that my D800 gives IMHO better focus than the D700. This is using the same lenses ranging from the 12-24 F2.8 to the 200-400 F4.

          • Reality Check

            Note that I said “fast” lenses in poor light. What I mean by fast is f/1.4 to f/1.8 lenses.. Your f/4 200-400 at f/4 is definitely in the safe zone. Your f/2.8 wide angle is not only relatively slow, it also has a huge DOF likely to hide focus inaccuracy. I can see you getting better focus in good light with either of these lenses on a D800. The D700, however, stomps all over it in low contrast lighting with the 85mm f/1.4.

            • Eric Eikenberry

              I agree wholeheartedly, and in my discussions with Thom Hogan, he’s voiced similar concerns. At apertures wider that f/2.8, ALL of the new Nikons have issues with AF accuracy. Simply put, results with the same lens on the same body taken in successive order do not produce consistent results. I have a D300 with recalibrated AF system by a shop in Phoenix, and I’ve tried three D7100s and a D600. The D300 is simply more consistently accurate at wide apertures. Period. Even on older AF-D lenses where the camera itself is controlling the motion of the focusing system, there is just something weird going on in the newer AF system logic. Going to f/2.8 will mask the problem, though you CAN still see the boundaries of the plane of focus move forward and back from one shot to the next.

          • Heitschmidt

            Having had the blessing of shooting both the D700 and D800e, it has been my experience that both cameras focus perfectly; neither has demonstrated an edge over the other. I’ve never had any trouble with either camera in any lighting condition with any lens (mostly f/2.8 lenses).

            But that is beside the point of this test. My takeaway from the shots is that the much more cost-friendly D600 is incredibly close to the big $ bodies. If only I could convince my wife that I really NEED at D600….. 🙂

            • Tooma

              D800 is slightly better.

          • Tooma

            Agreed thank you

        • El Aura

          And given that the 12-bit small RAW of the D4s is only 8% smaller than the full-size, 12-bit, compressed RAW file (13.1 vs 14.1 MB), those space savings won’t be very substantial.

        • E-Nonymouse A

          Having never owned a d700 I can’t say about that BUT owning a D800E with fast prime lenses in low light it does focus good enough for studio use. If your expecting it to auto focus properly pointing up at at star in the night sky, good luck.

        • Tooma

          I disagree and I shot both for many years

      • paolo

        Yes… I think so, if d800s will have the same d800’s sensor, probably its raw-s files will be 9 MP…(16:4 for D4 and 36:4 for D800s..

  • The D3s has a hint more contrast and warmth… both of which are easy enough to match. Noise and general rendering are what are critical here, and I think the D4s at least matches the D3s. Tonality is personal preference and easily adjusted.

    • Declan Gallen

      exactly. I own a D4, D800 and D3s. I find the D4 to generally produce better colour. D3s is too magenta. This is easily fixed though as you say. The D4 is better in low light in real world shooting. Images are less grainy and seem to have more colour depth and detail. It also focuses better in challenging lighting conditions.

      The main reason why the D4 is better though, and the same applies to the D4s, is the ergonomics, especially when shooting portrait. The layout is much better. The ability to assign the REC button to ISO is a god send and something that was inexplicably left out of the D800 firmware.

      • Tooma

        D4 buffer yes! Battery life no! And why not 2 CF slots???

  • Aldo

    This is a much better test. Thank you

  • D4Owner

    Wow. The D3S is no slouch compared to D4 and D4S.

    • Marcel Speta

      I had an oportunity to shoot with D4 and D3s which i own. I decided that upgrade to D4(s) is not necessary for my needs. I would rather get D610 because of better DR and Resolution and swap cameras when needed. I am convinced that 24Mpx sensor is great and this test just prove it

  • Arch Brook

    I bet people saying D3s is best didnot open the links. Commenting through 500×300 img size is really funny.

    • Terry Hansen

      Agreed. Even though it takes time to load large photo files, you really must view the Original, full size images to get a full picture of what is happening.

      • Loose Lucy

        No, you must ‘print’ to get the full picture which shows tiny marginal differences between cameras, and zero differences when hung on the wall….but that will take all the fun out of pixel-level scrutiny.

        • Loose Lucy

          and thanks for doing the tests also. That’s a lot of work and it’s appreciated.

    • Totally Agreed!

  • Rob

    So If you don’t need Good Video and a huge buffer the D3s Is still the best camera If you take current cost. I found a “like new” D3s for about $3300 which sounds just about right for me. I’ll save the extra cash and go on a trip to Thailand or wherever the hell I want! I wish Nikon would invest in doing something truly radical and Make leaf shutter lens line up or a midum format camera.

  • peteee363

    it would be cool to add the df for the high iso’s as well. I wonder how that holds up?

  • jvossphoto

    Flicker’s interface flat sucks.

    • There was no better way to share those images – it’s just too much for one post.

      • jvossphoto

        No worries Peter….Probably more my bumbling than anything.

  • lorenzo

    Uhm… i might be blind or dumb but honestly looking at all pics here I can’t see any differences. How is possible to magnify them? Don’t have a Flicker account.

    • Click on the image here. That’ll take you to the flickr site. Then click on “…” which opens a menu. Then you’ll have the possibility to view/download the large image.

      • Eric Calabros

        So what do you think InTheMist? I like to know your conclution, (as a D800 owner?)
        btw, its first time I see DxO score irrelevant

    • Go to flickr – you can also download them.

  • drfilms

    What is really amazing is how well the little very inexpensive D600 holds up.

    • Yeah, I quite like the D600 output.

    • Naval Gunfire

      The D600 sensor is great (when it isn’t covered in oil). 🙂

    • CSIROC

      I’m constantly amazed by the sensor in my D600. Absolutely incredible. Only thing I’d really like is a better AF system, though, for the price (especially the price I paid), I really can’t complain.

      • Declan Gallen

        I was impressed with the D600 in normal lighting but as soon as you’re shooting in low light it struggles to focus. All the noise reduction in the world isn’t worth a damn if your AF goes hunting when the lights dim. Even the D700 finds focus better.

  • Rafa R

    The Nikon Df should have been tested along this cameras!

    • Terry Hansen

      I agree! Alas, I only had access to the cameras I used, otherwise I absolutely would have included the Df in the shakedown.

      Now, just in case a Df could somehow magically appear for me to borrow in San Francisco for an hour or so, I have not yet dismantled the testing set up, everything is exactly in place and I could add it to the test (and, frankly, would love to see how it compares myself). Any magicians out there with a Df in San Francisco?

      • MyrddinWilt

        The problem with testing the Df is going to be sample variability. The Df and D4s share the same sensor, they are sorted according to how well they perform. The best 20% or so go in D4s bodies, the rest go in the Df. So the Df bodies are going to vary quite a bit.

        The DxO result is not such a surprise as a result. Early Df bodies were being made when there was no D4s production.

        It would also be interesting to see how well the D800 performs at higher ISO settings but that would require setting shutter speed etc in full manual mode then pushing the output since 25600 is the highest stop.

        • Rafa R

          I do not agree, a Nikon Camera is a Nikon camera, not such thing as ¨sample variability¨ how do you know that they are ¨sorted¨ according to how they perform¨ Bullocks! a sensor is a sensor, checked with scrutiny at Nikon Quality Control and the Nikon Df is a respected camera (at least for Nikon) and it has its own assembly line, and quality control.

          I own a D4 and I just bought a Df, and I LOVE my Df, there is something about its IQ that I truly love, I shoot dance photography and I am surprised about the Df performance in low light, no complaints on the AF department either, that little camera is going to earn your respect and others , sooner or later, believe me.

          • Rafa R

            my own personal review on the Df you may want to check:


            • Jerry Friedman

              “batteries last for ever even tree times as much…”
              Your review could use a little editing.

            • Rafa R

              thank you for your advice, English is not my main language, I´ll try to improve

          • Tooma

            All electronics have sample variety however Nikon does not handpick sensors as stated above.

        • Tooma

          They do not share the same sensor.

      • Rafa R

        sorry, I live in Mexico, but if we were closer, Id love for you to test my Df for sure, I hope someone raises a hand a lends one to you, great work btw! cheers

    • Terry may be able to get a Df camera and test it as well. I will update this post with the images.

      • Rafa R

        nice! thank you

      • Cool, I’d really like to see that too.

      • 3Horn

        I’d be curious to see a D7100 in that lineup.

        No, it isn’t full-frame, but it is currently at the top of Nikon’s DX offerings.

      • 3Horn

        I’d be curious to see a D7100 in that lineup.

        No, it isn’t full-frame, but it is currently at the top of Nikon’s DX offerings.

  • Terry Hansen

    As the guy who conducted this test, the biggest take away I got from it was the very pleasant discovery that the new RAW S (Small) format does indeed dramatically improve the sensor’s ability under low light conditions. Having studied the full size original RAW files extensively on my monitor (and prior to JPEG compression), I would subjectively rate the high ISO improvement of the smaller RAW S format as being a full stop improvement over the normal RAW L (Large) format.

    Now, granted, using RAW S is at the rather significant expense of dropping from 16.2MP to 4MP, but if you absolutely must shoot pics of a black cat in a coal mine at night, RAW S absolutely blows everything else away!

    • This is especially visible in the ISO 409.600 test.

    • Thanks for taking the time. Very well done!

      • Terry Hansen

        Thank you, and you’re welcome! It was a labor of love. And a bit of madness thrown in. 🙂

        • Zoot

          “It was a labor of love. And a bit of madness thrown in.”


    • Eric Duminil

      Thanks a lot for your test. But isn’t this full stop improvement expected, mathematically speaking?

    • I like your little paper notes, telling which camera you are using, right in the photos. Great idea!

      • Mark

        that method is good out in the field with manual lenses too, negating the need for pen/notebook

    • wonderingpi

      If I wanted 4mp images id buy a 4mp camera not 16. Why spend 6.5k for a 16megs when you only use 4….

      • umeshrw

        You should read the article in full (Especially the footnote) before posting.

    • rkas

      Do you have any full sized examples? 🙂

    • peterw

      Remarkeble it seems indeed.
      Since these pictures come from the same sensor and the same pixels, this implies that the camera is using a smarter algorithm to interprete these pixels and combine the data in sRAW than the software that handles the RAW-normal to which it is compared. sRAW can not be an improvement to the sensor, physicaly.
      However, It is possible that the improvement is due to in-camera – in the sensor – preprossessing, gathering photon-hits from multiple pixels before amplifying and read-out to file. The lower amplification is needed, the better S/N gets.
      If sRAW is done after reading the indiual pixels, after amplification, there should be a smart way to get the same 4MP data quality from a 16 MP RAW externally in software.

      This is so exiting. Imagine the possibilities in three years form now of 64 MP sensors. One body, in which you can choose for a slow extreme resolution of 64 MP (enough to deconvolve all lens faults), or a superfast extreme low light 16 MP.

      sorry, I got carried away.
      Nice work, thanks.

    • Declan Gallen

      The thing is, resizing a D800 file to 4mp can hide a multitude of High ISO sins. Comparing SRAW to 36mp/24mp images isn’t a fair test, really. It would be interested to see how SRAW stands up to resized processed images from the d800 and d610

  • Captain Megaton

    Many pixels. So peeping.

  • lord eels

    why the hell did you use A mode? the shutters aren’t constant.

    • Terry Hansen

      I’m not clear what you are suggesting.

      For any given ISO setting, are there not only two variables involved in getting a correct exposure, i.e., aperture and shutter? I wanted to achieve identical depths of field across all shots and settings, so I chose to fix the aperture and let the shutter be variable. Are you suggesting I should have instead used a fixed shutter speed and let the aperture change for each shot, and to hell with a consistent image? What am I misunderstanding about your comment?

      • lord eels

        there is this thing. it’s called manual. you adjust things. your test is bunk because you used a semi auto mode. the shutters and apertures are not exactly the same for all cameras at a given ISO.

        • Terry Hansen

          Ah. So I am not misunderstanding anything. You’re just, well, unpleasant. Got it.

          I set the ISO. I set the aperture at a constant f/8 using the same lens moved around to each of the bodies. Whatever you might mean by “…apertures are not exactly the same for all cameras at a given ISO” completely escapes me. It’s f/8, for all pictures. Please feel free to check the metadata.

          To achieve consistent exposure, the only setting left to change is shutter speed. And I let the internal meters, all set to the same parameters and metering the identical area of the scene, determine that shutter speed. And, frankly, I’d say these photos show an extremely consistent exposure.

          In any case, you clearly have a different opinion, and I respect that. I am sure we will all look forward to your series of tests done correctly. When might we expect those?

          • Chary

            When I check the ISO6400 images, I am quite surprised by for example D3s vs D600. D600’s image seems a bit brighter (the light things like the paper or the pompom starting to burn, the grayish background a bit brighter) but the exposure is 1/3EV darker (1/15 vs 1/13). How’s that? Did the lighting change or is D600’s sensor more sensitive?

            • Sylvain Larive

              I’m with you on this. To me, the D600 clealy looks brighter, the D4 as well, though slightly less. Not brighter by much but enough to notice right away, to me at least. Not saying it’s the tester’s fault, but it is there nonetheless.

          • Hi Terry,

            I’m very sorry but Lord Eels is right and as an author of almost 30 books on Nikon cameras and a trained photographer I do have some knowledge of testing. This is not a
            matter of opinion, but a matter of technical knowledge. Of course the only way
            to test noise between cameras is to take care that the amount of light hitting
            the sensors is exactly the same at the same ISO value, in other words: use the same

            That means:
            within the same ISO setting, use the same constant light source and check it using a light meter (I hope you did), use the same lens (you did that), using the same aperture
            (you did that) AND shutter speed (you didn’t) and exactly the same processing (did
            you???). Of course this can only be achieved using M (manual), not A (so only
            in manual, not in aperture-priority AE). Why? Because at A the shutter speeds
            are determined by the light meter and no two light meter systems are equal, not
            even between Nikon cameras, especially when using matrix metering (which you

            In A not even the shutter speeds displayed as equal in EXIF data are equal, they are rounded.
            In some pictures in your test the EXIF data even clearly indicate that exposures differ approx. 1/3 EV, e.g. at ISO 12800 where the D800 uses 1/25s and the D3s and D4s 1/30s. That would mean that the D800 effectively is at a lower ISO value – or the other cameras at a higher ISO value. It could also mean the light has changed while changing cameras but then your test is flawed as well. In other words: you failed to acquire
            ceteris paribus, your test has some value, but no conclusive value.

            Apart from
            that, within the limitations of your test, I get the impression the D4s does
            better than any other camera in this test. But yes, to be sure, you would have
            to do the test again. I’ll do it in the next two months, but I don’t know when

            • syd

              I would have thought that with 30 books you’d know now that setting everything constant, inc. light, but changing the iso would give you highly variable exposures I.e. bright, brighter….white….


            • dumbfounded

              It’s really shocking how many people are not getting this. Nobody said to NEVER change the shutter speed. They are saying that at a given ISO, the shutter speed should be the same for all cameras. For example, (and I’m just using easy numbers) they should all be set MANUALLY to 1/100 at ISO 3200 and then they should all be set to 1/200 at ISO 6400. This really is not a difficult concept.

          • lord eels

            cameras have meters that function within a 1/3 of a stop. clearly your test was on a borderline of consistent exposure because different shutters were selected by the different cameras at a given ISO and fstop. these different shutters lead to different noise signatures and one image has more light. it’s rather simple isn’t it. this test is WORTHLESS.

            • “WORTHLESS” is out of line. This test may not employ the ultimate in rigorous controls, but a reasonable degree of care was taken, and IMO the results are interesting and largely meaningful.

            • lord eels

              if you don’t understand what I mean at this point, just be quiet.

            • I understand exactly what you mean.

            • Jorge

              Don’t encourage the moron.

            • lera

              lord eels lord of my eel

            • Frank

              if you think it is WORTHLESS why bother joining the discussion? I guess to show that you have all the wisdom that is needed for proper testing….

            • Endre Tollár

              That is the point. Perfect comparison is not existing. Close to perfect or reasonable are available. For me a >200k ISO picture is valuable, because I have only a 25k ISO machine. If you do not accept this test, neglect. Or give supporting idea to improve.
              The easiest thing on the WORLD is criticize the others action. Do it better !!!

            • Jerry Friedman

              Although the guy did seem to be a dick, to reduce what he said to “The easiest thing on the WORLD is criticize the others action” is simple bullshit.
              The poster simply suggested a different, or in his mind, ‘more correct’ methodology. Yeah, he was critical (abusive?) of what he felt was inaccurate information. Suggesting ‘the truth of the matter’ or correcting some others misstatement has value and there is no need for him to “do it better” or do anything at all. It is information, facts and operational details – They are either correct or incorrect and there is no need to demonstrate them – They won’t be any ‘righter’ or ‘wronger’ for the experience, or opinion.

            • Jorge

              Man why are you so rude and arrogant. They guy did a test. I appreciate the information. If you can do better then by all means do so. or STFU

          • Endre Tollár

            I would rather thank to Terry to spend time for this test. Probably it is not perfect, but I can get information from the result which wouldn’t be possible without having all these camera.

            For a repeatable test probably the 11 step grayscale test chart would be more convenient. ( I think, even a single full 18% grayscale sheet would be enough.) But don’t misunderstand Terry, don’t do it for me. Any real life picture allow to judge the result by human eye only, which is less reliable in any case than measured units.

            Refer to the Exposure vs. Aperture change complain: the sensor generated noise have to be compared to the quantity of the incident photons. I do not see any difference of the Exposure change vs. Aperture change result if the incident photons became more or less. If someone doesn’t agree, please demonstrate the difference on a picture content. I’m really curious.

            • lord eels

              there is not set of 5 different camera meters in the WORLD that will auto meter a complex scene like this exactly the same everyone. wow. you are quite naive.

          • syd

            lord eels is a wanker but to be fair terry, shutter speed isn’t the only variable. one could set the shutter and fstop constant and adjust the light, either through flash or better yet, nd filters. truely this would be the best analog to real world situations. it’s great you tested high iso, but some people demand high iso low light comparisons, because they don’t have faith in high shutter speeds simulating low light.

            those of us that understand that high shutter speed simulates low light accept your test.

          • Thom Hogan

            If you want to lock down all variables, using f/8 on a modern Nikkor can be problematic. The aperture activation arm on even pro Nikon cameras can be inconsistent, and it is inconsistent across cameras. I don’t see much variability in your exposures, but I do see some very minor ones. Not likely to impact the results here, but still…

            A better choice is to use and older lens with aperture rings, where you can absolutely guarantee no aperture variation between shots.

            • Terry Hansen

              “Aperture rings”…what are these odd things of which you speak?

              I have not had an older lens with manual aperture ring for some time now. All I have at present are them thar newfangled modern Nikkors.

              But your point (here and other points elsewhere in these comments) is well taken, a physical aperture ring would have avoided the possibly minor inconsistencies. And your point is NICELY presented, thank you for that as well.

          • Jorge

            WELL DONE AND SAID TERRY. That LordEels is an arrogant and rude F&ck.

          • Maji

            Excellent reply to a rude poster. Let him post his findings.

            • Jerry Friedman

              LOL – He did – He found this methodology lacking.
              Which part did you miss?

        • Alwin Papegaaij

          Clearly keeping the aperture and shutterspeed the same while changing the ISO will not keep the exposure the same. The way Terry did the test is the only correct way to do it.

          • lord eels

            no, you numbskull. that not what I am suggesting. you would have to manually change the shutter to compensate for each bump in ISO. you must have the same shutter (at each given ISO and f/stop combo) as even slight over/under exposure at high ISO can produce radically different noise signatures.

            I am shocked you would even post such drivel. clearly you are uneducated and lacking critical thinking.

            • syd

              on drugs much or you checking if we are?

            • Joe

              I would have taken your posts more serious if you didn’t start responding with a disrespectful tone. You can make your case without calling people “numbskulls” or telling people to be quiet. You can easily make your case and rebuttals without that extra nonsense.

              That being said, I would like to see you post the next NR exclusive that performs a re-test of all these cameras. I am genuinely curious how drastic your suggestion would change the results.

            • Jorge

              This is what this guy does. insulting, rude, arrogant; and admin let’s him hang around for?

            • umeshrw

              The sad thing is that when we wish to upvote your post for good technicality; we still itch to downvote it for your rudeness and insulting. Is it really so difficult to speak/write civilly while disagreeing to other person’s point (however stupid or ignorant it may be) ?

            • Jorge

              Holy crap you are an a$$. I’m sitting here laughing out loud at your pathetic attempts to gain attention you f*cking rude ba$tard.

        • jaredpollo

          lord eels really? ahah lool camera noob go buy a toy-fuji instead

      • Eric Calabros

        fixed shutter speed and fixed aperture may not be realistic but as all of these bodies are Nikon I guess it would result constant exposure

    • guest

      I also agree that it would be easier to compare photos if they’re consistent. It seems as though even the WB is auto. When viewing two photos side by side, their exposure and even color temperature aren’t even the same so it’s hard to evaluate them.

      If it’s a general test for the cameras’ metering auto WB capability, then I would agree to leave them auto, but for low light high ISO test, the only variable left should be the sensor output.

  • Stan

    Why does the Raw S ( small ) look better? do you have more dynamic range with the 16.2 (large) file? and if so ,can you adjust the large file to look better then the small?

    • I assume Nikon is effectively making each pixel 4 times larger by combining 4 into one. This could be done in software, with some sort of 4 pixel average, or with hardware, by having a means to electrically connect 4 pixels together to make one (with four times the light gathering ability) when in RAW S mode. I suspect the former, but that is just a guess, I have no idea what Nikon is actually doing.

      • Thom Hogan

        They are doing white balance application, color noise reduction, and converting to YCC color according to a friend who should know. They do not just bin the data.

        What you get is a YCC, with each pixel having a Y value of less than 11 bits depth (no value above 2151). Adjacent pixels share a Cb and Cr value.

        In short, we’re a long way from “raw.”

        • Reality Check

          Agreed. Even if they were pixel binning simply to produce a smaller “RAW” file, it still wouldn’t yield the same low noise performance as a sensor with a larger pixel pitch that is native to the intended resolution. Binning may average the noise generated by amplifying small pixel wells to the sensitivity required but it can never eliminate it. Wouldn’t it be great to be able to simply swap a lower resolution sensor into the camera body whenever it’s needed for low light work?

          • Thom Hogan

            Maybe, maybe not. I think the problem here is that you really need to understand what the customer is trying to do and what they want. This is not as simple as it seems.

            For example, if I shoot 36mp raw and my intent is 9mp final output, downsizing tends to add acuity into the results. On the other hand if I’m shooting with no light at absurd ISO values, my basic interest is just in combating noise, including quantum shot noise. Then I’d rather have intelligent noise reduction being used. Then again if my main interest is simply to keep file size manageable, then, yes, I want a lower count sensor.

            This is one of the reasons why I argued that the pro body should be modular at the sensor, and why I argued that the camera itself should be programmable. Those two hardware changes allow virtually all needed user scenarios. As long as we get hard coded, proprietary answers from the camera companies, we get their answer to what they perceive to be the problem/need.

        • Thank you.

    • Maji

      Iliah Borg, who is the brains behind the libraw RAW conversion software has some interesting observations that he posted on DPreview. http://www.dpreview.com/forums/thread/3637319

      I hope this is okay with Admin because it points to another site, but I think Iliah’s observations as he writes RAW conversion routines are very interesting.

      • Of course it’s ok (unless you are spamming).

        • Maji

          I am not Iliah or have any financial interest in his company. So, it was just an FYI and not spam 🙂

          • No problem,I just say it because sometimes I get spam links which I delete right away. As long as the link is related to the current discussion, it’s ok.

  • lord eels

    I’ll take dxo tests over this guy. he buys S model pro bodies and skips a generation. that is a bad idea. buy the tick not the tock

    • mikeswitz

      You are not a photographer, so why would anyone care what you would take. I can assure you not even your mother would care.

  • bgbs

    D4s clearly exhibits less noise than any other gang member, although by not much. Whats even more important is that D4s controls color far better at highest ISO levels which gives it a further ISO leverage.

    • nick

      Saying ” D4s controls color better from ISO 6400 and beyond” means what I also can see ? A clearly more neutral WB?

      • bgbs

        No, the image still holds its color, WB or not, while ISO increases, other cameras start to fall apart in color as the ISO increases above 51K

  • hooman

    to me it is obvious Nikon D600 with that price has got the best performance out there. at ISO 25600 the image quality of all of them are pretty much the same with some negligible differences.

    • wonderingpi

      Its great but the gimped Af sucks

  • George Kalogeris

    anybody could enlighten us how RAW_s works?
    Do neighboring pixels team up to make an ultrapixel ?
    Or is it just a good resizing algorithm that we can simulate with Photoshop

    • Terry Hansen

      My guess—and it’s just a guess, no insider or technical information here—it’s as you first said, “neighboring pixels team up to make an ultra pixel”, a process also known as pixel binning. Four photosites are treated as if they were a single, larger photosite. More surface area per virtual pixel means better sensitivity. And, it was suggested to me, this yields a different and better result than merely downsampling a file after capture.

      Perhaps not coincidentally, pixel binning would also explain why the RAW S file is exactly one fourth the size of the normal file. This could even be what Nikon means by “a new sensor”, possibly the same/similar sensor as the D4 but modified to allow for the pixel binning.

      Finally, if it is pixel binning with its inherent 1:4 ratio of pixels, I don’t think we can look forward to any other intermediate sizes here or in any other future cameras. Pretty much “one fourth” will be the rule, meaning, for example an updated D800E might offer 36MP normal files but only 9MP small files.

      But that is all speculation in the absence of Nikon’s technical disclosures.

      • Thom Hogan

        It does not appear that Nikon uses pure pixel binning to create the sRaw. The values saved are not RGB, they’re luminance and color, and there’s luminance for every pixel.

  • Here’s the thing: this test is a high iso test without the low light. Why don’t you just shoot outside in the middle of the day with ND filters stacked on?People that are interested in a low light capabilities of a camera never shoot at F8 at 100k iso unless, well, they have no idea what they are doing. If you shoot night landscapes you can set the camera on a tripod, use 100 iso and shoot at F16 as it doesn’t matter if you shoot at 30sec or 15min. This camera (D4s) was designed with a fast response in low light situations, when one has to use a higher iso AND fast shutter speed to freeze action in very dim, close to darkness surroundings. Therefore, other than pure -oh, well they did another test that proves all nikons are good in low light- this is, for me, a little silly.

    • Jeff Hunter

      Not true. I live in Louisiana and shoot Mardi Gras parades at night, hand-held, as well as nighttime cityscape aerials. And depth-of-field and at least 1/250th is very important. Expand your photographic experiences!

      • Jeff, simply pointing out that in my opinion, an exposure of 200 @f8 shot at 102.400 iso is silly. This test would have been more accurate if shot in a real LOW LIGHT situation not the equivalent of an 125 F1.4 @1,600 iso which is plenty of light for most modern cameras to perform well. My iPhone shoots well in that kind of light. I am happy you find this test useful for your needs, however it is absolutely useless for mine, as I don’t have the luxury to stop down my lens at f8 when I shoot in a pitch black venue. Has nothing to do with my photographic experience, Thank You for trying to expand my horizons.

  • Stefan

    I wonder if Active D-Lighting was deactivated on all cameras? Looking at the pictures in low ISO it looks like that this test was not really carried out thouroughly (identical settings)…

    • desmo

      active delighting is part of in camera jpeg conversion…
      note, he said he output the files as raw ,
      then converted to jpeg

      • Chris Gullett

        Active D-Lighting makes adjustments during exposure. D-Lighting is used for In camera RAW Development. At least according to the D4 manual.

      • umeshrw

        If D lighting is activated in camera for raw files it is applied automatically to raw files in capture nx. So it should be the same for nx d too.

    • Terry Hansen

      D-Lighting WAS deactivated on all cameras.

  • george

    Nikon d4 is a winner

    • Pablo Ricasso

      Is it, darling?

  • dan

    D3S is the KING!!!

    • wonderingpi

      I agree.!

  • Morris

    nikon j1 is the queen !!!

  • Is there any side by side comparsion with this shots?

    • Spy Black

      Load them up into Photoshop.

  • Endre Tollár

    I downloaded the 200k ISO picture and no indication in the EXIF data about the applied ISO. Is it intentionally removed ?

    • Jan

      Probably because Nikon’s body shows Hi-something at higher than “recommended” ISO range.

      • Endre Tollár

        Unfortunately, the the ISO Speed Ratings field ( Exif ID 8827 ) is missing. I downloaded another D4s picture from Flickr with 25600 ISO value.

    • Terry Hansen

      Nothing removed intentionally.. Shot RAW. Opened in Capture NX-D Beta. Output as JPEG. Uploaded to DropBox. From there, NR grabbed the shots and posted them to Flickr. If metadata was lost in that chain, I don’t know where it might have happened.

      • IndyReader

        I tend to think it would be a bug with the NX-D software.


    All well and good to show the results based upon RAW BUT we should keep in mind that this camera was likely designed for pros who, due to their organizations not having the time to process RAW files, are going to be shooting in JPEG. (I’m referring to sports Journalism photographers) It is great to have these results, though, for others who will be shooting in RAW. Just thought it should be mentioned!

  • RichardAB

    I’m amazed, having previously had no idea of the quality achievable at such enormously high ISO numbers.
    I’m a humble enthusiast who uses a Nikon P7100 for all of my photography, shooting at ISO 100 unless there is no option but to choose a higher ISO. Not that I feel too restricted, I don’t.

    • Spy Black

      Although you’re not going to get this kind of performance out of the D7100, you can still get good high ISO from your camera at least up to 6400, which is a functional ISO in many low-light situations.

      • Alert

        P not D, wake-up.

        • Spy Black

          Oops, need coffee…

      • umeshrw

        I compared results from 7100 to my 800E. D800E- 3200 iso is equivalent to D7100 -1000 iso. But sharpness for 7100 is excellent. At par with that of D800e( Ofcourse MPs notwithstanding)

  • Julian

    would be really nice to get a RAW S option added to the D800 firmware…

    • Terry Hansen

      It might not be a simple firmware option, we don’t know, and Nikon isn’t saying (yet). But, even if it were possible, or a new “D800s” were released, my strong educated guess is that the MP count would be exactly one fourth of the native MP. Would you be happy with a 9MP RAW S file on a theoretical D800s?

      Frankly, given the full stop improvement in image quality, I would…but then again, for a 9MP improved ISO image, I’d just use pretty much any of the 12MP or 16MP cameras used in this test, get even better high ISO imagery, and with much higher native resolution.

  • Ineluki

    Hi Terry,
    thank you very much for your outstanding tests. The DF has been already mentioned but as a D4 owner I have to say there is no need to update for a D4s.

  • M

    D600, hands down…

    • Morris

      Oil up

  • Spy Black

    The best thing to get out of this comparison is that you’re doing pretty fine with any of these bodies when it comes to high ISO. However the camera you should really be comparing the D4s to is it’s Canon “equivalent”, because that’s it’s real competition.

  • Kynikos

    Excellent test.
    That junk that showed up a few days ago needs to be lost down the Memory Hole.

    • rt-photography

      the other vase test of the D4 vs D4s post?

  • Julio Tamara

    The D3S is better!

  • wonderingpi

    Jeez they all look the same… happy I didn’t buy a d4s jeez

  • Simples

    At this sensitivity the best way to conduct the test is count the number of photons entering each camera and make sure it is the same for each iso set. Alternatively just make a short exposure 1/1000 say, in the dark with the body caps on and compare the noise generated by the sensors.

  • jamez

    Thanks for an excellent test! I downsampled the d600 and d800e to d4 resolution and did a comparison. All cameras perform much the same at 3200-12800 I think, although there are some subtle differences in noise and dynamic range. Also interessting comparing them at iso100. The difference between the D800 vs the others at iso100 is not subtle.

  • UA

    It’s just amazing that 102k ISO is still well usable in web. Although it’s usable from D3s as well, if you tolerate some mild visible color noise. That’s were D4 and D4s are better. And by web use I mean resolutions below typical modern screen sizes, 800×600 for example.

    But still, the D800E is just amazing with all its pixels, the DR and the details. It’s just blows everything else away up to the 12k ISO. At 25.6k D800E seems to struggle a bit against others in some OOF details and noises.

    But still I am happy using my D700. I think the D800s might be a good upgrade if it comes some day (hopefully a bit more fps and that RAW S there).

  • Tom

    Great comparison. I would also ask one more thing – throw an image in there of a top APS-C camera like the D7100 or even D7000 for the sole reason to see the difference in ISO performace and get an idea of the “true” improvement from APSC to FF.

    • Terry Hansen

      I would be happy to do that! I would just need to borrow such a body in San Francisco for a couple of hours, if anyone wants to offer one. I still have the still life and lighting set up, and will maintain it for a few more days.

  • Maji

    Great work. Thank you so much for doing the tests.

  • Sandy

    it seems several of the images labed D4s (L) have a post-it note that reads D800e? Which is correct?

    • Terry Hansen

      I just went through every photo over posted on the linked Flickr site, and, to the best of my ability to tell, all photos were correctly identified, I am not sure where you saw an anomaly?

      If you get a chance, try linking over to the Flickr site and see if the problem you saw persists.

  • Manvin

    all looks good to me, I’m still happy with D4 apart from video lol, i never use high ISO just low as possible – light is everywhere 🙂

  • Jason

    Terry, you sure did put alot of effort into this, Thank you!!

  • Ray Justice

    My thoughts, from a peon low on the food chain. Corey Rich stated on his site “Don’t let the name of D4s confuse you. This isn’t just some small upgrade from the D4. The D4s is an entirely new camera that is so badass it blew me away, just like it blew away each of the three photographers featured in DEDICATED. We were all floored by what this camera was able to achieve.” And Dave Black stated on his site “The NEW Nikon D4s camera is fantastic! It is truly a significant upgrade well above its predecessor the D4. The D4s upgrade from the D4 far and away exceeds what we all thought was possible. From the very first picture I made with it I knew this camera was something special.” These guys have put the D4s through several weeks of intense hands on testing and make the statements seen on their website. So this is how I made the decision to order the D4s (received yesterday) and the same basis I used when upgrading from the D3 and then the D3s. The cost of my new D4s was $6496.95. I will now sell my D4, and research online shows most D4 bodies in fantastic, like-new condition with 37568 actuations are selling for approximately $4400.00. So ask yourself, is it worth upgrading from the D4 to the new D4s for a cost of about $2100.00. Not to mention in a couple of years when the D5 is announced, the resale value of the D4 will probably drop in half of what it is now. Of course just my thoughts of I made my decision…

    • wonderingpi

      If you honestly think its a big upgrade then your falling for good old maretking this is not the gains of d2 to d3 or d3 to d3s.

      Buy some glass instead.

    • lord eels

      why are you quoting nikon ambassadors about a new nikon?

    • John_Skinner

      This is the voice of a person ‘invested’ and justifying the purchase. Although I’ve NOT bought into this body, I have used it (IMO)extensively over a 3 day period.. In images shot between the D3s, D4 and this…. There isn’t 3K of value here by a long shot.

      You can work the numbers all you want to, it just doesn’t make business sense. If your a guy with ‘G A S’ then yes.. But a working guy with costs vs. value vs. return… No

  • Horshack

    Terry, thanks for the test. One important point – when performing these comparisons you should use a lens with a manual aperture ring rather than relying on the aperture control in the body. This is because the tolerance of the aperture levers vary from body to body (even of the same model). It’s not uncommon for it to differ by up to 1/3 stop or more. Since the High ISO performance of current Nikon FF bodies is so similar this means the aperture variability alone can easily sway the test in favor of one body over another due to a higher/lower absolute exposure than what is indicated by the nominal aperture dialed into the camera.

    • wonderingPI

      Its close enough to show a layman that the differences are so minimal.. marketing marketing… The test is ideal enough to show that you should spend your extra CASH on lenses then upgrading a d3s or d4 to d4s.

  • broxibear

    Can you also take photographs with the D4s or is it just a device for testing high iso noise ?

    • Terry Hansen

      I was told by the B&H sales agent that, if I wanted to use the D4s to take actual photographs, I would have to buy the optional 3-year SquareTrade Protection Plan. At an additional $699.99, I decided to bypass the plan and just my D4s for testing high iso noise. (That’s OK, because my friend Fred says he knows a hack that will unlock photo taking abilities for only $20! And for just 10 bucks more, he’ll make it pick up cable TV! Sweet!!!)

      • John_Skinner

        Are you taking the piss?

        HAVE TO BUY? HACK? Fred ? Holy cow !

  • chlamchowder

    I was just looking over the ISO 12800 images from the D600, D800, D3s, and D4s. Does anyone else feel like there’s some kind of NR going on with the D600/800 images at that sensitivity? In some parts of the image, the D600/800 resolve less detail than the D3s, which goes against my experience (I own both the D3s and D600. At high ISOs, I feel like the D600 still out resolves the D3s, but the D3s doesn’t lose color integrity as quickly).

  • Back to top