PhotonsToPhotos published new Nikon D850 tests results for different RAW options

PhotonsToPhotos published new Nikon D850 test graphs based on data collected from the different RAW files options (RawM, RawS, Lossy in 12 and 14 bit):

Dpreview also completed their D850 review:

Check D850 availability: B&H | Adorama | Amazon | BuyDig | Cameta | Focus Camera | eBay | WEX | Jessops

Follow D850 on Facebook: Nikon D850 Page | Nikon D850 Group

This entry was posted in Nikon D850 and tagged . Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • Steven

    Now I’m starting to feel some gas. Thanks. At least I have a collection of Nikon glass already.

  • Luca Motz

    Someone explain to me how the D850 is supposed to be between 1 and 2 stops better when shooting in some of the smaller RAW formats. I thought the measurements are supposed to be normalised for resolution. And on top of that It’s physically impossible for the camera to suddenly have more DR when I select something else in software. The data will be there in the full format files as well right? Seems to me like the dude measuring this doesn’t quite know how to bring that data out then?
    I’m genuinely confused

    • MB

      There is definitely some error there … there is no way 12bit jpeg like encoding (12 M and S) can have better DR than 14 bit RAW …

      • BVS

        D850 M/S RAW is not like S RAW in D810. It’s actually a true RAW file, albeit only 12 bit.

        • MB

          So it seems that they are just in camera re sampled images after all … but anyways they are 12-bit so how can we explain this results?

          • TurtleCat

            The question is how often do the 14 bit raw files have 14 bits of significant data? And at what point of gain will it never be possible to get 14 bits of data. I’ve read that the crossover is somewhere around ISO 400.

            • Thom Hogan

              Still trying to answer that empirically, but yes, that’s about right. At the low gain setting you would tend to get additional useful data with the extra bits. In the high gain values (>400) you certainly don’t. But exactly where that crossover is still needs some testing on my part to verify.

            • Allan

              Why doesn’t Nikon write software (in the camera) that automatically decides on 12-bit or 14 bit based upon the ISO used for that particular capture?

            • Thom Hogan

              I think you mean “why doesn’t Nikon allow us to program our cameras to do what we want them to?” ;~)

              I put it that way because there are plenty of things I’d like to group together as “a Thom-automated setting.” Not just one menu choice like that.

            • The answer is straightforward and as expected.
              You can see it by comparing 12-bit and 14-bit Read Noise in DNs.
              12-bits is not sufficient where the low values start to curve rather than being straight (2 below the 14-bit).
              This happens for values as they drop below 0.
              Remember, the y-axis is logarithmic, so 0 is 1 DN.

          • dave

            D850 M RAW stored 36mp, and asks RAW processing softwares do pixel binning to 24mp. That’s how it get less noise in lowlight. And 14 bit is better at base ISO, not much at high ISO though.

          • akkual

            The difference is that those 46mpix RAW measurements do not have any processing applied, where as the camera does seem to apply processing when scaling it down to RAW S or RAW M. You can achieve the same result by applying processing to a 46mpix RAW before scaling it down. So there is not mistake in those measurements.

            This is also what I presented here earlierly why I found D850 superior at high ISOs. Take 46mpix -> NR it -> scale down to 24mpix will give better end result than taking 24mpix and NR it from my D750. So those RAW M and RAW S do some processing before scaling down automatically for you. Apparently, the processing is double resampling.

            12bits is enough for 12 EV. The graph hits ~11.9 EVs. But of course you will loose color and maybe some overhead in your exposure.

        • Thom Hogan

          Well, it’s “raw” in the sense that it has Bayer type data in it. But because there are two transforms done to the data, it’s not the original data any more. Nikon seems to have picked a process that suppresses noise, but it also results in significant destruction of acuity.

          • I’m not sure we can take Nikon to task on this since any resizing to what is effectively a larger pixel will sacrifice acuity.

          • “Significant” in this case would be subjective?

            • Thom Hogan

              Would be “clearly visible.”

            • I would steer clear from it then. Thanks for the explanation.

      • ninpou_kobanashi
    • not too liberal

      This is really amazing if true (I certainly wish it is true)!

      • peter w

        It is data, so it is ‘true’. However, your interpretation / my interpretation of the data may be false. Or we may not see all relevant data.
        (I think it has been amazing from the start of it. Technologically speaking.)

    • At the risk of repeat myself. The downsizing to RawM and RawS also has a noise reduction effect. So even normalized the net effect is higher dynamic range (at the expense of the usual noise reduction smoothing).

      • Luca Motz

        Thanks for your clarification Bill but isnt the DR the same to start with? So I can go ahead and use the not downsized file and apply noise reduction after the fact?
        If so it seems the resolution normalised logic is kinda flawed. Don’t get me wrong I appreciate your work I just really can’t make sense of it this time. Basically what I’m asking is are you suggesting Im losing something by not shooting downsized

        • peter w

          Stacking is applied to increase dynamic range, reduce noise. Downsizing is a kind of stacking within availlable data, thus implies other kind of loss, which is obviously resolution.
          The basic dynamic range of the cells is enhanced by stacking data of several cells.

          • Luca Motz

            Okay but that means I don’t lose anything when shooting full size correct? Because I can still do the downsizing after the fact. If so the normalization for resolution seems incorrect in his data. Don’t get me wrong I’m definitely not totally stupid and I totally get that I get less noise when averaging over several pixels but to me that should be accounted for when normalizing for resolution

            • peter w

              Ah, see what you mean, clearly I am not a bit smarter than you ;). It should be tested. There is a possibility that double reading (and subsequent stacking) gives more noise than reading an analogue stacked signal. It could as well be the other way round. No idea how the read-out is working.

            • Luca Motz

              Exactly my thoughts. Probably easy to find out (and I will when I get home tonight) but something to worry about. I assumed there would be no real difference but I had the feeling the charts were suggesting a 1-2 stop difference which was now clarified by Bill. I stand by my statement then that the measurements are not actually normalized for resolution then.

            • Maybe the resampled images in M/S raw are losing image detail as well as size hence it is showing better NR characteristics? But even then what makes the DR better?

          • My comment doesn’t detract from your statement but strictly speaking it’s more like pixel binning or re-sampling rather than stacking.

        • At the pixel level the DR doesn’t change. But it’s as if you’re comparing an image with noise reduction (RawM or RawS) against one that has none (normal Raw). The noise reduction will boost the apparent PDR. So the question would become whether that “built in” noise reduction is objectionable to you or not. That’s a question that PDR can’t answer.

          • Luca Motz

            Got it thanks! So I’m not losing anything when capturing all the data and then doing the noise reduction after the fact right? Always assuming my software is as good as the one in camera.

            • My personal preference is to do all (any) noise reduction myself. But given the file size difference I can see someone using RawM or RawS if they don’t find the “built-in” “noise reduction” objectionable. It’s classic YMMV.

            • Luca Motz

              Same for me. Thanks a lot for your clarification. If you find the time would you mind publishing an article to clarify this for everyone?
              Anyways keep up the great work and it’s time for me to work for my money 😛

    • akkual

      What I observed from RAW M and S, they have internal summation applied during deduction of the pixels – basically a noise reduction. The thing with those photonstophoto measuerements is that they are made with unprocesssed RAW images, which are scaled down. But the true strength of oversampling allows you to do much more better processing at full resolution and then scale it down.

      That is, take a 46mpix shot -> process it -> scaled down to 8mpix will result to better end result than take a 46mpix shot -> scale it down -> process it.

      And this is the reason why more megapixels do matter.

      So there is no error, people just don’t realize this factor of postprocessing before scaling down making things much better.

      • Piooof

        Note that it’s the same thing in music. You don’t need 24bit/192kHz audio as a listener, but you do need them for post-processing if you’re a sound engineer.

        • peter w

          You don not neccessarily need 24bits in the recording, i mean if it is not there you don’t need to record it. But you do need oversampling in the processing.

          However, gathering better information on noise can be extremely usefull to better get rid of it. Like for lenscorrections.
          (Sony or Philips (?) made some terrible mistakes in the ’90s reprocessing recordings from like 1920-2930. )

        • Rick Jansen

          Very much true :)! However, 192kHz is a little overkill, I’m at 32bit float 88.2 kHz for my recordings, which is high enough. Oh well, that’s another discussion 😉

        • ZoetMB

          As an ex-recording engineer, I might challenge that assumption. 192kHz gives you a frequency response to 96kHz, but humans can only hear to 22kHz tops. That extra frequency response, if actually recorded, can actually cause harm to the audio quality, because when playing it back, one is using amp power and heating up the speaker coil for sounds that no one can hear anyway, if those frequencies even get past the microphone or direct box or mixing board.

  • Nikkor300f4VR

    GAS intensifies..

    • cst labs

      release some

  • JJ168

    Wow, smaler Raw now provide better low light performance. How time fly, i remember people was wishing for smaller Raw file in 2012. Nikon gave it to us via d810. Then people wished it would improve low light. Here we are. Well done nikon.
    What is next? 🙂

    • Thom Hogan

      Read the entire discussion. The higher DR for the smaller raw file sizes does not come without a penalty. You can’t noise reduce without impacting acuity.

      The question is this: can you do a better job than Nikon? That answer varies from person to person, but for me the answer is yes.

      However, the camera can do it as you shoot (with some buffer penalty), while I can only do it with post processing, which takes my time.

  • Charles

    If these numbers are correct… I may have to sell a few cameras so I can get the D850.

  • PhilK

    This “Best DSLR on the market today” conclusive statement at the end of the D850 review is extremely bold. Not sure I’ve ever seen DPreview make such an unequivocable assessment of a camera like that before.

    No wonder the interest/popularity stats for the D850 on DPreview are going through the roof.

    • John Mackay

      Well there are only 3 others in the running, and 2 of them cost over 6 grand and are much more niche. The other is the 5d4 which isnt that great.

      • PhilK

        Yeah, it did occur to me that most of the other significant players are making mirrorless primarily.

        But Sony’s SLT cameras like the A99 II are still technically DSLRs, too. And one could conceivably still award “best DSLR” status to a non-FF DSLR, too. (Like the D500)

        • John Mackay

          I thought A mount was SLT not SLR? And for lens reasons I would never give it to an apsc camera, but perhaps medium format like phase one could make a case :p

          • Martin Francis

            Single Lens Reflex is a universal term referring to cameras with a mirror in the optical path between the lens and shutter. This includes Sony’s SLTs.
            Single Lens Translucent is Sony marketing language for an SLR with a semi-transparent (not translucent) pellicle mirror that doesn’t pivot with each exposure.

          • PhilK

            As Martin Francis replied, Sony’s cameras with a pellicle mirror are still SLRs.

            Various SLRs with non-moving pellicle mirrors have been produced during the film era, but were never re-classified as non-SLRs. (Canon Pellix, various limited-production high-speed flagship models from Nikon and Canon, etc.)

            • John Mackay

              Yea, I looked it up after he replied. I thought reflex was to do with the mirror moving but apparently its light bouncing. Still, as A mount is essentially a pity camera for a dead system it never had much of a chance vs a 1 year newer model from Nikon.

            • PhilK

              A “pity camera”?? You must be joking.

              Sony just released the A99 II less than a year ago, and I assure you, corporate management of massive multinational corporations doesn’t spend millions of dollars in development and production costs for something like that because they “pity” their customers. They are running a business.

            • John Mackay

              I am not. Management do spend millions of dollars on advertising, PR and brand reputation. Sony cutting off support for all their A mount customers without first developing an equivalent E mount camera would do huge brand damage and affect their e mount reputation.

              A mount has a very tiny market share and they have said publicly that they currently have no plans to bring out more lenses, and will just update camera bodies. The R and D put into the a99ii also does not seem huge, it is a great camera but it is in essence an a7r2 with a turbocharged sensor and a good but not fantastic off sensor phase detect system. I also believe the Video mode also can’t use af at wide apertures. It does not seem like they are expecting A mount to grow bigger and bigger over the next 5 years and make them lots of money when it is one of the least popular cameras in a rapidly declining market segment.

            • PhilK

              Gigantic, publicly-owned multinational corporations do not make products because of “pity”. They make them for business reasons. Sony is not a charity.

              One of those reasons could certainly be “brand reputation”. But that is not “pity”, that is a strategic business decision. I’d rather we didn’t attempt to anthropomorphize faceless business entities.

            • John Mackay

              If the reason is not to make money, or to give a halo effect, or to advertise, but instead to stop A mount users looking sad and then get angry at Sony, it is basically pity. Pity incidentally is a product of evolution, and those that exhibit it to some degree are more successful. That is why it is still around. While Sony may not have a face, at least in the west like Apple do, it is run by people, they are not idiots, and they know the value of showing pity. If you think pity is by definition a detriment to the exhibitor then you do not understand pity or evolution.

            • PhilK

              Dude, repeat after me:


              If a corporation’s management and board decide to produce a product because they think that, for whatever reason*, it will benefit the corporation in the long-run, that is just a cold-hearted, strategic, business financial decision. It is not “pity”.

              People can feel pity. Corporations cannot.

              * (Including catering to the misguided fantasies of customers who are
              emotionally dependent on the fantasy that their product is some kind of
              magical faerie dust, or any other silliness)

            • John Mackay

              Again, you are confused as to what pity is. Pity is just an emotion to prompt a rational, beneficial to the self decision to help another. Pity is a perfect fit for Sonys decision. Pity did not evolve because it helped other people, it evolved to help the person(/genes of the person) with pity. Helping others is simply a side effect.

              Corporations cannot feel pity, neither can they make decisions of any type, in the same way that a house does not make the decision of a house hold. People run the corporation, people make the decision. Sonys decisions are simply the aggregation of a few individuals. Pity as a describer is apt. It is a perfect fit both logically and very likely a genuine emotion felt by the decision makers of Sony.

  • VanHoff

    Someone can explain why “12bit Raw M” seems to be the very best option to get the most of the DR and “Read Noise”?

    • ninpou_kobanashi

      You can read here – it looks like since it’s downsampling the noise would be reduced.

      • Eledeuh

        But… shouldn’t the tests be normalized? Surely downsampled 14bit RAW should behave slightly better than downsampled 12bit.

        • TurtleCat

          If all 14 bits are significant. I’m sure more than most 14 bit images have less
          than 14 bits of significant data.

          • Eledeuh

            That’s a practical consideration (a very valid one), but it has no place in a testing environment 🙂

            • Even at 64 ISO the max DR doesn’t exceed 12 as per the ISO v/s DR graph. But I think there is some catch in it as according to nikon DR is much better.

        • El Aura

          I guess the ‘downsampling’ used for RAW M & S includes a bit of noise reduction as well (which could even be due to a different downsampling method). Or to simplify: Don’t trust any downsampling you haven’t done yourself.

    • akkual

      It is not the very best option. The very best option is to shoot 46mpix RAW and post-process it by yourself and then scale it down. What you see in those graphs is basically post-processing done in camera (actually it’s double resampling for scaling down, but that is post-processing as well).

      • Thom Hogan

        I’d agree with all of that.

  • ninpou_kobanashi

    So… always shoot 64-100 or 400 ISO? (^_^)

    • Rick Jansen

      It seems like that, interesting..

    • Tony Beach

      I typically start using Auto-ISO as soon as I’m in varied lighting and no longer exclusively using base ISO (on my D800, but I would do the same with a D850). Losing a fifth of a stop of DR between ISO 250 and ISO 400 wouldn’t change how I do that, and I would rather lose that small amount of DR at that very narrow ISO range than constrain my exposure to what works down to ISO 400.

  • Brubabs

    How am I to interpret the graphs? Does the S850 actually have less ISO noise at ISO 400 than it does at 200, and about the same dynamic range?

    • Yes, less at ISO 400 than ISO 320; because of dual conversion gain technology that kicks in at ISO 400.

      • peter w

        Quite interesting.
        Do you have a link about this dual conversion gain thing? Would be interesting to know why it is shut off below 400 iso.
        Great work. I love data.

  • Azimuth1

    my personal GAS is going down down down …. and luckily I must say!

  • Spy Black

    “Omission of built-in pop-up flash may turn off some users”

    That’s entertainment!

    • Harold1966

      Everybody and his/her/x cat said the same about the D500, still it sold a few. Now it’s ‘missing’ on a 4000 euro camera, and that market niche will miss it? Hmmss.

    • drororomon

      I don’t have a fancy flash setup, so I use the pop up flash as a commander for my speedlight. I don’t get to do that with my D500.

      • Allen_Wentz

        Buy a radio commander like those from Yongnuo or Pocket Wizard. Don’t wish a crap built-in flash on to D850 or D500.

      • Aldo

        I have used it recently just to avoid walking to the car to pick up transcievers and load batteries etc… I must say it works well as long as you are aware of the limitations.

      • Spy Black

        Oh I’m not against pop-ups at all. They’re quite handy for me. Some people however feel that a “real” professional camera should not have one, which I think is comical. I thought it was rather funny that DPR made a point of it. 🙂

        • ZoetMB

          Well, they’re correct insofar as using it as a flash is concerned, but I’m also in the court that found it useful (on my D800 and D200 and D70 before that) as a trigger.

          • Spy Black

            Yes, but you just contradicted yourself. The point is they are useful in any number of ways including, believe it or not, as a flash.

    • Nikkor300f4VR

      LoL, these dpr folks! For sure D850 will be purchased by cat owners..
      “Whaaat? No flash? I’m out!”

  • Jay

    Does this essentially mean I should be using RAWM whenever I am shooting higher ISO, lower light, sports type events? I want to say those figures make it look awfully comparable to the D5 at higher ISOs.

    • Harold1966

      Turn the dial to the running man.

      • Jay

        LOL…. You win!

    • not too liberal

      This is really interesting. Seems too good to be true …

    • The trade-off that isn’t conveyed by the PDR measurement is that a side-effect of the downsizing is a form of noise reduction that might destroy detail you care about. It’s a trade-off that you would want to investigate before committing to RawM.

    • akkual

      No. Results only show that the camera applies some processing for you automatically when shooting RAWM or RAWS. You can achieve similar, or even better result by shooting 46mpix full RAW, process it (like apply NR and sharpening) and then scale it down to 25mpix or 12mpix. Though, if your event is not that meaningful that you need to produce top quality prints, you may well shoot RAWM to safe some space.

    • Nope. Unless you are ok with limited buffer in your sports type events. That conversion requires heavy processing so buffer is affected.

  • sickheadache

    DP..They are just as useful as that guy at 1600…No clue on what they are doing..And just like film flopwell..Can yap from both sides of their mouths at the same time and still make no sense.

    • Ric of The LBC

      Making photography great again

  • I encourage people to place with the interactive charts (links above) since with all series shown the screenshots are hard to interpret.

    • mohammad mehrzad

      Bill, Thanks for the really hard work, much much appreciated. as another poster asked, it would be nice to include a few other comparison point cameras in this chart. D5, D750 and a couple DX contenders, like D7200 and D500, to compare DX results.

    • MB

      They basically converted 14-bit readouts into 12-bit output expanding DR and effectively reducing noise … you can do a similar thing in PS, you can even go from 14 to 16 bit, but that will inevitably lead to some clipping and loss of data … it is OK if you dont mind it but the problem as I see it this isnt a RAW image anymore … they could have done the whole shebang in camera and made it a compressed tiff, that would produce smaller and more usable files IMHO …

  • Cynog

    So, the tracking autofocus is not quite as good as the best in the market, and that’s a con…? Coming from a D810, it’s a definite pro for me.

    • I think it’s a con according to them because they were expecting D5 level performance, because of that AF module.

      • Thom Hogan

        This, of course, is a marketing problem. Nikon set that expectation with their constant “D5 autofocus system” blathering. Had they instead used words like “uses the D5 autofocus parts and algorithms to provide near D5 performance, and far better than D810 performance,” we wouldn’t be having this discussion.

        • TurtleCat

          They could even simplify that by saying, “approaches D5 autofocus performance”. It’s vague enough to be true yet specific enough for marketing benefit.

      • Cynog

        Probably, but that’s not a good basis for an objective test.

    • David Gottlieb

      It’s almost as good as the D5 – but not quite…. I wish Nikon had not been so misleading. I wonder if it got lost in translation or they thought that the marketing would help sell cameras. Why not just say it like it is?We are all going to find out eventually.

  • peter w

    Sorry to ask, Bill. How does your dynamic range 12 – 2 relate to the DR given by Nikon and DXO 14,5 – 6 ? a link is enough.

  • scott800

    Is there a way to see D850 mRAW vs D750 on photons to photos?

    • silmasan
      • silmasan
        • KnightPhoto

          Aha! So the D850 12RAWm shoots like a D5 – very interesting result.

          And as far as “loss” of resolution, we just need to compare the mRaw to the D5 native output.

          • Allen_Wentz

            Not really; read what akkual said above.

            Comparing mRAW against D5 NEFs only applies if no substantive PP will be applied. And if no PP is applied one might as well just compare JPEGs.

            Sorry but I generally see no value to adding by-definition-compromised smaller RAW files to the workflow. We buy a high MP camera to have lots of image data, and arbitrarily having the body immediately and _permanently_ toss data for us does not make sense to me, because when the rare hero image presents it sucks not to have max image data to work with.

            The exception would be certain automated wireless-transfer sports/PJ scenarios where seconds matter. But other than that I say photogs should just upgrade to Thunderbolt 3 and SSDs, and learn to deal with larger file sizes. Heck, I did that with my Mac laptop in early 2011. Even a few Win boxes have fast i/o available today if one goes looking for it.

            • Exactly the reason why I don’t even crop in camera.

            • Allen_Wentz

              I never thought about the DX crop issue, but now you have me thinking…

          • silmasan

            D5’s better 3D tracking for moving subjects is its key advantage. Other than that, the D850 is just so much more versatile in terms of image quality. And btw I think that chart only illustrates what has been said about downsizing D850’s 45mp files (i.e. no need to set at RAW M up front).

  • kyalami

    Is any of this available in plain English? I have a degree in science and my brain is hurting.

    • peter w

      it should be…

      (it is like running, if it doesn’t hurt, you are not running fast enough)

  • Yawn………..

    I live in the real world not a lab, all this means little when in the field

    • peter w

      In one thing you are right. The data show that mostly it doesn’t mean that much in the field.

      However, you could consider showing respect for the people who show with data that your assumption is right.

      (From these data I take – for now – that I will shoot eigther iso 100 or iso 400. Nothing in between.)

      • I will skip 320. That is the one that takes a dip.

      • Allen_Wentz

        Heck you could do that with my D3 and get good pix. We should trade.

        • peter w

          Don’t get you there. Do you want to trade your D3? I could give you my D300, and you would make good pictures with that. Don’t know though, D300 weigths so much less. 😉

          • Allen_Wentz

            I was referring to “I will shoot either iso 100 or iso 400. Nothing in between” for a D850. Because the D3 even D2x rock at low ISOs.

            No thanks on the D300… :~)

            • peter w

              … for low iso situations I meant ;). It may be higher than 400, if required.

    • Nikkor300f4VR

      No Sir, your “real” world is a biggest lab in the matrix, as your are also part of an intergalactic experiment set up by alien lifeform. Your lifegoal must be buying new cameras every year.. [Beep..beep.. initiate launching GAS sequence]

      • you really need to get a life

        • peter w

          Hey Jeremy, what are you doing here, hanging around in the lab? Get out into the field ;). Make some pics. Show them to the world.

      • Shhhh… Those mice will hear you . And then they will have to terminate the experiment and start it again. There goes the earth for intergalactic highway.

  • No comparison on how much file size you actually save by going with RAWm or Raws?

    • You can refer the manual for that.

      • And break my 10+ year long of never cracking open a camera’s user manual? No thanks! 😛

        • peter w

          You can download it 😉

        • LOL. I also didn’t open the manual for my D200 or D800e. But I did for this one. Technically not as I referred to the soft copy.

  • Nikkor300f4VR

    Please explain..

  • akkual

    People, important PSA: when ever you refer to these normalized figures of photonstophoto, remember that the normalization done there is just downscaling. The images have not been post-processed in any way before the normalization.

    If you shoot full 46mpix -> do the post-processing (like NR and sharpening) -> scale it down with the best algorithms, you will achieve better result than with this double resampling done in camera to get those RAW M and RAW S. (BTW. this same applies when you compare different cameras, your D850 46mpix RAW has more data to do better PP than your D750 24mpix RAW, and you may end up having better 8mpix end result by doing the PP over those 46mpix, even if graph tells you D750 has more DR).

    So the graphs do not mean, you actually get any other benefit from RAW M and S than that they make some post-processing for you automatically in camera. They do not magically make your sensor have more DR, and you can always achieve the same, if not even better, result by starting from the 46mpix RAW.

    But if you shoot a lot and want to safe space and time, RAW M and RAW S are very good options, especially if you publish them only in web.

  • TurtleCat

    Yeah, I knew all that stuff already. I’ve worked in software development for 20 years. And all of that is true in theory. The question isn’t whether 14 bit has more range of values available but whether the image itself has that many significant digits. If you’re shooting something very low contrast you may not even have an 8 bit image so the rest of the 14 bits is just waste. And as gain increases your SNR decreases thus you’ve lost significant digits. So setting 14 bit NEF doesn’t mean, necessarily, that you’ve recorded 14 bits of *good* data. You may *potentially* have 14 bits of data but you’re not guaranteed that.

    • Thom Hogan

      Dead on true. You get 14-bits of useful data only in certain scenarios, with certain ISO settings. The rest of the time you should be shooting 12-bit and saving the file room.

      • Sure you can get a sense of that by comparing 12-bit and 14-bit PDR and you can get an understanding of why by comparing the 12-bit and 14-bit Read Noise in DN.

      • Tony Beach

        Yes, but I nonetheless leave my D800 at 14 bits lossless compressed as a default because I never can be sure that the next shot isn’t one of those scenarios. When I know I will specifically not need it, and particularly want the larger buffer (an issue on my D800 even at its paltry 5 fps in 1.2x crop mode) then I switch over to 12 bits and maybe even compressed.

        • I think many of us stick to 14-bit because we can and it’s a “safety net”.
          But as file sizes increase it’s an issue that resurfaces.

          • Tony Beach

            Increases in camera file sizes are following and not ahead of the drop in the price of data storage. I bet it’s less expensive now to archive a 14 bit D850 file than it was to store a 12 bit 12 MP file back in 2008.

        • Same. And because I frequently forget to change my settings in different situations and then remember to after some time.

  • Parampreet Dhatt

    I’m honestly a little disappointed with the dpreview review score.
    Although 89 is of the highest scores ever obtained, it’s still behind Nikon D750 (90), Sony A7R2 (90) and Nikon D500 (91). Considering its image quality, dynamic range (both probably the best in its class) and well-rounded overall capabilities/feature set, I was expecting it to get the highest score ever.

    • David Gottlieb

      Take these scores with a grain of salt…. Go out and test the D850 against cameras you own. You will see the difference…. Enjoy. Be Happy, Don’t worry.

    • If you’re listening to preview, you’re listening to the wrong thing. Trust your own judgement like @disqus_lF8qlooiWA:disqus says.

    • You should just check the areas where D850 lags and where it scores in individual scoring. Judge which of them are more important to you and then decide if this camera is for you or not.
      This looking at a final number is always deceptive. Same happens when you check DXO score for lenses.

    • Michiel953

      Form your own opinion. All information, which you have to assess yourself btw, is readily available, and you could try out the camera for a day as a loaner or renter.

  • peter w

    Wow, you figure is interactive ;).


  • decentrist

    They should get a 93 for removing the built in flash…brilliant!! How about deleting the delete button followed by not allowing the camera to work on government holidays

  • Photoman

    Those charts are confusing. Why even bother.

    • The screenshots are confusing but if you visit the interactive charts you’ll probably make sense of them.
      (Particularly the Photographic Dynamic Range (PDR), it’s less technical than the Read Noise.)

  • notsosmart

    from these results and comment resposne – now I understand why average Joe Photographer does not want “too many megapixels” and why he thinks that: “more megapixels = more noise”, “real low light cameras has few megapixels” – average Joe Photographer has no idea how downsampling works (and what that is and where to get one and why he should need one)… well here Nikon did it (with those M and S raw’s) specially for you Joe – so he can enjoy those megapixels not fearing from “too many megapixels” and Joe does not need to think about that fearfull downsampling thingy.

  • Back to top