What exactly is the new Nikon sRAW file format?


Image credit: WEX Photographic

The D4s was the first Nikon camera to support the new sRAW format (aka NEF RAW size S file format) and the rumored D800/D800E replacement is also expected to have it. In the latest [NR] blog's discussions there were some confusion on what exactly sRAW is. Here are some interesting findings on the new Nikon sRAW format from rawdigger.com (there is much more detailed technical information on their website):

  • It contains not RGB but YCbCr data much like a JPEG
  • The data is 11-bit
  • The tone curve is applied to the data
  • The in-camera white balance is applied to the data
  • The pixel count is 4 times less than with regular NEF files
  • The color information is shared between two adjacent pixels (in other words, chroma is recorded for each other pixel)
  • The file size is insignificantly smaller than full resolution 12-bit lossy compressed NEF
  • Compared to a regular NEFs the data needs additional processing (linearization) while converting it to a TIF/JPEG, that may cause additional problems during the conversion (see Study 3 below) as well as some additional computational errors
  • There is some loss of color accuracy in shadows, which negatively affects the usable dynamic range if color accuracy is important.
This entry was posted in Nikon D4s, Nikon D810 and tagged . Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • Herman

    wow… speechless. they’re missing the mark by a trillion miles on this one.

    • Neopulse

      Do explain, curious to know what you meant by that? No sarcasm, just wanna know.

      • Jon Ingram

        I don’t want to presume, but I think he probably means that sRaw isn’t that that much smaller than lossy compressed 12 bit NEF, but loses significantly more data and has the potential to cause more problems in post-processing work-flow. In other words, the new file format is worse than the alternative and doesn’t save any space over the alternative.

        • Herman

          yea, pretty much Jon.

        • Eric Calabros

          But has less chroma noise
          I wonder how its noise compares to downscaling and converting to DNG 1.4

          • Sad Nikon Man

            It’s not referring to chroma noise. That’s the colourful noise you get when you put your ISO super high. Chroma in a digital image file refers to the colour information. An uncompressed RAW file will have full chroma and luma information for each pixel with 14 bit of depth. To compress an image they use algorithms to average out data between different pixels, which means you will lose some colour accuracy.

            • Thom Hogan

              Not exactly. An uncompressed raw file will have a single 14-bit value for each pixel value that is filtered (red, green, or blue). Color and luminance information is interpolated to create an RGB pixel value.

              sRaw does that interpolation, but then imposes further interpolations of linearity and white balance, and then uses techniques to keep all this additional information from bloating the file size, such as storing chroma values across pixel pairs.

              Whether people find sRaw useful or not is a personal choice. But my objection is that it is not “raw,” but rather cooked data. You have no access to what the sensor saw: you are seeing what a converter designed by Nikon was designed to give you.

            • waterengineer


            • Photo-Jack

              Hi Thom,
              let me ask you a question which you as an expert may consider as stupid:
              At nowadays pixiel density, what would actually happen, if the’d switch all information of every second pixel off. What remains may be less resolution but original data without interpolation.
              And what would actually happen, if they’d take the full information from one pixel for the next neighbor too?
              I’m really interested but am not familiar with such electronic stuff.

            • Thom Hogan

              Problem is these are Bayer sensors. So you have a row of photosites that are RGRGRGRG and the next row is GBGBGBGB. Skipped every other photosite means you lose color information.

              As I noted elsewhere, you could “bin” a group of four photosites (RGGB), but that doesn’t actually gain you much, as you’re storing one 42-bit value where you were storing 56-bits (4 14-bit values).

            • The big question is whether it’s 14-bit float or 14-bit int. (or 11-bit) I believe it’s int which means that you could use the exact same technique to compress the “RAW” image as is used for JPEG compression (but preserve dynamic range, etc.). The problem with JPEG is that it intrinsically relies on the response of the human eye (e.g. less sensitive to variations in blue) for its payoff which is why they need to bake in white balance, tone curves, etc. If this were all executed competently it would be a worthwhile tradeoff (e.g. you could expect decent compression with no visible quality loss), but “RAW” is not the word I’d use to describe it. HDR-JPEG would be more accurate.

            • UnknownTransit

              I never heard of 14 bit float and 14 bit int. If they were 14 bits, they wouldn’t be a float or an int. It really doesn’t matter. This depends on how the data is interpreted.

              A float (floating point) is a decimal point number stored in IEEE 754 single precision format. It is always 32 bits in Windows.
              An int (integer) is 32 bits in Windows.

              The 14 bit value represents the colour intensity for that specific pixel. That could be red, green or blue depending on what pixel in that bayer pattern. The other two colours are averages on nearby pixels. It’s up to Nikon to decide how to get those values. It has nothing to do with integers or floats.

              JPEG looses all the details when the DCT process is applied. The image is converted to frequency of changes from white to black (or RGB). Then the high frequencies are discarded. In photography terms, it means discarding the details. All JPEGs are the same, there is no such thing as HDR-JPEG. HDR just gathers details from multiple images. Once it’s stitched together, it’s the same thing.

            • geek or artist?

              with all this expertise in IT, BIT, TIT, IEEE…. just want to see some of your images!

            • Spelling Nazi

              ‘geek or artist?’, I believe you are browsing the wrong porn site. This is one is for camera porn, for the other kind go to pornhub.com

            • Spelling Nazi

              ‘geek or artist?’, I believe you are browsing the wrong porn site. This is one is for camera porn, for the other kind go to pornhub.com

            • mark

              I can’t help but think they would be better off releasing a number of options for the same camera body, similar to Sony’s recent releases. They could release a 36, 24 and 16 Mpx version of the D800 and I am sure it would sell bucket loads. The D700 replacement arguments are awash on every Nikon based blog site. If one side is right or wrong it doesn’t matter, people want a lower resolution D800.

            • Thom Hogan

              Been my point for awhile, though I don’t know that they need three variations for each body. They certainly need two variations for each body.

            • El Aura

              Two variations of the D4 body, two variations of the D800 body, two variations of the D600 body, two variations of the D7100 body, two variations of the D5300 body, two variations of the D3300 body, two variations of the Df body. Yeah, that’s the way to fight bloat.

            • Thom Hogan

              Not quite what I’ve been suggesting:
              D4s, D4x
              D800xs, D800hs
              Which is 8 models. Nikon currently has 10 models, many more if you could previous generations still available new.

              The D610 and Df need to be one model targeted at the economy FX buyer (consumer).

              Note also that I suggested modularity at the top of the line a long, long time ago to keep from having exactly that body proliferation. But if we’re going to have it, it needs to be logical and rationalized, not random.

            • El Aura

              I consider the D3x and D300s as previous generations still available new. They are not models consciously created to be part of the current line-up. They are models still being produced (if they still are being produced) because Nikon considers the opportunity cost of still building and selling them to worth it at the moment. Nikon does not present them differently than its (other) previous generation models.

              It’s also a bit disingenuous to present on the one hand your proposal as one of an eight-camera line-up but then still add that the D610 model has its place (which would make it nine cameras).

              Your discussion of offering different sensors in the same body so far is always a commentary on a particularly body type not being available with one of the already existing sensors. Thus, you consider there to be room in the market for a 36 MP D4x besides a 36 MP D800, but not for, eg, a 16 MP D9300 besides a 24 MP D9300.

              And the ‘modularity’ on the Sony side seems to be taken in one moment as an argument to criticise Nikon and in another moment to criticise Sony. I for once am struggling to see the point of offering both 24 MP and 36 MP in the Sony A7 line. It looks like they wanted to match Nikon with the 36 MP but also add on-sensor PDAF (which they didn’t manage to offer with one sensor).

              I have no problems with criticising companies and offering advice on how to improve their products and their line-up. What I have problems with is the certainty that ones own solution is so obviously so much better that it justifies scathing criticism.

            • Thom Hogan

              D3x -> D4x in my lineup
              D300s -> D9300 in my lineup

              Just because a model hasn’t been updated in a long time doesn’t mean it won’t and shouldn’t be considered a placeholder for the next model. We’d need to see real evidence that Nikon is never going to move forward on those. I think we see that evidence on the D700 never moving forward, which is a shame.

              Personally, I’m not a fan of a lower FX model (D610/Df). I don’t think it needs to exist. But it currently does, and in two different forms, basically. If it’s going to continue, that needs to be rethought.

              I base many of my comments on surveys that I regularly do, which now number in the many 10’s of thousands of responses from Nikon users (some of which I’ve shared with Nikon, and which they are now attempting to mimic with some of their surveys). Since 2007, Nikon has drifted away from their old policies of moving specific lines forward with updates and has more and more targeted “new users” with cameras that aren’t actually an update. This is a risky strategy for a company whose reputation was built on word of mouth from the legacy user base, and one that doesn’t seem to be working well. For every new user they pick up they’re losing an existing user these days (what I call “leaking”). The trick is to try to define a line you can entice new users with AND retain existing ones. I don’t believe Nikon is doing that.

              Which brings me to Sony. Sony is doing the same thing. Essentially they’re competing with themselves to a large degree. As Alpha DSLRs wither, they have grown the NEX/A mirrorless line to about the same size as Alpha used to be when they bought it. They have leakers out of the Alpha system into competitors, and are sucking in new customers to replace them.

              As for the A7s, A7, and A7r, I like the idea behind doing the three cameras, though I think they’ve not made it clear why you’d want one over the other (especially given the price differentials). I also think their initial responses to the shutter shock on the A7r were disadvantageous to them. If the purpose of buying a 36mp camera is to get “best possible results” then their denial of a real problem in doing that just sets up a cognitive dissonance that impacts sales. Sony should address that, add an uncompressed raw format, and a couple of other similar “quality oriented” features to differentiate the camera more.

              Back in 2001 Nikon got it right: for sophisticated shooters they want both a “performance” camera (low light, fast focus, high frame rates, etc.) and a “studio” camera (high pixel count, high quality, etc.), and ones that share the same UI, accessories, and most common features.

              I suspect that Nikon keeps getting the wrong message from the wrong data point, though. By pricing the D3x at a ridiculously higher price than the D3 when the only real change was the sensor, they essentially lost sales in the “x” line. Big time. They then looked at “we didn’t get x sales” and concluded “we shouldn’t do x.” Hmm, maybe it was the price, not the sensor difference that caused the problem?

              If you look closely at Nikon’s practices, they always try to re-establish price points with upgraded models, and they often will try to push the price point up these days. New entries (Nikon 1, Coolpix A, Df) are coming in at ridiculously high price points and then failing to sell well. Their problem is getting to be simple: they’re often not able to justify the prices for the features/performance. The D800 was an exception to that, but remember that it was being judged against the US$8000 D3x when it came out.

            • El Aura

              It doesn’t matter how something ‘should’ be perceived. It matters how something ‘is’ perceived. And I would say almost all potential D3x or D300s buyers know that they are previous generation models and perceive them as such.

              And yes, one can argue that a camera that is more expensive than another one shouldn’t have worse key spec than a cheaper camera (AF in D600 vs D7100). But only offering one affordable FX model at $3000 when competitors offer FX models at $2000 doesn’t seem like the ideal strategy either. Part of the problem is naturally that DX spoilt consumers in regard to PDAF frame coverage which makes the D600 AF look particularly bad but then the 6D and A7r (PD) AF is even worse.

              In regard to Sony one could argue that their spray and pray approach is having some success in that some of product variation strategies (A7) is meeting your (partial) approval. Offer enough different options, and some of the product variety will be considered a good idea by some people and another part by other people.

            • martin

              I believe Fuji are trying to achieve something like that, with their EXR technology

            • Paul

              so how is this substantively different from JPEG?

        • Neopulse

          But, isn’t is good to have variety (I suppose). You can choose 12-bit instead of 14-bit, with compression or no compression right? At least there are options.

        • Neopulse

          But, isn’t is good to have variety (I suppose). You can choose 12-bit instead of 14-bit, with compression or no compression right? At least there are options.

      • Herman

        the reason we shoot RAW is because we want the data capture preserved so that we can manipulate them. RAW has really been showing its value in recent camera sensor advancements since we’re capturing 14 bit of data, of which is beyond what our monitors can even display. I shoot professionally, and I shoot everything in RAW and file size has been an issue ever since my studio switched over to D800 because we’re dealing with 35-50MB RAW files without a good way to reduce the size of them. If you want RAW you have absolute no way to lower the file size; unless you go the jpg route which takes away too much comparing to RAW.

        Take the other camp, Canon (ahem, the forbidden word here), they have different RAW sizes by lowering the pixel count and not sacrificing important things like dynamic range, color accuracy loss, and for pete’s sake 1/4 the file size so that your cell phone shoots higher pixel count than that? I mean, if you take the D800 36MP and do different sizes, I don’t mind having 1/4 the pixel count. But for the D4s to downsize to 1/4 and losing the actual advantage of RAW file, what’s left is basically like a jpg file in RAW format.

        In other words, they’re missing the mark by a trillion miles.

        Clarification, okay it might be worth it for the supposedly new camera, but gosh, Nikon do it right please.

        • Jon Ingram


        • Eric Calabros

          It has more data than jpeg: 11.3 bit vs. 8 bit

          • andre_

            Both the bits count and the in-file WB and gamma are interesting.
            But the rendering issues and the file size make the more data useless.

            In other hands, I completely agree with Herman and Jon.

          • Thom Hogan

            With less and compromised color information.

            Technically, we can have 11 bit JPEGs. Personally, I’d rather see that than sRaw.

            • Eric Calabros

              I’d rather keep some ability to recover highlights behind of say, tennis player, than his hat color.. It will be different from what Canon shooters send to their agency anyway 😉 I personally never shoot sRaw or any lessRaw, I just guess maybe Nikon thought this way

            • El Aura

              Apart from being easily understandable by any application that can read JPEGs (once that applications gets over the 11-bit-ness), is there really any meaningful difference between RAW S and 11-bit JPEG?

        • Roak

          Photographers. oh. When the first thing you have to complain that it is wrong. If you get this way is also wrong. Clarify priorities and shoot small RAW with D4S or completely uncompressed RAW with the D800.

          I also shoot weddings with D800 full raw and I do not have any problem. And I buy the D4S for small raw.

        • groucher

          True except that we’re not capturing 14 bits of information. The D800, as with the D700, captures 11 bits of info at optimum ISO. If you examine the LS 3 bits, you’ll find that they contain purely random data. This is easy to confirm using spreadsheet statistical functions.

        • Michael
        • Thom Hogan

          While there seems to be a lot of agreement with you, Herman, I’d disagree on the wording. We don’t want raw so that we can manipulate it. You can manipulate sRaw files with a converter that understands them. Heck, you can manipulate JPEG and TIFF files, too.

          The reason we want raw data is that we don’t want to be locked into one interpretation of the data. Moreover, we don’t want to lose any data, either.

          One thing that ought to be clear to everyone now but was clear to me back at the turn of the century is this: software approaches to raw data change over time and get better. We haven’t even really gotten to fully multi-threaded comparison approaches yet, either, where a converter applies multiple demosaic techniques to the same data and alters its approach in areas it detects as detail versus those it detects as not having detail.

          Once raw data gets “cooked” in any tangible way–and sRaw cooks data quite a bit–you lose the ability to later pull more out of the file. So sRaw has to be thought of only as a convenience format.

          As you might be able to tell, its the event shooters looking at it mostly, and mostly because they are convenience oriented when they do things like shooting five or six weddings a weekend and have to process tens of thousands of images quickly.

          Is it good to have another convenience setting? Sure. Add it to the ones we already have: JPEG and TIFF. Does it do everything most of us would want from a raw file? No.

          • Eric Calabros

            we need more of these in-camera-cooked features, but Chef Nikon should ask us our taste. I want skin color from Portrait picture profile, color boost from Vivid, DR from Natural and some micro-contrast sauce! is it too much?

          • Herman

            Thanks for the insight Thom.
            I hear you. It sure adds to the already wide array of choice of format but I guess the issue here is that we’re calling for a smaller real raw instead of another variation of a smaller “cooked” format of data. For the photographers who shoot a lot and actually need to retain the raw files, the ever increasing resolution is creating more than just a headache. I can’t help but feel envious of the Canon camp having different raw sizes they can choose. Although you may point out how it may be inferior or they may not be real raw also because they have to down sample, it does, like you said, provide the photographer a choice and in this case a much more valuable one than what Nikon is doing with their new sRAW format.

            Let’s just say the D800 has made a splash and changed the way a lot of us shoot, but at the same time the excessive resolution (probably for 99% of us) and the huge file size along with the inability to shoot at a lower resolution do make a lot of us frown. I’m probably just more disappointed that they’re not addressing this, and when they do seem to be making an effort, they’re not really tackling what really needs to be addressed.

            Oh well

        • Neopulse

          Thanks for the reply. Cleared up what I had trouble understanding. Not very adept in multiple size RAW file formats.

        • Neopulse

          Thanks for the reply. Cleared up what I had trouble understanding. Not very adept in multiple size RAW file formats.

      • Chef Nikon

        Cooked vs. Raw
        Thats all…

        all original nutrients, but hard to digest
        boiled down less nutrients, but easy to digest

        • Neopulse

          Funny reply.

  • Marc Aberdeen

    Wow!! A whole 4mp??? Damn, what would i do with all that resolution Nikon?

    • OO7MIKE

      What did anyone do with the 4mp images that the original D2hs or how about the 2mp D1?

      They took images and made great prints with them. The 4mp sRAW images from the d4s are light years ahead in quality.

      You don’t have to use every new feature that comes out. Some folks are still bent out of shape over Live view and video capabilities in a DSLR. Nobody is forcing you to use it.

      • phil

        and vehicles used steam engines and they worked fine – that does not mean it would make sense to produce a steam car these days

    • Eric Duminil

      You could downsample it again to watch on Full HD.
      It’s more than enough for websites and small prints.

  • koenshaku

    Hmm.. Well if you are shooting sports it comes in handy be cause it takes the load off your buffering. However if you’re doing landscape work or standard portraiture you’re going to want to use RAW large full erection not semi lol.

    • bonem

      Not sure if that was a typo or not, but we are laughing! hahaha.

    • Rudi

      When I shoot sports I shoot jpg anyway. I mostly shoot raw and jpg (to be honest I didn’t even try out the sraw on my D4s).

      At least the OOC jpg is better than what I get from my RAW in Lightroom. And when shooting sports on assignment there is no time to use NX2 to develop raws.

    • Thom Hogan

      Are you sure? Because sRaw requires EXPEED to do lots of calculating, the raw data sits in the buffer longer while the chip is doing its work. I’m not sure that there’s a buffer benefit to sRaw at all, and it could be a buffer drop. Have to admit, I haven’t tested that yet.

      • Eric Calabros

        Iliah Borg:

        “With the buffer dropping from 176 frames for 12-bit full size lossy compressed to just 36 for small raws – I doubt it is good for event shooters.”

  • eric

    I doubt “chroma is recorded for each other pixel”. More likely, it is averaged between two pixels. There is a big difference between these two things: one will lead to more noise, the other to less. I can’t believe Nikon would use the first one.

  • Paul

    I’m still at a loss as to why one would want sRAW. wouldn’t the person wanting to shoot this format be better off with shooting RAW with a smaller resolution dSLR such as the 610? of course, it’s an option that doesn’t have to be used exclusively. In addition, given some of the mentioned color inaccuracies or little change in file size from lossy, compressed 12-bit RAW, why bother? flexibility or options are fantastic but I’m missing something on this one.

    • Beso

      I don’t think you are missing anything. From the description sRAW appears to be much more hype than anything else. And the price one pays in quality far exceeds any concept of what a benefit might be.

    • Eric Duminil

      Let’s dream a little :
      a well implemented D800 sRAW (14-bit, 9Mpx, ~10MB, with less noise and more dynamic range, real RAW with nothing cooked in) would give you the best of both worlds for any application (studio/landscape/sport/events/weddings/low light/…). It would basically be two cameras in one (think D3X and D3S in a small body), and the D800 wouldn’t have any drawback. An sRAW D800 would basically be the D700 successor.
      Let’s dream a little more : in sRAW, you might get more FPS and better buffer.

      As implemented in the D4s, it’s a completely useless feature. Many drawbacks and not a single advantage.

      • Thom Hogan

        Yes, you’re dreaming.

        One reason why raw files are small to begin with is that they only store one 14-bit value per data point. Even JPEG stores THREE 8-bit values per data point, though it does so with lossy compression.

        Thus, something has to give in your dream. To get the file size down, you still need to be only recording one value per data point. But to do so you have to toss information, a lot of it as it turns out.

        People ask why aren’t things “binned”? Well, you’d end up with three 14-bit data points per final pixel, a net loss of only 25% over the original raw file size, yet you’d only end up with 25% of the pixels ;~). Not a good tradeoff (and there are other problems with this approach).

        Part of the problem with sRaw is that people WERE dreaming. They demanded “raw” be far, far smaller in file size. What they got is essentially a variation on JPEG, but without the Fourier compression.

        • arachnophilia

          as i understand it, the bayer interpolation is the problem. if we could chuck 3/4’s of the pixels, or bin 4 into 1, without any regard to color, it would give people what they want. but the color filter array isn’t laid out in a way that is conducive to this. if we were dealing with a monochromatic camera, this would be simpler.

          here’s a question. what if we came up with a better pixel-skipping pattern? say, only used every other BLOCK of four RGGB pixels? or maybe something more creative?

          • Thom Hogan

            That was my point: we get one-quarter the pixels by binning, but three-quarters the file size. Not a good trade off. Now if you did 11 bit binned data, you’d get 33 bits out of 56 bits of data. You still haven’t saved half, and remember, NEF compression can manage to do that.

            My problem with what Nikon’s doing is it’s just a “Canon has it so we need to have it” approach. That’s not exactly solving real user problems. Because so many people won’t understand the details of sRaw, basically it becomes a way of Marketing to Dummies type of feature.

            A more useful feature for those event shooters would be to have a built-in raw conversion routine (there is one) that batch processes while shooting (it doesn’t). So, you take a shot, process it in camera and then tell the camera: create a JPEG using what I just did as a guide for every raw I subsequently shoot.

            Nikon thinks they have this (called Picture Controls), but that’s only part of the puzzle, unfortunately.

            • El Aura

              If Nikon can shut up part of the online critics by adding another option to their menus, I’d say that is a very cost-efficient PR move (well, it pushes some of the cost to raw converter developers for supporting sRAW). It’s like Obama putting a US flag pin on his lapel during his first presidential campaign, it shuts up some of the critics at very low cost. Sometimes doing something irrational can be good PR.

            • arachnophilia

              i think the problem is that most of the critics are smart enough to figure out that this isn’t what they were asking for.

            • El Aura

              The critics I am referring to were the ones who said: ‘Why doesn’t Nikon offer sRAW and mRAW as Canon does?’, immune to any explanations that Canon’s sRAW is about as useful as Nikon’s new RAW S is.

            • Thom Hogan

              Sure. But what if cost efficient PR moves are the primary thing that you’re doing? Moreover, what if all these types of simple moves just add up to more complexity and clutter in the menu systems?

              We call it “bloatware” in the software industry for a reason. At some point it gets tough to digest. Even Microsoft eventually recognized this with Office and changed their focus from adding features to changing UI (not that I’d argue that was done all that well).

              Cameras don’t need BandAids. They need some real surgery.

            • El Aura

              I guess, cameras also don’t need commercials. The need real improvements. Any money spent on commercials should be diverted into improving the product in a meaningful way.

            • Thom Hogan

              I don’t follow your argument. Adding sRaw was not a commercial, it was development and engineering time and money spent on that in order to claim parity with Canon for something that most users have no idea what it really is.

              The operative question is this: if Nikon spent X dollars creating that feature in development, is there a different feature that cost the same that would have better served the user base?

            • El Aura

              Whether you spend money on a commercial or on feature development, it is still money. And if you spent money on feature development that serves purely marketing purposes, that is very similar to spending money on commercials (since commercial serve the purpose of marketing).

              Yes, very short term, any human resources used for adding this feature get diverted from the development of other features (or feature improvement). But, comparing the amount of resources diverted for this with the amount of resources that Nikon can add efficiently, let’s say, per month, this is a very short term effect.

            • arachnophilia

              right, but what i’m wondering is if we couldn’t just actually throw away 3/4s of the pixels, and leave the 1/4 still uninterpolated. basically, there has to be a smarter way.

              for instance, taking a 4×4 grid of 16 photosites, instead of 4, and then binning the 4 red photosites into a single red photosite, the 4 blue photosites into a single blue photosite, and the 8 green into two green photosites. and then letting the post processing computer still do the interpolation from 1 channel per binned site.

              or just using one 2×2 grid per 4×4 grid, and throwing out the other 3 2×2 grids.

            • El Aura

              Here are some sketches to illustrate how to bin a Bayer pattern while still retaining the properties of a Bayer pattern in terms of file size:


              While this can be done, the resulting ‘virtual’ pixels now overlap. This doesn’t necessarily has to be a problem but at the least, existing de-mosaic algorithms would need be changed to take that overlapping nature into account.

              It’s certainly doable, but as with the X-Trans pattern, raw converters need to figure out how to best do this while balancing detail, aliasing, noise reduction and other aspects. I’m sure somebody must have tried to do this but that I have seen nobody talking about such a pattern might be an indication that there are some important drawbacks (in computational power needed, visible artefacts, etc.) and that for the sole purpose of reducing file size a lossy compression of the full-size raw data has much better cost-benefit ratio.

              And I would love to hear Iliah Borg comments on this, as he knows how to write de-mosaic algorithms (as do other people working on raw converters but only the open-source developers might be willing to discuss this openly).

            • arachnophilia

              i’m just positive that there has to be a more intelligent way to go about this, rather than making a larger file with less information. perhaps the answer is that as we make more pixel-dense sensors, we should find a better interpolation pattern that allows for it.

            • El Aura

              Nikon’s lossy raw compression reduces the file size already by about 50%, going from 14 to 12 bit reduces the size by another 20%. That is so far the smartest way to reduce file size. The higher your ISO setting, the more you can reduce the bit depth further. This could be automated, ie, adjusting the bit-depth based on ISO automatically. Equally, the lossy compression could be cranked up, a Bayer sample of the scene projected by the lens is nothing but a lossy compression of the information (only recording a third of the information for each pixel). When you, during processing, downsample, you automatically gain bit-depth. Thus, if your size needs are small, reducing bit depth when recording the image can go a bit further.

              The second half of the equation, reducing processing time in the computer (up to the point of improving responsiveness of image adjustment tools) could be improved by raw converters don’t applying a traditional de-mosaic algorithm but just a 2×2 binning as a very simple ‘de-mosaic algorithm’. RPP, eg, offers this. Another option would be to downsample first (directly after de-mosaic) before applying any further adjustments and not only at the end right when exporting images. But these are functions added to raw converters, not to cameras.

            • neonspark

              There is nothing wrong with matching canon on it. After all they are not in the business of making cameras for the good of the planet, they are in the business of making money for their shareholders: and if doing this helps them sell more, they can continue to make more cameras. That is how every business works.
              They can off course implement this differently, or better yet, produce a camera that is more budget friendly to event shooters. However with 50+MP bodies on the horizon, how soon until 36MP becomes the new sRAW he he.

  • uggyy

    I can see a press tog using it. Think smaller file size Adv but in raw for sending to editor. Can see some uses but not what people want IMO for the d800 upgrade.

    • Byron B.

      Canon guys have had sRAW and even sRAW2 for a long time. I think it is supposed to be useful to boost buffer limits in situations where you need more shots per sec, and you know you won’t need the resolution (web publishing). Also it effectively is 2/3 less space but just half the resolution, by some weird mathematical magic.

  • Todd

    I’m a newbie at all this stuff. I don’t understand investing a big chunk of change for a camera that is capable of collecting an awful lot of data….and then tossing 3/4 of it away.

    • csmith

      if you don’t get it then you don’t need a d4s. not trying to be mean but the d4s is for sports and press pros, not hobbyists and rich dads. rich dads want to stare at their computer monitor for hours and count pixels…pros want files small enough to upload straight to a smartphone in real time where they can be ftp’ed straight to their agency

      • Todd

        That’s precisely my point. I read the article as an ‘improvement’ to the D800. While I take no offense, my point stands. If you have a reason to collect all that data on an 800, why give it away in an algorithm.

  • It’s probably useful for a small handful of photographers. But, I don’t see it as a feature Nikon would spend a lot of time talking about. It’s an option to add to the image quality list, just like uncompressed vs. compressed RAW. Kind of a ‘tweener.

  • Beso

    Just reading this post tells me all I will ever need to know about sRAW – you get smaller files but at such a huge cost it makes no sense whatsoever. People who are bothered by file size from a 36MP camera shouldn’t be looking at 36MP cameras. Pretty simple.

    • fjfjjj

      Yeah, they can buy the D710 instead.

  • To quote just about everyone below me, I’m glad that all your years of hard work in electrical engineering and digital signal processing has paid off.

    • csmith

      HAHA +5000

  • Tim

    Am I reading that right it says the files are INsignificantly smaller (as in almost as large)? Is that supposed to say significantly smaller? I’ve never seen one of the files so I can’t say, but I would assume the whole point was to reduce file size.

    • Jon Ingram

      You are reading it right; the files are Insignificantly smaller than lossy compressed 12 bit raw. Some people might like that some in-camera processing is “cooked” into the lossy raw file with the new format. Not me, but somewhere out there maybe

    • Eric Duminil

      You read it right, sRAW files are 1MB smaller than RAW.

      • arachnophilia

        one of iliah’s posts below seems to indicate that they are 1mb BIGGER than the equivalent 12 bit lossy compressed raw.

        • El Aura

          It’s not exactly rocket science to Google: Nikon D4s manual pdf and extract the actual file sizes from it:
          – 14 bit uncompressed raw: 4928 × 3280 -> 33.6 MB
          – 12 bit lossless compressed raw: 4928 × 3280 -> 15.4 MB
          – 12 bit compressed raw: 4928 × 3280 -> 14.1 MB
          – 12 bit small uncompressed raw: 2464×1640 -> 13.1 MB

          D4s manual

          • arachnophilia

            tables in manuals and actual tested numbers are frequently not quite the same.

            • El Aura

              Sure, (lossy) compression can be dependent on image content but the size of uncompressed images is simple math. But until we have comprehensive test data, taking the official ones as the basis of our discussion is much better than building arguments on nothing but thin air.

            • arachnophilia

              iliah posted this elsewhere in the thread. i wouldn’t call it “nothing but thin air”.


  • pete

    who says “four times less” instead of 1/4th????

    • factor four less resolution!

      • pete

        yeah. in school we were taught that was 1/4th.

        • peterw

          some are non-natives, went to school and learned French, Swahili, Thai, Chinese or German
          Some didn’t go to school and yet became intelligent photographers
          Some didn’t even become photographers, but are trying to make themselvers clear on internet.

          And some just make incidental mistakes.

          so many possibilities

          thanks Nikonrumors for presenting this sRAW thing.

    • waterengineer

      Engineers do, thank you very much.

    • Kyle Farris

      “6 in one hand, half a dozen in the other.” It’s the same meaning said a different way. Division is just the opposite of multiplication and it’s equally as easy to understand either, IMO.

  • Arthur Tazo

    The “s” in sRaw stands for the four letter word that rhymes with hit.

  • How does this compare to Canon’s sRAW scheme, I wonder?


    This is a step up from a jpg file and that is about it. This is very similar to sRAW that cannon came out with a few years back. Its good in a pinch when you are just about out of space on your card. It came in handy on my 40d.

    I really wish they would have made sRAW a pixel binning format. Smaller file size, very low noise, sharp images. Infinitely more useful.

    The best part about a new feature… You don’t have to use it!

  • Getho

    wake me up when current nikon management are all sacked

  • Mardock

    So basically there’s absolutely no reason to want it. Got it.

    • Thom Hogan

      The reason to want it is that Canon shooters have it. ;~) That’s also the justification Nikon saw. That’s not the way to design “best possible” products.

      Design is tricky. You have to listen to users, but not literally. Nikon users have had serious ego hits because Canon users had something they didn’t that sounded like it was more useful than it was.

      There were other approaches to solving the user problem (less file space, smaller image size, more buffer) than either Canon or Nikon chose.

      • Canon sRAW is implemented in essentially the same way? If I had know that, I don’t think I would have coveted the feature so much. I don’t think I quite had the ego about it, but I did think it was legitimately a great function. Now, not so much.

  • Louis-Félix Grondin

    Is there anyone here that actually used the format? I’d like to know if it’s as bad as it sounds(reads)…

    Did someone find a good use for it? Maybe for some people who send their images quickly to different medias that function came in handy, but I struggle to see any other good usage of this format.

  • rt

    I can see this being useful for burst shooting / high frame rate shooting, assuming that what they’re doing with the sRAWs is a “half-size” demosaic, treating each bayer 4×4 group as a single pixel (1 red (Cr), 1 blue (Cb), two averaged greens (Y), though they might incorporate luminosity values from all 4 pixels rather than just green) – which theoretically should give you the least amount of chroma noise/false color – Most of what you would be losing out on is luminosity resolution – short of a magical demosaicing algorithm (though I suppose many are) you shouldn’t be losing much, if anything in chroma.

    Where you’ll run into the most problems is if you want to edit the white balance because you’ll be running two functions over the data, where with the raw you would be only editing for one – loss will be inevitable.

    In other words, if you are used to RAW, keep shooting in RAW. If you want extended bursts of something that holds more information (at the very least way more chroma information) than the usual camera produced JPG and yet can be calculated much faster, while still being small enough to not fill up the buffer, sRAW sounds like a viable option.

    In -theory- at least.

  • Well I shoot Nikon JPGs and never bothered with RAW at all but guys let me tell you have you seen a JPGs from FUJI X series? Nikon learn from competition how it can be done 🙂 . Now this sRAW nikon thing does not make any sense to me at all or am I missing something here?

    • rt

      That of course all has to do with the lack of an AA filter combined with a sensor array that has -slightly- more emphasis on luminosity (greens) which is more what our eyes tend to care about, while still being clever about how it arranges R and B, and having to go crazy and back to the drawing board when it comes to demosiacing. / It’s impressive, but I doubt it’s anything Nikon will ever touch. I suppose it makes sense as Nikon was traditionally more about optics and Fuji more about film. (?)

      • Thom Hogan

        I think he’s referring to color and saturation choices Fujifilm makes in their JPEG files. Nikon is much more neutral than almost any of the other camera companies. Fujifilm uses a double hue shift similar to what they did with their films.

        • Coloretric

          Also Thom, isn’t Nikon using an old style of JPEG algorithm in their bodies compared to some other manufactures? I could be incorrect of course and can’t fact check as I’m in my mobile in a laboratory 😛

    • DMT

      You are DEFINITELY missing something but it is *not* sRAW.

    • gizzo

      Make sure that you already know that Fuji cheats with their ISO settings Files…Iso rated 3/4 stops slower than Canikony..;)


      Nevertheless I cannont understand why Nikon not introduce different smaller raw formats (Sraw, Mraw, Lraw) without less impacts on the IQ/bit so that everybody can choose depending on his situation which one he wann use….

      Is this impossible….?

      Additionally i cannot understand why they not introducing the d7200 (d7100 first) and the d800(e) successor later with more/major improvements….

      Nikon why not be innovative….standing out of the competition…even it is not common for so called “pro” dslrs…with features like

      -fully swivel/articulating lcd screen ( like d5300/ Sony a99 ff pro dslr)
      – integrated wifi (less stuff to backpack, you wanna more money? just make the dslr 100-150 dollars more expensive…)
      – put in useful u1 and u2 modes (as in d7XXX /d6XX)
      – real quite shutter mode (d610 still loud could be easily less noisy)
      – crop modes of the d7100
      – less shutter lag
      – larger viewfinder magnification
      – long time exposure 60 seconds like omd em1
      – panorama stiching mode
      – better video functions/resolutions/bit frames (internal 4 k would be a dream)

      so many things which could easily been improved and make the d800 a nearly perfect camera….

      dont tell me panorama, wifi , u1/u2 modes, fully swivel display is not for pros or they dont use this/need this….

      it would be so useful in al ot of situatuions (not lying on the ground for shoots, architecture, nature landscape shots etc., quick change of settings e.g. one video, one for stills, less stuff to carry Wt /wu 5 fuccing adapter…)..

      But i know all of this is not coming….nikon doesnt have to be innovative anymore too big (money, power) to fail….

      why do more when it is enough to do less???

      • nwcs

        Most places report 1/3 stop difference max at higher ISOs. This guy didn’t exactly do scientific analysis.

        • NoMeJodas

          I didn’t know that one would need a “scientific” method to judge that one image is seriously underexposed when shooting with two different cams under the same circumstances and exposure settings. Can you please post a link to one of those places?

          I discovered accidentally that the X-E1 underexposes by a far margin when I was using it side by side with the D600 sometime ago. After doing some tests I came to the conclusion that the X-E1 was underexposing by 2/3 stop at ISO 6400 and 1/2 stop at ISO 3200 compared to my Nikons (D600/D7000 at that time). In lower ISOs it underexposed by up to 1/3 stop which I consider “inside the tolerance limit”. All in all I thought that this is yet another bug in the long list of X-E1 bugs then. That’s why I was surprised to see that this seems to be still a current issue. OTOH no reference to the used X cam was made in the YT video. So who knows…

      • Well I look at this from the most important point of view and that is the image quality which is outstanding. If the ISO settings are pumped up then still at 1600 there is no noise when used with f2.8 lenses, so imagine if I use f1.4 or 56 f1.2 I could use ISO 400 instead and still be able to work… impressive to my opinion and files and dynamic range simply amazing :)…. but my workhorse is still D3s just to make clear myself Nikon please follow fuji customer service and innovate new stuff like they do 🙂

  • How about inventing a sRAV format that is actually useful? For example: using half of the pixels count instead of just 1/4th. Maybe then even someone would use it – especially if it would result in a better ISO performance by the lower resolution.

    • Nice idea, too bad it’ll then compete with D4s, that’s why Nikon will never do sRAW the way most sensible people think it should be implemented.

      • I would probably end up using only sRAW as a standard setting on the D800 if it would provide 18MB and better high-ISO results 🙂

    • stoooopid

      Flow – the better high ISO performance of lower MP sensors is because of the larger photo sites of lower pixel density sensors. Discarding half of the pixels captured would not equate to having larger photo sites. Now – maybe the could play some tricks with averaging those photo sites and cancelling noise – but I am still not sure that would work the same as larger photo sites.

  • Why is there no option to only save the actual sensor data? Every pixel below the bayer filter only returns one value, not R, G and B, so saving this would reduce the file size to 33%, wouldn’t it?

    • Thom Hogan

      That’s the definition of raw. A raw file stores 14-bit values for each photosite, not RGB data.

  • DeepCVisualArts

    I want to drag the little away though it is important that I inform you all. I have been using d7000 since 2010 and d800 since 2012. I have not seen any difference in quality of 12bit and 14bit from my d800 yet (please make the word yet, which means old ACR vs new ACR). On the othe hand when I processed my old d7000 files from 2010, in new ACR 8.3…..wow I saw a huge boost in the dynamic range in the same photo. I am sure new ACR can get a DR upto 16 stops easily. So mind it before you settle for an sRAW. You are not being future safe in this case.

    • Coloretric

      8 bit or even 10bit via FRC is the limit of the monitor to display files. In print there may be a negligible difference.

      • DeepCVisualArts

        I use 10bit monitor for processing my images and I can confirm that I got much more dynamic range from the latest software. And 10 bit can give a huge DR if processed carefully with new ACR.

      • DR range is important for processing, not display. When pushing shadows, my d800 files could be pushed a stop farther then the raws I see from the Canon 5D MK ii.

        Let’s say you’re photographing a high dynamic range scene – setting sun in the frame, below a cityscape. You’d probably underexpose a bit to get the details in the sky, while the rest of the frame will be underexposed by 1-2 stops. Then you decide to push the shadows a bit to get some details in the city. This always add noise. The more DR your sensor can capture, the less noise you’ll get.

        What your monitor can display is irrelevant. Jpegs (8bit) will simply look cleaner from a rich raw file when you need significant processing.

    • nwcs

      There are some differences in the shadow area. Subtle, but there. And the differences can show up in other types of photography. In astrophotography, for example, there are lots of extreme manipulations done and having 14 bit data gives you just a bit more leeway when you do image stretching.

      • DeepCVisualArts

        Agreed and probably future will bring much more obvious differences.

    • Iliah Borg

      > I am sure new ACR can get a DR upto 16 stops easily
      Given the lenses have 12 stops max, hardly.

      • DeepCVisualArts

        Of course lenses have to be pro grade ones. But DR is limited on the best lense also? How?

        • Iliah Borg

          It is limited because of flare, with any lens. It is something lensmakers know for decades. The moment you go past 11 stops (remember Adams, and think why he limited everything to 11 Zones?) veiling glare is huge, linearity is completely lost, and the image starts to look artificial CGI.

  • Creamcorn

    Oh sh*t..this hurts me head??I will continue to shoot..large raw files. Sheeze.

  • avidsiman

    sRAW as described here seems pointless.

    I have a D800e. When I want RAW image less than 36 MP, I change to a DX lens and let the camera crop the image to something easier to handle which usually ends up around 20-22 MP. While DX lenses are not quite as sharp as the FX lenses, they’re still plenty sharp for most purposes and easier to tote around. I usually use this technique for situations where I need a steady shot and don’t have a tripod or stable mount available and must shoot handheld. The DX lens does a better job of hiding camera shake and the savings in file size allow me to take multiple shots without giving up space on my flash cards.

    If I’m shooting anything under controlled conditions such as a landscape or portrait with the tripod – 36 MP full RAW it is.

    I avoid JPEG at all costs because it shows it limitations really quick in the editing room as I like to push color and brightness a lot for custom looks. JPEG breaks apart to easily ruining the shot.

    • dude

      Why not just use the 1.2 crop? Gives you 24mp, wider than DX, and no fumbling with switching between DX and FX. Oh and you get 5fps, just like the D800s

      • jr456

        It has to be a bit harder to compose shots in either DX or 1.2 crop mode though…right? The viewfinder is still showing the entire full frame I would think.

    • Sheikher

      “The DX lens does a better job of hiding camera shake”. How is this possible? Or is it because the image is not sharp now because of poor lens sharpness instead of shake?

    • nwcs

      The resolution is what shows camera shake, not the size of the image circle from the lens. And DX lenses are not inherently less sharp than FX. They are as sharp as they are designed to be. Some DX lenses will be sharper than FX and vice-versa.

    • Just FYI, you don’t have to put a DX lens on your camera to get DX size. You can change the format any time, you can even assign a button combo to switch between 24×36, 24×16 30×24 (5:4 crop), etc. Unless you want to duplicate the exact focal range – but then, the DX line-up doesn’t would not support that (maybe 3rd parties).

  • Eric Duminil

    Yes, we know that the current sRAW is shitty.
    Is there any reason it has been implemented this way?
    Are there better alternatives?
    I cannot believe Nikon engineers are that dumb, and there must be some reasons that we armchair experts don’t understand.

    • Iliah Borg

      > there must be some reasons
      that Nikon are not willing to demonstrate for some reason

  • brunoalmeida

    2.5k raw, maybe for use in video mode in a future firmware update.

    • Yes, because Nikon is famous for adding new features in their firmwares :~)

  • SportsPhotoGuy

    “Four times less…” So it is -3x as many pixels? It sucks pixels into an alternate universe?

  • Honestly

    new sRaw File/Format summarized with one word = “useless or bullshit” and in terms of Nikon language.. I AM Bullshit….

    But they can say we’ve done something….superficial measure and bluff package ….

    Bravo, Chapeau Nikon …all of us making a bow!

  • Thom Hogan

    I’m struck by many of the comments here and in conversations I’ve had with many other photographers about sRaw that we’ve hit another of those “good enough” barriers.

    Too many don’t seem to fully comprehend what happens with Bayer sensor data (at every stage of the process from collecting data to displaying it). Thus, they get into their assessment totally visually. And thus, the default Adobe ACR conversion looks “worse” than the default Nikon Capture NX2 conversion, so Capture NX2 is “better.” Or “I can still change things after the fact with sRaw” so it works just like I think raw does.

    The film analogy is that the lab you use to process your film just changed the chemical mix they use to develop it. But you’re okay with that because “it looks okay.” Maybe you didn’t even see the change.

    So in that sense, sRaw makes sense. Unfortunately, it really doesn’t save any file space over Compressed NEF, so you really aren’t getting a benefit there. The only place you might get a benefit is if you batch process the sRaws, and that’s because they’re 9mp output versus 36mp output. In other words, its a convenience format where your computer will seem faster on batch processing tasks.

    • jtra

      Another disadvantage of sRaw is that it reduces buffer space significantly according to D4s manual (36 images for sRaw, 176 for 12bit compressed raw while file size difference is just 7%). This may not be issue in D4s where buffer is already large, but it could be in D800s unless Nikon has improved sRaw implementation or made buffer really big. My guess is that D800s will not be a dream event/wedding shooter camera even with sRaw as long term storage requirements (raw archive) are not improved much and buffer will be significantly compromised.

    • Reality Check

      I don’t get the feeling that Nikon users are satisfied at all with sRAW in its sad, current state. The consensus to me seems to be that it’s a general disappointment because it fails to provide users the full image quality benefits of RAW post processing even while not saving any card storage space. And who cares if their PC can process the sRAW files faster if tone, white balance, and bit depth data have already been compromised?
      The sRAW that the community really wants is one that intelligently bins groups of pixels together to a achieve better signal to noise ratio for high ISO scenarios, outputs at 12 or 14 bit depth, doesn’t apply white balance or tone curves, and stores as at an appropriate fraction of the full RAW size from the sensor. What are the technical barriers that would prevent Nikon from giving this to us?

  • Sebastian Rasch

    I think instead of “insignificantly smaller” it should be “significantly smaller” since sRAW files have about half the size as RAW files @NikonRumors:disqus

    • Iliah Borg

      For D4s:

      ImageWidth ImageHeight BitsPerSample NEFCompression filesize

      4936 3288 14 Uncompressed 32 MB

      4140 2736 14 Uncompressed 23 MB

      3244 2144 14 Uncompressed 15 MB

      4140 3288 14 Uncompressed 27 MB

      4936 3288 14 Lossy (type 2) 15 MB

      4140 2736 14 Lossy (type 2) 10 MB

      3244 2144 14 Lossy (type 2) 7.0 MB

      4140 3288 14 Lossy (type 2) 12 MB

      4936 3288 14 Lossless 17 MB

      4140 2736 14 Lossless 12 MB

      3244 2144 14 Lossless 8.1 MB

      4140 3288 14 Lossless 14 MB

      4936 3288 12 Uncompressed 25 MB

      4140 2736 12 Uncompressed 17 MB

      3244 2144 12 Uncompressed 11 MB

      4140 3288 12 Uncompressed 21 MB

      4936 3288 12 Lossy 12 MB

      4140 2736 12 Lossy 8.3 MB

      3244 2144 12 Lossy 5.6 MB

      4140 3288 12 Lossy 9.8 MB

      4936 3288 12 Lossless 13 MB

      4140 2736 12 Lossless 9.5 MB

      3244 2144 12 Lossless 6.4 MB

      4140 3288 12 Lossless 11 MB

      2464 1640 12 Small 13 MB

      2048 1360 12 Small 9.0 MB

      1600 1064 12 Small 5.7 MB

      2048 1640 12 Small 11 MB

      • arachnophilia

        am i reading this correctly? the small raw filesize is actually BIGGER than the equivalent 12-bit lossy fullsize real raw?

        • Iliah Borg

          Yes you are reading it correctly.

          • arachnophilia

            well, that’s especially useless then. what were they thinking?

            • Iliah Borg

              You may note that lossless 12 bit is about the same size too. I can upload the files if you wish to see for yourself.

            • arachnophilia

              this may be the single dumbest thing i’ve ever seen nikon do. what possible advantage could this have?

            • El Aura

              Shutting up the critics that demand that Nikon add sRAW because Canon has sRAW. If years of reasoning have not managed to convince ‘the Internet’ that Canon’s sRAW is of little use (if the option of a lossy RAW file exists, as it does with Nikon but not with Canon, where thus sRAW makes a bit more ‘sense’) just adding the bloody thing seems like the only way to achieve this.

              (And there are situations where 11-bit JPEGs can be useful, simply because they need no raw converter, they just need software that can read 11-bit instead of 8-bit jpegs. But maybe marketing stepped in and said that if only a few applications can deal with 11-bit jpegs than those few applications act as the same bottleneck as raw converters, why not call the 11-bit jpeg a RAW S.)

            • Iliah Borg

              The files are here http://updates.rawdigger.com/data/D4s/
              and the text file contains descriptions so that you can decide what files you want. See the dimensions of sRAW files are not exactly half of regular RAW files, but in fact they exactly the same as for out of camera JPEGs.

            • Ernesto Quintero

              Ah, they were thinking marketing to match Canon’s offering.

      • Sahaja

        What is the difference between the Nikon implementation of sRAW and Canon’s?

        Is Canon’s implementation any better?

  • decisivemoment

    So it absolutely mangles the file while barely saving any size.

    On a D800? I’ll shoot JPEG over this, any time.

    On a D800s, with probably comparable RAW but a full stop better JPEG if it follows the D4s pattern? What I said above, times a hundred.

  • Phozo-Jack

    Now that Sony has supplied MF-makers with a 50MP CMOS sensor it may be another game than back then when DXOmark testet the MF cameras with CCD sensors.
    CCD sensors have the name to have a better tonal response (but that of course is a factor Dxo mark ommits!)
    With Sony’s new 50 MP sensor there is a clear gain in high ISO and Dynamic Range compared to the old sensors.
    However, following up the comparison between the D800E and the Leica S2, what would motivate me to take the higher costs and weight in account. I think, Leica needed to do a lot of homework to maintain their MF position, and probably they did as a Leica S3 probably with a 50 MP sensor is rumored for the Photokina.
    In an above mentioned article a Nikon Manager is quoted, that Nikon doesn#t have sufficient MF experience to got to market. But the solution may be easier than he thought. Nikon has a lot of DSLR experience. That points to something like the Leica S2 or S3. Sigma for one, did not gun for Canon or Nikon but for the best 50mm there is, namely the Otus. Even if they can’t really surpass the Otus, at a third of Otus’ price point it will be a success.

    As far as I can see it, both Canon and Nikon may come in difficulties to maintain there mount and the reqired IQ of lenses, if MP goes up to 50 and more MP. I assume they will need to build a larger mount sooner or later anyway.

  • Spy Black

    Well, technically that may explain some things, but why would you want to use this format?

  • Lee Myers

    So I am supposed to pay north of 3 grand for less quality? Are you sure we aren’t talking about Obamacare?

  • Greg


  • DeepCVisualArts

    Why can’t nikon average out 4 pixels into 1 and provide a 14-bit raw? Will it incur a smooth gradient absence?

    • Iliah Borg

      Averaging 4 pixels eliminates just one pixel. File size is 3/4 of the original. Not sure it improves things. 10 to 12-bit jpeg might be a better option.

      • El Aura


        How difficult would it be to ‘de-mosaic’ a ‘Bayer-like’ matrix obtained by averaging four red, blue, and green pixels each to simulate larger pixels as illustrated in these drawings?


        • El Aura
        • Iliah Borg

          Not difficult at all. But displacing pixels one looses more acutance compared to just a lower resolution sensor.

          • El Aura

            Thanks, makes sense. My other idea was to de-mosaic the raw data, downscale the image and then do a reverse Bayer de-mosaic action to ‘compress’ the image by a factor of three.

            This seems like the most straightforward solution to the question of how to reduce the file size by the same proportion as the image resolution. But apart from the computational resources needed for that, I would expect that to generate artefacts (incl. loss of acuity, etc.).

  • It takes some getting used to but the viewfinder does show the DX crop area in the middle of the frame and as long as you keep your image inside that box you’re OK. I sometimes use it when I’m shooting sports with my D800E. At first it was very distracting but after a while you hardly notice it.

  • Ernesto Quintero

    No, you’re thinking of 401k fees in your retirement plan.

  • Jordan

    Couldn’t there be an mRAW? Something in-between that gives us half the resolution but without the other drastic cutbacks?

    • El Aura

      No. Can you take a car and only use it with three wheels to save money and weight?

      • Jordan

        Err they’ve already done it with sRAW, so why not something in between?

        Also, terrible analogy.

        • El Aura

          The reason that sRAW is pointless is that a re-sampling of the resolution of Bayer data requires a de-mosaic of the data first, which triples the amount of data. So, any resolution reduction by less than a factor of three will result in a file that is larger than the original raw file.

          Can you tell me how anybody who has understood the above two sentences (the key message of the original linked article as to why sRAW makes little sense and of half the comments here) doesn’t see clearly that cutting the resolution into half makes zero sense because it actually increases the file size?

  • neonspark

    that looks like a very compromised implementation. I’ve avoided sraw like the plague before but seeing how Nikon took this direction makes me glad I never consider it worth it. Then again I don’t belong to the spray and pray crowd for if I did, I’d be shooting with a D4s or a D700.

  • mikeswitz

    No, they used sRaw to cover up Benghazi.

  • In short, it’s not raw…

  • JonS

    I have a D800 and a D4, S-Raw seems a bit pointless, if you have enough money to buy D4s and the upcoming D800s/810 your going to buy good memory cards and shoot RAW uncompressed as a pro photographer.
    128GB is plenty for the D800. When the card fills, dump the files to a PC.

  • Back to top