The difference in RAW files. 12 bit raw VS 14 bit raw

It is often said that RAW image format provides maximum image quality. On the one hand, RAW is the raw data from the camera matrix, on the other hand, RAW has its own special settings.

RAW - thoughts from Radozhiva

RAW - thoughts from Radozhiva

I will write only about the Nikon system, since I work with it, and over the years I have gained some experience working with RAW files.

Nikon DSLR cameras have some RAW features. RAW files themselves have an extension NEF (Nikon Electronic Format - Nikon Electronic Format). In the camera menu, in order not to confuse a person who has read a lot and heard a lot about RAW, marketers carefully write NEF (RAW).

In the menu of some cameras there is only one item that is responsible for setting the RAW. It "Record Image NEF (RAW)". Depending on the type of camera, in this menu you can choose color depth and compression type. And in some cameras there is generally no way to select the RAW setting.

For example, in advanced cameras by type Nikon D300, D700 and even in Nikon D7000 You can select the color depth for RAW.

Also, in some advanced cameras, for example in the same Nikon D300, D700 you can select the compression level for RAW files, the “TYPE” setting is responsible for this:

  1. Lossless compression
  2. Normal compression
  3. Without compression

Thus, we can use a combination of 6 RAW recording options

  1. 12 bit lossless compression
  2. 12 bit regular compression
  3. 12 bit uncompressed
  4. 14 bit lossless compression
  5. 14 bit regular compression
  6. 14 bit uncompressed

Attention: 14-bit RAW reduces the burst speed in some cameras, such as Nikon D300, D300s in RAW mode, 14 bit burst speed drops to 2.5 fps. D7100 with a color depth of 14 bits can not shoot at a speed of 7 fps in 1.3x mode. D5300 и D5500 with a color depth of 14 bits can not shoot at a speed of 5 k / s.

The main question is how much compression affects image quality?

It is logical to think that with Lossless Compression the camera simply compresses the source material using algorithms that do not lose data after decompression. For example, this is how an archiver works on a computer, with which you can archive or compress files, and then unzip them back without losing data. In practice, Lossless Compression practically does not 'truncate' the original data. "Normal compression" implies compression using incomprehensible algorithms, and, most likely, with a strong loss of quality. "No compression" - this option allows you to get the real raw data obtained from the camera matrix without any compression or cropping.

Attention: the frame counter correctly shows the number of frames that can fit on a camera card only in the “WITHOUT COMPRESSION” mode, since in this mode all files have the same weight and the camera knows exactly how many pictures can be placed. In the modes “Lossless compression” and “Normal compression”, the frame counter on Nikon cameras is very wrong, since the camera does not know exactly how much it will be possible to “shrink” the resulting image, therefore, it takes the maximum possible volume and calculates quantity. It is sometimes unnerving.

What type of compression to choose?

  1. Lossless compression - the best option, no loss in compression, strong space savings and preservation of the full potential of the RAW file
  2. Normal compression - the option is suitable for those who, on the one hand, want to shoot in RAW, and on the other hand, greatly save on disk space
  3. Without compression - the best option, no loss in quality, the absolute maximum that the camera can give out. A serious drawback of the format is the huge file size.

For example, not to talk about abstract things, my Nikon D700 in RAW 14 bit mode and without compression produces files of 25MB each. When switching to RAW 14 bit with lossless compression, files average 15MB. Get 10MB savings in one picture, you see, the gain is simply colossal. Well, if you use RAW 14 bit with conventional compression, the file weighs from 10 to 15MB.

What color depth to choose? 12 bits or 14 bits?

14 bits are better than 12 bits. True, the psychological aspect can play a huge role here: when shooting at 14 bit color depth, you can be calm that you use the maximum that the camera is capable of. Personally, from my practice, 14 bits seems more flexible for processing.

Personal experience:

On the Internet you can meet a lot of battles on the topic 12 or 14 bits, but I’m sure that if you show two pictures with different color depths, it will be very difficult for professionals to distinguish from each other. I use RAW 14 bit + lossless compression on cameras by type Nikon D700, D300, D300s, D7000 etc. and RAW 12 bit + uncompressed on cameras by type Nikon D200. Well, on amateur cameras like Nikon D40, D40x, D60, D80, D90, D3000, D3100, D3200, D5000, general there is no choice, it uses the default 12 bit + lossless compression. In the cameras Nikon D5100, D5200 uses 14 bit + lossless compression. You can also read my thoughts on the topic software implementation of RAWand RAW benefits when editing photos.


RAW has different compression levels, with which you can seriously save space on the card and the hard drive of the computer.

Material prepared Arkady Shapoval. Training/Consultations | Youtube | Facebook | Instagram | Twitter | Telegram

Add a comment:



Comments: 84, on the topic: The difference in RAW files. 12 bit raw VS 14 bit raw

  • Kostya

    Clearly written! And people also forget that when picking at compression or proving the advantages of 14-bit, there is also a monitor on which to watch all this, and if it is not a super expensive IPS monic or at least a good old ray tube - you simply will not notice all these little settings. on Monica for 1500 UAH!

    • Alexey Solovey

      Kostya, I completely disagree with you: if everything could be seen on the monitor, then there would be no need for serious tests - the same DxO software for matrix tests was developed by serious scientists and costs very good money. In addition to being an amateur photographer, I am also a professional film and television operator: no one believed in film cinema, for example, even such a simple thing the distance to focus was measured with a tape measure and not by eye, there was a densitometer to test the film in any film processing workshop , the adjustment and adjustment of the densitometers (for example, the calibration of the color temperature and brightness of such a banal detail as the light bulb of the densitometer) was carried out by a special all-Union service, which sealed the densitometer after adjusting the densitometer. In television - despite the very expensive broadcast-class monitors (they don't use design Deeds-Nekami-Apple in television) - no one believes them by sight either: without an asselograph-vectorscope, there can be no question of evaluating the technical side of a picture. When filming a serious movie with a TV camera, a standard gentleman's set of operator: a modern professional color meter, a modern professional bright luxury meter, a modern professional asselograph vectorscope and a modern professional broadcast-class monitor - don't believe your eyes!
      So referring in many cases to the test results of the same DxO, Arkady is absolutely right.

      • Alexander

        Well, where with all this b .. technology are good films with a deep understanding of how to use color, density, etc.? everything is there and the result is not enough .. What is it?

        • root

          The point is not that visible on the monitor or not. When you draw shadows against the background of a bright object with 14 bits in the shadows, something will appear and with 12 bits you will not pull anything but noise.

          • Anton

            When you draw shadows against a bright object, you will see that they are hidden by highlights and hares. And that 12 bits, that 14 does not play a role here - optics will be the bottleneck

      • anonym

        Sharish brother)

  • Mogwaika

    + can be stored in dng, the format is universal and compresses without loss, convenient for LR users.

    Another would be jpeg with different compression rates for comparison ...

    • Arkady Shapoval

      JPEG everything is clear, there is a strong linear dependence on the compression level. DNG is a really good standardized format.

  • Anatoly

    Arkady, have you switched to FX?

    • Arkady Shapoval

      No, I didn’t go over. The D700 is still an addition to my D200 and D80 and the crumbs of D40, since I do not have lenses (especially wide-angle) for the D700.

  • rescuero

    As always, it is very sensible and interestingly written!
    A small correction - Nikon D5100 uses NEF 14-bit with compression.

    • Arkady Shapoval

      Corrected :) thanks for the vigilance.

  • Novel

    oh and when will the D700 review be ??? very interesting to read =) I will look forward to!

    • Arkady Shapoval

      As soon as I prepare, I’ll do it right away.

  • Nicholas

    Arkady, thank you very much for the interesting and, most importantly, very useful information! I am always pleased to read your analyzes and tests.

    • Arkady Shapoval

      Rather, reviews and tests :) not at all :) today there was an MTO review, take a look at the main page of Radozhiva

  • Alexey Solovey

    As for the topic, they explained to me this way: when looking at the source, the difference between 12 and 14 bits is really not visible, but the 12-bit RAB when processed in the editor quickly - literally with every operation - loses its properties. In contrast, a 14-bit file is much more tolerant of lengthy multi-pass editing. Therefore, if long processing is expected, 14 bits are unavoidable.

  • Alexey

    “In the menu of some cameras there is only one item that is responsible for setting RAW ...” and the article is called “setting RAV for cameras” =) thanks for your work, I ordered a camera via the Internet, I'm waiting, I have already read almost all your articles, I learned a lot ... The camera will come, I will start my experiments not from 0-la

  • Eugene

    To the author: the difference between 12 and 14 is not what you see on the monitor, but how they stretch in the converter, 14 is much more flexible, more likely to sculpt, unlike 12 bits, and of course, you can’t see anything on correctly exposed frames )))

    • Arkady Shapoval

      Actually, in the article the whole bias is made on pulling.

  • Alex V

    After the converter, a JPEG is obtained, which is 8-bit, i.e. for the monitor, this is 8 bits per color. The presence of 12 bits in equal amounts allows you to extend the exposure by -2 ... +2 stops without losing quality. 14 bits allows pulling -4 ... + 4 stops. This is the whole difference, you will not see the difference on the monitor - it will be almost the same in 8 resulting bits, but it is much more difficult to pull out the lack of light in 4 stops at 12 bits than at 14.

    Of course, there are still scenes with wide DD and tonal transitions, where with a higher bit depth we have more opportunities to compress some ranges and expand others - but this is for pro and gourmets.

  • Andrei

    As I understand it, with 14-bit files, you can “tinker” more without losing much in quality, then the photos of the recently released Nikon d 7100 with 42 bits can be made unimaginable at all?

    • Andrey Dolzhenko

      Can you link to a source about the 42-bit 7100 range? Nikon engineers don't know ... XD

      • Arkady Shapoval

        No one in the know :)

  • Levani

    Arkady, please enlighten, as in the ViewNX editor, when working with raw, draw details in the shadows to get a larger DD. I never worked with raw, not literate at all :)

    • Arkady Shapoval

      There is shadowp rotection, highligth protection, D-LIGHT, play with these sliders.

      • Levani


      • Andrei

        oh, these sliders are sad there. Almost nothing is pulled. (compared to LR)

  • Alexey

    And I generally think so far the most reliable and best camera is a device with a film!

  • Alexey

    Full manual mode, red eye, blurring ... that's what everyone needs in the house!

  • anonym

    Thank you!
    -how simple and lucidly they pushed for 12 and 14 bits
    To the owner of the D3100.
    - it turns out that I need 14Bits, and look after another Camera.

  • Andrey Dolzhenko

    Well, you learned it with your test. The main difference between 12 and 14 bits is the color space coverage - i.e. dynamic range. Lossless RAW compression is done by cutting off some of the tones in highlights, which also affects the dynamic range. Anyone who is fond of photography, and especially landscape photography, is well aware that when the ISO is increased, the DD of the matrix is ​​automatically narrowed. The higher the ISO, the narrower the DD. (proof - In your test, you significantly narrowed the DD from the very beginning, by increasing the ISO. Next, you try to see the difference (attention !!!) in the dynamic range of the images. But you have already leveled this difference using ISO 3200. This, sorry, is the same as testing sheets of paper for whiteness, having previously painted them with one gray paint. Take it hard - shoot a landscape with difficult lighting at minimum ISO settings - and you'll see the difference between 12 and 14 bits, between compressed and uncompressed RAW. All the best. :)

    • Arkady Shapoval

      It was an amateur household test. In order not to introduce people into any kind of misconception, I removed the extra information on your recommendation.

    • Anatoly

      Ta-ah-ah-ah ... However, I (and all sensible photo proletarians) do not understand the meaning of your frighteningly prickly remark, in other words, but could it be possible to explain all of the above in Russian? We are waiting not for naked theories, not rantings, but business practical advice.

  • Oleg

    Hello Arkady! When opening a RAW file in the RAW camera, the inscription below:
    Adobe RGB (1998); 16 bit; 3872 at 2592; (10,0MP); 240ppi
    Does this mean that RAW is 16 bits?

  • Basil

    Your D7000 is also in an amateur class, you do not need to raise it to a class because of some new functions and improved logic. Yes, and a magnesium bar with the declared moisture protection instead of polycarbonate as on D70, D70s, D80, D90 cameras also does not make the camera professional (although the front panel in it is still made of polycarbonate, although it could make the case more molded.)
    And so this is the same camera of the upper Mid-range range, as the models I listed above. Just added features that many fans have been waiting for. Besides mine, for example, the D80, in addition to 1920X1080, can record video with a stereo, large screen, CMOS sensor, some functions such as a virtual horizon, a large number of points, the ability to work with ancient manuals without dandelion, 85% of the magnesium blank is globally different. And outwardly, when directly comparing (comparing it many times, putting it side by side) almost 1 in 1. From the newer same D90 it is even less than different.
    Although the same newest D7100 also differs little from the D7000, I just held both cameras in my hands and compared them. But I liked the fact that they finally placed a full-fledged two-channel microphone - video lovers rejoice)).

    • Bogdan

      I am not yours. Only cameras for D90 products cannot be taken for granted. Of course, you’ll be amazed at the functions, keep your eyes on the 4.5-second serie for D90 and 6-second for the second one is not big, but + the shutter speed is 1/4000 and 1/8000 can be closed in the studio, the building is located at Yashchsho say that for other people for whom they clung to the chamber of another class. I am not a є proclamator of the camera ’ale dіisno the camera turned out to be a success. First, I thought it’s more important to change the matrices, if the Nikon D7000 has 3200 noise and I remember 6400 noise. I need to remember the camera D300 I want and the old D300 , D80.

      • Volodimir

        To drive noise: the technical moment of the CCD and CMOS matrix. On some of the lesser noises - on the other colors he has accepted. To my skin)))

  • Volodimir

    Convector ViewNX2 with RAW processing conversion to jeep. If I put the size on the long side
    800-1600 image is not high quality. If more than 3000, then excellent. So then compress.
    I am new to RAW. Correctly, it’s a pity that excellent pictures are obtained, and they are advised on recommendations
    long-side conversion 800-1000.

    • Volodimir

      Complain Lightrum and I will be happy for you)))

  • Bogdan

    I liked Andrey Dolzhenko's comment. Nikon's RAW compression is kind of weird. If it is compression, then the compressed files must also have the same weight relative to the source. In Nikon, the weight of a compressed file is lossless, fluctuates around 3MB, which indicates not so much about compression, but rather about throwing out “unnecessary” information by the camera's processor. Nikon D200, D80, D3000 have the same matrices. But here's the question. Why is the compressed equal without losses with the same matrices different for D80, D3000? D80 - up to 12MB, D3000 - up to 10MB. By the way, the uncompressed RAW weight of the D200 also fluctuates between 14,9-16MB.

    • R'RёS,R ° F "RёR№

      wrong conclusion.
      There are data types that are difficult to compress.
      Especially it shakes lossless compression.
      Losless compression is highly dependent on the picture, filling in the bits of the cells,
      as well as similar sound compression depends on the amplitude of the signal.
      Therefore, lossless compression cannot greatly compress the picture and the size floats
      (Well, the enclosed jpg of different sizes comes out).
      Regarding conventional compression: it also first uses mathematical archiving algorithms and only then begins to remove data at the edges of the range, but it has its own limit so that it is not very noticeable to the eye. Therefore, even with the usual compression of miracles, there will be no space saving on the map.

  • Igor

    And no one paid attention to a simple arithmetic inconsistency. Nikon D700, 12.1 million pixels. With an 8-bit ADC, we get an uncompressed RAW size of 12.1 megabytes. Increasing the ADC capacity by one bit doubles the file size, that is, at 9 bits we get 24.2, at 10 48.4, at 11 96.8, at 12 183.6. Nothing personal, simple math. The camera makers are bullshit. All their bit depth is interpolated.
    I decided to do the calculations after I read what size files Leika S. produces

    • Andrei

      > Nikon D700, 12.1 million pixels. With an 8-bit ADC, we get an uncompressed RAW size of 12.1 megabytes.

      8 bits per color, respectively, is not 12.1, but 36.3 megabytes.

      > Increasing the ADC capacity by one bit doubles the file size, that is, at 9 bits we get 24.2, at 10 48.4, at 11 96.8, at 12 183.6.

      It does not double the file size, but the dynamic range. With 12 bits, there were 4096 levels of each color, with 14 (two more bits), four times that, 16384 levels. And the file size grows just for this very bit - with 12 bits per color, a 12.1 megapixel matrix will give 12 * 3 * 12.1 = 435.6 megabits, or 54.5 megabytes, and with 14 bits per color - only 14 * 3 * 12.1 / 8 = 63.525 megabyte.

      By the way, some cameras at 14 bits per color write extra zero bits to the file in order to align the space for one color of one pixel to a round value - 16 bits. It is believed that this simplifies and speeds up processing, although in my opinion they could solve this problem in a less dirty way.

    • Maximum

      There is no arithmetic "inconsistency".
      About matrix megapixels:
      Most modern cameras use a Bayer filter (more on Wikipedia). Thus, out of 16 MP in the matrix there are 8 MP “green”, 4 MP “red” and 4 MP “blue”. To obtain the color image we are used to, the demosaic algorithm is used.
      About the size of the RAW file:
      From the matrix, the "brightness" value of each pixel is obtained. If we have a 16,2 MP matrix, then the data size will be 16,2 * 16 ^ 6 * 14 = 226,8 * 10 ^ 6 bits or (226,8 * 10 ^ 6) / 8 = 28,35 * 10 ^ 6 bytes or (28,35 * 10 ^ 6) / 2 ^ 20 ~ 27MB. Approximately the same amount of RAW file in Sony A35. If you apply lossless compression like ZIP, the RAR size will be reduced - as Canon and Sony always seem to do.
      Regarding Nikon:
      Nikon has NEF lossy compression - only ~ 9,2 bits per pixel are written to the NEF file. In fact, a tonal curve is superimposed and information is discarded from the "overexposure" area.
      In models up to 7 × 00, only NEF lossy compression is applied - Models D7x00 and older allow you to select NEF lossy, NEF lossless and without compression (the last 2 options store honestly 14 bits per pixel in the NEF file). There is a patch for the D3100 / D5100 firmware that allows you to use NEF Lossless - for experimenters I give the link __http: //

  • Roman

    Arkady, for the concept of “lossless compression” characteristics like “almost lossless” are not applicable, at least if we draw an analogy with ZIP, RAR and other algorithms for compressing files on a computer. After unpacking, we have exactly 100 percent match to the original. Of the minuses add. resources and time to unpack.
    The same applies to various lossless audio file compression formats (such as APE)

  • Anatoly Snezhan

    Arkady, reading the comments on your review articles (hundreds, thousands of questions of the same type!) Came to the following conclusion: you need to supplement all your reviews of more complex cameras such as D3200, D5100, D5200, D7000 with full versions (options) of their initial settings. On points, screenshots ... You have it smoothly, like in school. So that after their execution, you can immediately take up effective, efficient photography - to get sharp color pictures. I think you can do that! By doing this, you will extend the service life of thousands of brand new cameras, save them from the primary (pre-purchase, purchased and post-store shamelessly ineffectual barbaric use, tormenting expensive equipment.
    And how many nerves of the unfortunate owners of the long-awaited Nikonov will you save from destruction!

    • Arkady Shapoval

      There are simply no settings for all occasions. Let people think a little head :)

      • Anatoly

        I agree, but ... The fact is that the circle of Radozhiva's readers is already very, very ... that ... has grown. I mean what you need, Arkady, as a channel repairman (there is such a massive profession in land reclamation) not only to maintain the channel channel in a normative manner, but also to divide and sort everything that is floating ... just like you, and beginner photographers, with little experience, and amateur photographers, again, both experienced and just formed, and for the first time, who use a camera in their hands ... There are many of us already, all different. “Let people think lightly with their heads” is an appropriate phrase. But for mature people, already thinking ... These will think out, I agree. And the rest (there are more of them!) - who are just studying ... Their thinkers still need to be prepared, awakened to thinking. Is not it?
        I often return to old comments months later. To those who THEN, immediately, demanded reflection ...

    • Lynx

      If a person does not want to spend time studying his apparatus, its capabilities, settings, testing, understanding the mechanisms of work and other things, there can be no question of any “immediately creations of shydevras” (that is, “effective photography)”.
      We drag people who do not want to learn and delve into the essence - all these “superb settings” will not help. Yes and no them, and those that are - and so are protected in the camera menu "by default".

      • Anatoly

        Nda-ah ... really ... It seems to be like that, but not all at once, immediately to the touch and will grab onto the right Iichko.

  • sergey

    I recently watched showexiv the number of photos taken and accidentally noticed such a line Bits Per Pixel: 12
    I have Nikon D5100 how to make it shoot 14 bits and not 12, thank you in advance for the answer

    • Anatoly

      Sergey, I answer on the occasion, in a year +
      Even our Arkady will not catch the difference between 12 and 14 bits with an experienced eye. Ask him yourself! And one more thing. You can't make amateur cameras “love” like a pro. Hehe.

  • Bogdan

    Above they wrote about 42 bits in the Nikon 7100, it is clear that 42 bits are not. Ale pitanya camera yaka maє 42 bit yak yak і mzhlivostі vіdkriyutsya її RAW files? Mene zaq_caviv Fujifilm FinePix S3 Pro in the first characteristics entered in the new glybin color of 42 bits! What is so plus plus?

Add a comment

Copyright © Blog author - Photographer in Kiev Arkady Shapoval. 2009-2023

English-version of this article

Versión en español de este artículo