1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

How can I tell whether RAW files have blown highlights?

Discussion in 'Digital Image Editing & Printing' started by ChrisNewman, Sep 25, 2017.

  1. ChrisNewman

    ChrisNewman Well-Known Member

    Hi,

    I usually bracket my exposures and record both RAW and JPEG files, mainly with a Nikon D800. A precept of the technique of “Exposing to the right” is that, in the absence of other factors such as varying sharpness, from my bracketed set, I should be able to develop the best image from the brightest RAW file that doesn’t have significant highlights blown out by over-exposure. So I would like to save the files for that shot, but I might as well delete any brighter ones. But how can I tell whether highlights are blown in a RAW file?

    I copy the files onto my computer with Nikon’s ViewNX-i. Pressing the h key will show blown highlights, but these are on a scale of 0-255, and I understand they refer to the JPEG file. (I understand the same applies to displaying blown highlights on the camera’s monitor.) I read that there is considerably more headroom for exposure in RAW files than JPEGs, so presumably, just because ViewNX-i is showing blown highlights for the JPEG doesn’t mean that I couldn’t develop a properly-exposed JPEG from the RAW file by reducing the exposure of the brightest areas.

    Nikon’s software includes Capture NX-D for developing RAW files, and <Shift>+h is claimed to show lost highlights. However, this is again on a scale of 0-255, and the histogram changes if I alter the colour temperature, so this seems again to be referring to the JPEG that I might develop from the RAW file with the current settings, not to the values in the RAW file itself.

    How can I tell whether the highlights in a RAW file are blown, or only too bright for the current JPEG processing settings.


    With thanks in advance,
    Chris
     
  2. Roger Hicks

    Roger Hicks Well-Known Member

    Dear Chris,

    It helps if you think of RAW files as transparencies. Blown highlight are forever blown -- there's just no information to be recovered -- but you can dig a LOT out of the shadows.

    Film negs are the exact opposite. It's very hard indeed to overexpose so badly that you can't dig something out of the highlights, but very easy to record insufficient detail in the shadows.

    In technical terms, digi and slide exposures are keyed to the highlights (don't over-expose or you'll lose the highlights) while neg exposures are keyed to the shadows (don't under-expose or you'll lose the shadows).

    Cheers,

    R.
     
  3. ChrisNewman

    ChrisNewman Well-Known Member

    Thanks Roger, I’m aware of all of the above (although I didn’t appreciate the different approaches appropriate for transparency or negative film when I was using that medium). But using exposure bracketing, I get 3 differently-exposed versions with each shot, and I want to know which of these to keep, and which to delete. The brighter the exposure, the more detail will be available to be dug from the shadows. It’s easy to display which versions have blown highlights in the JPEGs, but I read that there is considerably more headroom for exposure in RAW files than JPEGs. So I’m looking for a quick way to find out which RAW files have blown highlights, and which retain all their detail for potential future editing.


    Chris
     
  4. Roger Hicks

    Roger Hicks Well-Known Member

    Dear Chris,

    How do you tell? Well, just look. Or, most editing programs (and all cameras that I have tried) can show a histogram for part of the image. That will show you if the part in question is "blown". Sorry if I'm misunderstanding but this works for me.I'd rather lose (a little, often theoretical) shadow detail than highlight detail.

    Cheers,

    R.
     
  5. spinno

    spinno Well-Known Member

    In simple terms (for I am a simple man) when I open the raw file if the image is overdone at the right hand edge of the histogram the graph goes up in a straight line. As I pull back the whites and highlights (slide left) the straight line usually falls and the warning of blown highlights diminishes or disappears, for under the requirement is to move the shadows/black sliders to the right.
    I am using Adobe CC photoshop
     
  6. ChrisNewman

    ChrisNewman Well-Known Member

    Thanks again Roger. The only histograms I can find seem to show the exposure levels in the JPEG, not the underlying RAW file. As I read that there is considerably more headroom for exposure in RAW files than JPEGs, it seems likely that even if a JPEG shows blown highlights, the RAW file from which the camera developed that JPEG may well have recorded the full exposure range without blowing the highlights. I’m hoping to find a way to check whether or not the RAW file has blown highlights without going through the process of developing another JPEG on the computer and juggling with the settings to see if I can produce an image with all highlight details preserved.

    Chris
     
  7. spinno

    spinno Well-Known Member

    What software do you use to process your images? Does it not show a histogram when you open a raw image?
     
  8. ChrisNewman

    ChrisNewman Well-Known Member

    Recently I’ve been using Nikon’s Capture NX-D. As I explained briefly in my opening post, this does claim to show lost highlights; it shows a histogram, and <Shift>+h blacks out the image, except for showing blown highlights in colour, which I find clearer than the histogram. But the histogram is on a scale of 0-255, and the display changes if I alter the colour temperature or level of exposure, so this seems again to be referring to the JPEG that I might develop from the RAW file with the current settings, not to the values in the RAW file itself.

    Chris
     
  9. spinno

    spinno Well-Known Member

    Once you start to change settings can you see the change in your image and in the histogram. If you can you have effectively created your own jpeg (if you were to save immediately with no more processing). The in camera jpeg will remain as the camera created it. I suggest that you play around with a raw file that you have using the various settings and look at the affect that they have on your raw file. Remember the raw file can be worked on as many times as you like without degradation.
     
  10. ChrisNewman

    ChrisNewman Well-Known Member

    Thanks, but what I want is to be able to work quickly through each batch of photos I take, identify which ones from each bracketed set have blown highlights (or other faults) and so should be deleted, and which ones should be kept, give the ones to keep an explanatory filename, and then save them for possible further development in future. If I only shot JPEGs I could do this very quickly in ViewNX-i by pressing the h key for the ‘Show Lost Highlights’ function. But as RAW files are said to have more headroom for exposure, I don’t want to delete the RAW file of a version of a photo just because the JPEG has areas of overexposure. So I’m hoping someone can tell me of a similarly quick and easy function that will show whether or not a RAW file has areas of overexposure.

    With thanks,
    Chris
     
  11. PeteRob

    PeteRob Well-Known Member

    The raw file will have some recoverable detail in the highlights compared to an in-camera jpg as long as the highlights are not overexposed. i.e. the histogram level is less than 255. What you gain is information between, say 254 and 255, that is lost at lower resolution.

    The in-camera histogram and exposure indicators such as the "blinkies" are generally conservative (you can sometimes configure the level at which the blinkies trigger - typically they may turn on at 250) but that doesn't mean that you can wantonly over-expose a shot and hope to always recover highlight detail. When you set the exposure you basically set a white point. As Roger said if you set this high (so the picture is underexposed) you can recover but if you set it low and let too much light in for the sensor to cope then there is no recovery.
     
  12. PeteRob

    PeteRob Well-Known Member

    I don't know the Nikon tools but define a preset that reduces highlights to minimum. Import the raw files and apply the preset. Turn on highlight warning and scan the files. If they show warnings with maximum highlight recovery they are truly blown. Make your selection and delete. Then undo the preset.
     
  13. GlennH

    GlennH Well-Known Member

    You'll have more exposure latitude if you process the raw files in a ProPhoto RGB colour space - a choice that is not available with in-camera JPEGs or settings, at least in part because it must be edited in 16-bit mode. Colour profiles, saturation & exposure are all interrelated, as you may know. Though a raw file doesn't have a colour space, by the time it's gamma-corrected and rendered as a JPEG or TIFF in an editing program on a monitor, it does. That choice of colour space affects clipping in RGB channels and, consequently, exposure and the ability to recover highlight detail.
     
  14. Terrywoodenpic

    Terrywoodenpic Well-Known Member

    As a normal practice I would work to the Jpeg's histogram, and use the raw to supply a further safety net.
    This way the raw will always contain more recoverable highlight and shadow detail than the Jpeg.
    While it is possible to shoot even further to the right it is a "Risk" in terms of potential highlight losses.

    If the Highlights in the particular shot are of no importance to you, there is then no reason why you should not increase the exposure to get cleaner noise free shadows.
    But you will not be able to change your mind later. Clipped highlights are lost
     
    Roger Hicks likes this.
  15. Terrywoodenpic

    Terrywoodenpic Well-Known Member

    As does the choice of Film mode when using Fuji X cameras. each film type clips colours at different points. in much the same way colour balance can.
     
  16. El_Sid

    El_Sid Well-Known Member

    The issue here is that the RAW file is not an image file. A RAW file consists of the largely unprocessed data from the sensor expressed as non-specific luminance values along with all the relevant camera settings and picture data in use for that exposure. It does not become an image file until it has passed through an image converter according that converters pre-set values.

    In Nikon's own software the preset values appear to be based those set in the camera and the image you see looks the same as it would as a JPEG - I can never see any difference between the NEF and JPEG images in ViewNX at any rate. Third party converters default to their own settings - although they usually allow you to set your own custom profiles that can be linked directly to both camera and ISO, possibly you could create a profile that produces a neutral result possibly with reduced contrast and no brightness adjustment that may give you a more 'accurate' view of the highlights.

    From my own experience I find that RAW files showing only moderate clipping can often be pulled back successfully but if the right end of the histogram starts to rise above about 25% of the scale height then the picture is likely in terminal trouble. Some highlights, such as specular highlights from reflective objects, are often fundamentally unrecoverable unless the image was drastically underexposed to compensate.

    My own experience with expose to the right is that it has never really worked for me. I don't know whether it's the kind of images I take but when I try it I generally end up with hopelessly over exposed images that are difficult or impossible to recover...:(
     
  17. ChrisNewman

    ChrisNewman Well-Known Member

    I usually bracket ±⅔EV, look at the histogram or for "blinkies", and if I find no blinking on the darkest shot, but some on the brightest, I consider that exposure to be on target. (If not, and if it’s a particularly good subject, I’ll change the exposure compensation 1EV and shoot another set of 3.) But trying to see detailed information on the monitor is tedious, as I need to put on my reading glasses, and in bright conditions it’s hard to make out any detail anyway. I find it worryingly easy to miss a column at the extreme right of one of the histograms if the others slope down towards the right. But my current priority is trying to make optimal decisions on which shots to keep, and which to delete as over-exposed. I don’t think the "blinkies" on my D800 can be adjusted, but I assumed that the values you quote refer to JPEGs, that 254 is a genuine exposure level in the JPEG, whilst 255 means that the processor has assigned that pixel the brightest level available in the limited palette of the JPEG. But if there is more headroom in what the sensor has captured and has been recorded in the RAW file, a JPEG value of 255 won’t tell us whether the maximum recordable exposure of the RAW file has been exceeded.

    Thanks for your suggestion of making a preset that gives a low level of exposure. When I get a chance, I’ll investigate whether I can create presets in Nikon’s Capture NX-D, and if so, experiment with a sequence of shots of varying exposure and see what I get. But I had hoped to find a procedure that will check RAW file exposures as quickly and easily at the tools provided for JPEGs.

    Chris
     
  18. PeteRob

    PeteRob Well-Known Member

    I wouldn't get too worried about following the "expose to the right" logic. It essentially says overexpose then under-develop. For scenes lacking highlights but having dark areas that would have to be lifted if the scene were exposed for the mid-tones it can give less noise in the shadow areas. For "normal" scenes you risk blowing the highlights. You could try using a spot meter. This would reduce the empricism implicit in bracketing.
     
  19. ChrisNewman

    ChrisNewman Well-Known Member

    I’d be grateful if you could expand somewhat on the processing process. I appreciate that the 14-bit RAW files from my D800 can hold much more detail than an 8-bit JPEG. If there is more headroom for exposure in a RAW file than an out-of-camera JPEG, there will be more highlight detail recorded in the RAW file than is shown in the JPEG. But I had assumed that, with appropriate processing, it would be possible to create a JPEG that showed that highlight detail, although either the tonal range would need to be coarser, or shadow detail truncated, to compress the information into an 8-bit file. When you state “choice of colour space affects clipping in RGB channels and, consequently, exposure and the ability to recover highlight detail”, do you mean that it may not be possible to include all of the highlight detail of a RAW file in any JPEG?

    I’m also wary of the terminology. “Exposure latitude” and “headroom” suggest there is information in the RAW file that isn’t shown in the out-of-camera JPEG. However, “recovering highlight detail” suggests the exposure was too bright for all highlight detail to be recorded, but careful processing from a RAW file can make amends to a certain extent. A JPEG from the RAW file of a brightly exposed shot in which highlights have been “recovered” sounds as though it could be inferior to one from a less brightly exposed shot.

    But if RAW files which where exposed more brightly than the brightest out-of-camera JPEG can be developed to give better results, because of their greater exposure latitude, I still want to find a quick way of identifying which are the best files to keep, and which I may as well delete as over-exposed.


    With thanks,

    Chris
     
  20. ChrisNewman

    ChrisNewman Well-Known Member

    I rarely use spot metering; even with matrix metering I notice a tendency for the camera to overexpose if the subject I focus on is dark, and underexpose of it is bright. My question isn’t about how I should expose my photos when taking them. With bracketed exposures I’ll typically have a shot where the JPEG histogram looks just right, and another where the JPEG histogram shows some blown highlights. But I frequently read that the RAW file accompanying that JPEG will have more headroom, and so it is likely to be possible to develop it into a better image than the out-of-camera JPEG or the RAW file accompanying it. So I want to be able to identify, quickly, whether the exposure of a RAW file that accompanies an over-exposed out-of-camera JPEG has captured the full range of tones, with the potential for being developed into a better JPEG, or whether exposure has also been blown in the RAW file, so that I might as well delete it.

    Chris
     

Share This Page