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. PeteRob

    PeteRob Well-Known Member

    Indeed but I was looking at it as a way of assessing RAF files before going through the LR import process. When I assess a picture (to decide to delete it and not bother with processing) I don't much look at the histogram, I'm more interested in composition and focus.
  2. ChrisNewman

    ChrisNewman Well-Known Member

    Thanks Andrew, I did have a look at that page, but I’m afraid I was lost. I failed to find anything that explained which section referred to the output data from the sensor that would be processed to form the image (is it the “lossless jpeg grayscale picture”?), or what format that data was in, and in particular, how the output behaved if the photosite received more light than the sensor was intended to handle at that ISO sensitivity. But from PeteRob’s later link to RawDigger, I found their 3-part article on RawDigger Histograms. Part 3 (https://www.rawdigger.com/howtouse/rawdigger-histograms-overexposure-shapes) shows a histogram from a D800. The sensor does produce RAW values for each pixel up to almost the 16,383 limit for 14-bit digital numbers, but the clipping point for over-exposure is slightly less, and apparently clipping doesn’t necessarily occur at a precise number, in the way I had expected.

  3. ChrisNewman

    ChrisNewman Well-Known Member

    Sorry for the delayed response, but I’ve had a relatively busy week, and have had to do some studying in this area, so I haven’t been able to keep up with the helpful replies I’ve been given. You probably realize by now, but what I’ve been looking for is a straightforward way of finding out whether or not the highlight detail in the RAW file has been clipped. GlennH seems to have found that the FastRawViewer program does exactly that (and I see that you have now tried it), although it may not work for me at present.
    I don’t remember much comment about ISO settings when various experts have referred to RAW files having more headroom for recording highlights than is shown in out-of-camera JPEGs. I realize that retaining as much dynamic range as possible is even more valuable at high ISOs than low, because the sensor’s dynamic range is reduced. On the other hand, I assume the smaller the dynamic range, the less detail will be lost to the JPEG.

    My dilemma regarding whether to save or delete a RAW file when the corresponding JPEG has burned-out areas, and bracketing has given me a pair of files with a lower exposure, applies primarily to low ISOs. I realize that the sensor in my D800 is relatively ISO-invariant, so that above ISO 800 it loses approximately a stop of dynamic range for each extra ISO stop. The D800 also has an excellent Auto-ISO program, so I normally use this when shooting hand-held and bracketing. This means that in fairly low light, my bracketed set will have the same aperture and shutter settings for each shot, but different ISO sensitivities. So if, for example, the set covers ISO 1600, 1000 and 2500, if I brighten the ISO 1000 RAW file by 1⅓ EV, it should show almost identical shadow detail to the ISO 2500 file, but of course it will have retained far more highlight detail. So at higher ISOs I aim to keep exposure compensation more negative than I would in brighter conditions, and take particular care to check that I have an out-of-camera JPEG that doesn’t show burned-out highlights. At these ISO levels I’d have little hesitation in deleting the ISO 2500 and ISO 1600 pairs of files if their JPEGs didn’t look better than the ISO 1000 one.

  4. ChrisNewman

    ChrisNewman Well-Known Member

    Brilliant thanks Glenn! The display in FastRawViewer’s “Detect the Over/Underexposed Areas from RAW Data” feature appears to be just what I hoped might be available, although I had hoped it would be an element or plug-in for popular image editing programs, rather than a fairly detailed program in its own right. The catch for me is that it runs on 64-bit Windows 7 onwards. I’m still working primarily on a desktop with 32-bit Windows Vista (I do have a Windows 10 laptop, but the screen and keyboard on the desktop are better, and I find Windows up to 7 more user-friendly than Windows 10, and have an even stronger preference for Office 2003 over subsequent versions). I intend in future to get a new PC, a large high-quality monitor, and a better image editor (probably Lightroom; I’ve seen great praise for DxO, but unfortunately they don’t seem to cater for those of us who sometimes mount an APS-C lens on an FX Nikon body, although using full frame lenses on DX bodies is very well supported). But my main priority at present is sorting out which shots I want to keep, giving each of them a filename that identifies as clearly as possible what they show and where they were taken, and archiving them. I’ve allowed a large backlog to build up, and I want to clear it before I take on the distraction of choosing setting up and learning to use a new system, and the temptation of having a better editor to work with.
    My best option might be, once I have a new system, to try Lightroom’s Melissa RGB histogram, and then use the offered trial of FastRawViewer to see if it works significantly better for me. Meanwhile, the ‘Beware the Histogram’ RawDigger article that PeteRob referenced shows comparisons between a Canon 5D MkII’s JPEG histogram and the RAW file contents. The results seem to vary with the subject and image settings, from failing to show over-exposure when it has occurred (I find this surprising) to showing over-exposure about a full stop before it happens. I didn’t find any equivalent analysis for Nikon JPEGs. So I’ll probably continue with my guestimates, unless I’m drawn to try FastRawViewer on my laptop, transfer to making my current assessments on that, and hope I will still be able to install the other licensed copy of FastRawViewer on my new desktop when I eventually buy it (perhaps unlikely, as they’ll probably have moved onto a new version by then).

    With thanks, Chris
  5. GlennH

    GlennH Well-Known Member

    You'll see a similar difference between a camera's histogram and a Lightroom histogram, since the colour space used in the latter is large enough to contain most or all of the camera raw data. This difference will vary according to the colour contained within the image (i.e. "document colour space") and how it's subsequently treated. Boosting in-camera contrast/luminance/saturation might easily push the colour out of gamut and beyond the edge of the histogram, which affects textural detail in individual channels and ultimately exposure latitude. Shooting raw obviously allows a little more headroom in all of this, but you would have to "uglify" your camera preview to make it more representative of raw latitude. Personally, using the camera's ARGB histogram as a buffer seems to work well enough, though I'd surely use an in-camera raw histogram if it existed.
  6. ChrisNewman

    ChrisNewman Well-Known Member

    It seems I should check my ViewNX histograms carefully to see whether those for the RAW file show more headroom than those for the JPEG even without changing any of the exposure or other settings - I had just assumed their content would be the same. I certainly don’t want to adjust the camera’s Picture Control settings just to give more informative histograms - I take the resulting JPEGs as the main output, reserving the RAW files for those few photos where the JPEG has failed to produce a reasonable result, or I think it is an exceptional photo that warrants any improvement I can make.

    With thanks, Chris

Share This Page