RAW dynamic range vs out of camera JPG dynamic range

Discussion in 'Photography' started by PeteR, Aug 13, 2009.

  1. PeteR

    PeteR Guest

    One of the mysteries of out of camera JPG files is why if the full dynamic
    range of an image that is available in the camera data isn't automatically
    included when the image is processed to make a JPG file.
    Typically, the highlight and shadow detail is there as I can manually
    process a RAW file to recover either or both these by adjusting the
    exposure/recovery/fill light/curves etc. in Photoshop RAW into two image
    files then combining these in PS to recover the maximum dynamic range of the
    image. Then, if I save the final image as an eight bit JPG file the dynamic
    range I've achieved remains.
    So, as the data is there, why won't my camera's (Canon 40D, 300D, and
    Minolta 7i) do this for me in the first place?

    One friend I discussed this with says that this doesn't occur in Olympus
    cameras. He maintains that there is very little difference in the dynamic
    range between an Olympus RAW file and an out of camera JPG file (he owns
    Canon cameras), so is it just because the algorithms used to do the JPG
    processing in other brands aren't up to it or is there some other reason?

    Regards PeteR
    PeteR, Aug 13, 2009
    1. Advertisements

  2. PeteR

    PeteR Guest

    Sorry for the delay responding.
    Changing those settings is certainly an improvement. I've a bit more
    experimenting with them to do before I settle on this procedure.
    Regards PeteR
    PeteR, Aug 18, 2009
    1. Advertisements

  3. PeteR

    Paul Furman Guest

    I don't believe this is correct... depending on the camera and setting
    perhaps or it could be mostly a matter of semantics I've latched onto.
    For one thing, the raw data is not gamma (mid-brightness) adjusted for
    human vision. The raw data is linear and humans can only distinguish a
    narrower range of tones. The data has to be corrected on an exponential
    curve (wrong term?) or else the image would look nearly white. When this
    correction is made, lots of data is thrown out. Also, doesn't raw have
    more bits per pixel?

    I know my explanation is a bit sketchy & lame but I believe it touches
    on the key issues. Histograms are key in explaining this and I can't do
    that in text. In any case it's easy to test with any raw converter and
    image editor: convert a raw file to jpeg (or tiff) with a large exposure
    compensation difference, then try to correct them back to a middle
    point. They won't look the same, the dark one will have more noisy
    banded shadows and the bright one is likely to have blown highlights.

    I used to think a low contrast jpeg was always the way to go and you can
    always add contrast but if you look at the histogram for an extreme
    case, the data for a very low contrast image is all jammed into a narrow
    band in the middle. If you crank up the contrast (without blowing
    highlights or muddying shadows) that stretches the data across the
    histogram so there's more latitude and more range of tones available. A
    common example of this is a telephoto shot on a hazy day of distant

    Another consideration is even with jpegs, the brighter tones are
    represented by larger numbers so there is inherently a finer gradation
    of values that can be stored and later adjusted without seeing banding.
    This plays into the raw conversion/gamma correction in a big way but
    it's still true for jpegs. The shadows will already be crude and more
    banded in nature once you gamma correct from raw so there isn't much
    more that can be done if you try lightening shadows on a jpeg. This
    normally isn't a big problem because we don't see dark tones as well but
    if you make big corrections it gets ugly pretty quick.

    Paul Furman

    all google groups messages filtered due to spam
    Paul Furman, Aug 19, 2009
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.