ASA setting in digItal SLRs?

Discussion in 'UK Photography' started by A.Lee, Jan 25, 2014.

  1. A.Lee

    A.Lee Guest

    Reading an article today about a chap who took a pic with his DSLR, the
    pic was taken at 6400ASA, then pushed 2 or more steps in PS, so the
    outcome was something like 25000ASA.

    How does ths work on a digi?
    On film the emulsion would be different for higher ASA films, and the
    quality fell off as the grain size increased - presumably the
    emulsion needs bigger grain to absorb the smaller amount of light
    entering the camera.

    Is it that clumps of the sensor in the camera are joined together for
    the higher ASA ratings?
    Or am I missing something?

    A.Lee, Jan 25, 2014
    1. Advertisements

  2. A.Lee

    NY Guest

    My experience with winding up the ASA on a digital camera is that the
    picture has the same level of spatial detail but each pixel is subject to
    random "noise" (coloured speckle).

    Here are two photos (part of a magazine cover) taken with a Nikon SLR at 200
    and 3200 ASA (the opposite extremes of my camera's ASA range): (200 ASA) (3200 ASA)

    Allowing for the 200 ASA one being slightly blurred because of camera shake
    (shutter speed was only 1/10 sec!), the two pictures are pretty identical in
    sharpness and detail, but the 3200 one has a lot of random coloured speckle
    over it, which is most noticeable in the plain white and blue areas.

    With film you'd have the equivalent of a lower-resolution picture since the
    size of the grains would be larger for the faster film.
    NY, Jan 25, 2014
    1. Advertisements

  3. A.Lee

    Woody Guest

    Think of ASA in the old water circuit principle. You have a
    bucket the size of which is determined by the ASA. The
    aperture used is the size of the hose used to fill it. Big
    hose needs means short duration, small hose mean long
    duration. The lower the ASA the bigger the bucket.

    In other words a film of given speed requires a certain
    quantity of light to expose correctly, and the same applies
    for a cell and the software behind it in the camera. Thus if
    you make the quantity of light needed to make the cell work
    very small (i.e. high ASA) there will be less detail from
    the scene but the thermal noise generated within the cell is
    essentially constant and the software will be unable to
    remove it, i.e. you get a speckled image. Conversely use a
    low ASA and the quantity of light available is much greater,
    the scene contains very much more detail and the thermal
    noise level is insignificant by compariosn.

    I hope that makes sense?
    Woody, Jan 25, 2014
  4. A.Lee

    Paul Giverin Guest

    I'm not sure I understand either. As I'm sure you know, pushing is the
    process of underexposing the film and then compensating by

    With the DSLR, you don't develop as such. The only thing I can think of
    regarding the example you gave was that the max ISO on the camera was
    limited to 6400 but by shooting in manual mode and in RAW, you could
    underexpose by a couple of stops and then increase the exposure by a
    couple of stops in PP, giving an effective ISO of 25000. The result
    would be quite noisy I'd imagine as most DSLR's favour exposure to the
    right i.e. overexposure.
    Paul Giverin, Jan 25, 2014
  5. but by shooting in manual mode and in RAW, you could underexpose by a
    couple of stops and then increase the exposure by a couple of stops in PP,
    giving an effective ISO of 25000. The result would be quite noisy I'd
    imagine as most DSLR's favour exposure to the right i.e. overexposure.

    Yes, this is what my thinking is too to achieve ISO 25K and, as you point
    out, it would not be a nice image to work with. I'm going to have an
    experiment with this tomorrow, I'm intrigued to see how "un-nice"!

    I essentially *never* shoot above ISO400 and even that's rare! ISO100
    almost all the time, I simply cannot abide digital noise/grain[1], it's
    ugly in a way that real film grain isn't. In fact, I find real film grain
    rather beautiful!

    [1] if I want a "grainy" look, to images that end up B&W usually, then I
    add noise in post. Controllable, and usually a much better look than
    natural digital noise!

    Stephen Thomas Cole.
    UK Usenet Head of Social Media and PR.

    This post was sent from my iPhone, likely whilst walking, so may have typos
    or bizarre auto-corrects.
    Stephen Thomas Cole, Jan 26, 2014
  6. A.Lee

    Paul Giverin Guest

    In message
    Sometimes you just have to shoot at high ISO, either at night or when
    indoors and no flash is allowed. The most recent medium to high end
    DSLR's just take ISO1600 in their stride. After following a "high ISO"
    thread on a Canon forum some time ago, I decided to experiment with
    ISO12800 which is the highest my camera (EOS 7D) will shoot at. I used
    ETTR and then used noise reduction software in post processing. I
    thought the results were surprisingly good.

    This is the original ISO12800 shot:-

    This is a 100% crop of the original:-

    This is after noise reduction:-

    And 100% crop after noise reduction:-
    That's why I still shoot b&w film.
    Having said that, I loaded a scan of this grainy film shot onto my
    flickr page and someone commented that they liked the "noise". You just
    can't win :-
    Paul Giverin, Jan 26, 2014
  7. It amplifies the output of the sensor. Higher ISO setting = higher

    It doesn’t join together the individual pixel sensors as such, though
    since any subsequent noise reduction necessarily involves combining data
    from nearby pixels in some way, some loss of detail is to be expected
    (analogous to using larger film grain).
    Richard Kettlewell, Jan 26, 2014
  8. Oh, I don't disagree, circumstances sometimes dictate. But I do have an
    almost pathological aversion to shooting above ISO 400 and will bend over
    backwards to get a result without resorting to increasing it. Luckily, my
    commercial work rarely calls for me to shoot in those conditions.
    I use a Canon 1Ds, so a few years old now and, being the "studio" version,
    it's built to work best at low ISO, which it does really well. But even a
    current "pro" Nikon I used recently (forget the model now, I'm a Canon
    man!) felt like it had similar limitations above 400, the quality of the
    image produce was markedly poorer and "fell off a cliff" in the same way as
    I perceive my Canon to. I learned to shoot digital with Hasselblad cameras
    (university's, not mine. I'm not yet able to afford to invest £100K in a
    camera and 4 lenses...) and they made incredible images at ISO 50 and 100
    but were almost unusable above 200, so this way of thinking seems to have
    stuck in my head!
    I've yet to be convinced with noise reduction in post. I've tried various
    bits of software, the most recent being Noise Ninja, and have never been
    happy with the result. It always feels like too much of a compromise and I
    get, I feel, better results from noisy images by being a bit fussy in
    Camera RAW and then spending a bit of time retouching it in Photoshop,
    mostly with lots of curves layer masks.
    I've got that lens! Nice bit of kit, real good "all rounder".
    Yeah, I see the improvement, certainly, but it has that almost artificial
    quality to it that I find comes from noise reduction software. I can't
    really put my finger on it even, but I just don't like the way it treats
    the pixels.
    Yeah, that does look real nice, a good demonstration of the organic wonder
    of film! Black and white digital images with fine grain often look better
    and more natural and "filmic" (is that a word?) for having noise added in
    Photoshop. I get really nice results from adding it in several layers and
    then masking each slightly different noise layer (both in terms of amount
    of noise and layer opacity) to different areas, as added noise that looks
    nice in the shadows might not in the highlights! Well worth experimenting

    Stephen Thomas Cole.
    UK Usenet Head of Social Media and PR.

    This post was sent from my iPhone, likely whilst walking, so may have typos
    or bizarre auto-corrects.
    Stephen Thomas Cole, Jan 26, 2014
  9. A.Lee

    NY Guest

    Is all of that film noise? Would a scan of a photographic print look as
    grainy? I've found that B&W negatives that are a bit grainy to begin with
    look a lot more grainy if scanned from negative than if printed and scanned
    from print. The two main B&W films I used to use were Ilford FP4 (125 ASA, I
    think) and HP5 (400 ASA), and HP5 shows a lot more grain on a neg scan than
    a print, whereas FP4 shows less difference. The Ilford B&W film that used
    colour chemistry, XP2, (you had to process it like colour neg film and the
    silver was removed leaving an image in dyes) had unbelievably fine grain for
    such a fast film.

    Those were the days: shutting myself away in the loft with my hands in a bag
    while I loaded the film in the spiral, then the making up of the developer
    and getting it to the right temperature (and keeping it there!), and
    inverting the tank every minute. The relief when the fixer goes in, that the
    film is now safe from getting accidentally fogged.

    Then back into the loft, freezing cold in winter or roasting hot in summer,
    breathing in the dust and glass-wool insulation, on my knees on the floor
    next to the enlarger.

    Everything is so much easier nowadays when it's all digital. Being able to
    make geometric parallelogram correction because I've had to shoot something
    from an angle to avoid getting my reflection or the flash reflection in
    shot - that's fantastic. Being able to correct contrast. And of course doing
    it all in colour - I never attempted developing and printing colour neg
    NY, Jan 26, 2014
  10. It depends if the print has the grain in it or not. If it does, a scan of
    it will too!
    Grain is naturally "softened" or diffused slightly just by the action of
    the print being made by light that's been projected through the film and
    had to travel a couple of feet, whether you've focussed it pin-sharp or
    not. A contact print of the same frame will be comparatively grainier (not
    that it's noticeable in a 35mm contact print! You can see the difference,
    though, in a 5x4 contact) than an enlarged print, as will a film-scanner
    produced file, largely because each of these methods records data directly
    from the film without it travelling anywhere.
    Agreed. Stunning film. C41 B&W is an aesthetic all of its own.
    I still run a "pro" equipped darkroom with a friend, we hire a studio at an
    art centre for it, not that I use it as much as I'd like, but it'll be a
    long time before I give it up! Ah, the fetid stench of stale fix in the
    carpet... Beautiful! :)

    We're actually in the process of commercialising it, there's a couple of
    relatively expensive darkrooms for hire in London and a small community
    darkroom in Folkestone but that seems to be about it for the South-East of
    England! If you're anywhere near Kent, maybe we should talk!

    Stephen Thomas Cole.
    UK Usenet Head of Social Media and PR.

    This post was sent from my iPhone, likely whilst walking, so may have typos
    or bizarre auto-corrects.
    Stephen Thomas Cole, Jan 26, 2014
  11. A.Lee

    Paul Giverin Guest

    Its APX100 developed in Rodinal, which does give quite large grain.
    I'm not sure. Thinking about it, we all sit fairly close to our computer
    monitors but stand back from a print on the wall. I haven't got round to
    wet printing the seagull shot yet but behind me on the wall I've got
    this shot:-
    which is Neopan 1600 pushed to 3200. The grain does look more pronounced
    on the scan compared to the 16"x12" wet print, which supports your
    Paul Giverin, Jan 26, 2014
  12. A.Lee

    NY Guest

    I've done some tests to show how averaging a number of identical photos can
    reduce the noise. It works on the principle that if the exposure, focus and
    framing of the photos is identical for all of them, the only thing that is
    different is the random noise - and averaging reduces the effect of that.

    I mounted the camera on a tripod, locked the exposure and focus, and then
    took eight photos at 3200 ASA and one at 200 ASA. I then combined them:
    1+2 -> 12, 3+4 -> 34 etc, then combine 12+34 -> 1234 etc, and then finally
    1234+5678 -> 12345678.jpg.

    Here are the results (look especially at the out-of-focus grey area at the
    top right): (200) (3200 average of 8) (3200)

    No doubt if I'd had the patience to average 16 or 32 frames I'd have reduced
    the noise even further.

    These are full-size crops from 4288x2848-pixel images.
    NY, Jan 27, 2014
  13. A.Lee

    Woody Guest

    I have a Sony HX5V compact which was advertised as having
    optical image stabilisation (as distinct from digital.) I
    was surprised to find that when used in the anti-vibrate
    mode what it actually does is take five or six frames
    rapidly in sequence (fast enough for shake not to be an
    issue) and combines them into one picture. Whilst it does
    have the disadvantage of seeming to use maximum aperture, I
    have yet to see any form of noise on any pics - and I've
    taken some in horribly dark places that came out like
    Woody, Jan 27, 2014
  14. Wow, that is actually quite impressive. Great technique! How are you
    physically combining these images, what's the workflow here? Obviously,
    it's a labour intensive process and is limited to tripod situations, but
    this could be something well worth bearing in mind for the right

    Stephen Thomas Cole.
    UK Usenet Head of Social Media and PR.

    This post was sent from my iPhone, likely whilst walking, so may have typos
    or bizarre auto-corrects.
    Stephen Thomas Cole, Jan 28, 2014
  15. A.Lee

    NY Guest

    I'm using Paint Shop Pro (I've got both Version 5 and Version X - either can
    be used for this).

    I open pictures 1 and 2 and then do Image | Arithmetic, with
    Function=Average. I save this as 12.png (I save using a lossless format to
    avoid accumulating JPEG compression artefacts).

    Likewise for all the other pairs of original photos.

    Now I load 12.png and 34.png and average those to 1234.png and so on for the
    other combined pictures. And then average 1234.png and 4568.png

    So for 8 photos I had to do 4+2+1 = 9 averaging operations to get the final

    In theory the same technique would work with film, averaging several
    exposures, though the difficulty would be getting all the photos to register
    exactly over the top of each other. I'm not sure how reproducibly a film
    scanner would scan four successive negatives on the same strip - always
    assuming they were on the same strip and not, by chance, on two separate
    strips when returned from being processed.
    NY, Jan 28, 2014
    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.