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

Bob Newmans Low Light Article Brilliant!

Discussion in 'AP Magazine Feedback & Suggestions' started by Blind Pugh, Jan 29, 2012.

  1. Blind Pugh

    Blind Pugh Well-Known Member

    What a great article of Bob Newmans on Low Light.

    I have been interested about exposing to the right , and have read many articles regarding this (for and against!). The crucial point regarding finding the sweet spot in the iso range is the missing bit of information that I have not seen mentioned in many of the forums, and magazine articles regarding exposing to the right.
    It makes so much sense to calibrate the camera first so the maximum dynamic range can be had from low light images when using Raw , and even Jpegs .

    Brilliant !!



  2. mark_jacobs

    mark_jacobs Retired

    Thank you for your feedback Ian.
  3. Blind Pugh

    Blind Pugh Well-Known Member

    You are most welcome.

    I have tried it on my Lumix FZ38 , and it even works on that!.

    (By the way the sweet spot for that camera is 800 iso, for anyone who wants to try it.)


  4. swanseadave

    swanseadave Well-Known Member

    What a thoroughly absorbing article it was.I`ll add it to my scrap book of other usefuls removed from AP like the raw series.

    I`ve not tried it yet but I will.

    Full marks to all concerned.:cool:
  5. Richard Sibley

    Richard Sibley AP Deputy Editor

    Glad you all enjoyed the article, but would just like to point some of the credit in the direction of Tim, who actually wrote it.

    It was written under the guidance of Prof. Bob, who provided us with the scientific/electronic info and the outline of the test to conduct. It is so valuable to have Bob as part of the AP team to consult on these matters.

    Some of the analogies Bob provides are really useful in understanding how sensors and noise work.

    For example whilst Tim and myself were talking to Bob about the article he described photon-shot noise as being like putting cocoa on cappuccino through a stencil. The light photons don’t all reach the sensor at the same time. In this example cocoa represents the photons.

    If you give the cocoa just a quick shake through the stencil, the stencil shape on the cappuccino froth will be speckled - noisy. Shake it a few more times and the solid outline of the stencil is visible and there is no ‘cocoa noise’.

    The simple analogy explains just one of the reasons why giving an image more light will produce less noise.

    We hope to do more features like this in the future.
  6. Learning

    Learning Ethelred the Ill-Named

    I enjoyed reading the article and some sections more than twice. It is good to provide something that demands some thought on the part of the reader.
    That analogy was worthy of inclusion in the article.
  7. Terrywoodenpic

    Terrywoodenpic Well-Known Member

    It would all seem to suggest, that a camera that can collect and convert near 100% of the photons that falls on the sensor in shaded moonlight will be the ultimate.
    However there are two ends to that scale... the true ultimate sensor could also count all the photons arriving from a magnesium oxide target lit by full sunlight.

    Long before we can achieve these aims, we will need to have sensors that are self limiting in much the way a graduated filter can be. An automatic, electronic, variable geometry and density, graduated filter, comes to mind.
    We could than be virtualy free from ISO, shutterspeed or aperture limitations, except perhaps in the most extreme conditions.
    What a fantastic fantasy.
  8. beejaybee

    beejaybee Marvin

    We're there already ... sensors with quantum efficiencies in excess of 80% have been around for a couple of decades.

    No, just a deeper "well" so that clipping is avoided. Linear response is the crowning glory of digital; in fact good linearity is what makes it possible to manipulate images at all. If you want to simulate a filter, that's best done in post-processing software ...
  9. Terrywoodenpic

    Terrywoodenpic Well-Known Member

    It is that "well" as you call it that I do not see straddeling that brightness any time soon. nor do I suppose sensors with the full range of other qualities are available even for pro cameras with a QE of over 80% Though probably are for astro use.
  10. Bob Newman

    Bob Newman Photo-Science Consultant

    Yes, as Richard said, it was Tim's article - I advised and supported. It was a very interesting experience. Like Most photographers, Tim has been trained essentially around a film based idea of how to manage exposure. Those basic ideas are so deep rooted it takes a lot of mental agility to work back from first principles, but Tim did it. He did an excellent job.

  11. beejaybee

    beejaybee Marvin

    And for those who still don't "get it": "The Handbook of Astronomical Image Processing", section 2, then graduate to section 6 if you really want to get into the theory of digital imaging.
  12. AlexMonro

    AlexMonro Old Grand Part Deux

    Bob, may I take this opportunity to thank you for all your excellent articles, some of the best parts of AP for me.
  13. Blind Pugh

    Blind Pugh Well-Known Member

    It makes me ponder on why manufacturers don't appear to give us some insight and information regarding getting the best from their sensors and cameras.
    Simply providing a sweet spot / expose to the right RAW shooting option in the camera setup menu with accompanying accurate histogram with information relating to that mode for exposing for maximum signal to noise and dynamic range , surely would enable us to get the best from each individual camera because the manufacturer has calibrated the camera for us?


  14. Bob Newman

    Bob Newman Photo-Science Consultant

    Thank you - it's good to be appreciated. Writing for AP has been great fun, and doing that article with Tim very interesting. I'm glad people liked the end result.

  15. Bob Newman

    Bob Newman Photo-Science Consultant

    The film exposure model is pretty deeply ingrained, with camera manufacturers as well as photographers. Until people begin to demand it, they likely won't provide it. Still, it's interesting to think about how the camera UI could be developed.

  16. P_Stoddart

    P_Stoddart Well-Known Member

    I don't know if what Prof Newman is saying is similar to say push and pull work in processing. I used to shoot film doing say 400ISO then push to get higher settings. Some photographer would shot say 200ISO and even pull if the condition got too bright. But push if light fell. Of cause you could only do it for the whole film. :)

    There was even film designed to be push from 100 - 1000ISO from Fuji (MS 100/1000). They gave you chart with the film as a guide if you was doing your own and to tell the lab. You can still look up the chart online.
    Last edited: Feb 11, 2012
  17. beejaybee

    beejaybee Marvin

    That's completely different: the sensitivity of a film can be affected by the development but the sensitivity of a sensor is fixed .... a basic piece of physics which is terminally obfuscated by the provision of a "sensitivity setting" on the camera.

    ISO speed settings are much like "digital zoom" - the difference is that most people recognise that simply scaling an image up doesn't change the resolved detail in the image but few recognise that twiddling the ISO setting is much the same.
  18. P_Stoddart

    P_Stoddart Well-Known Member

    I see, I was not sure because I remember that when you pushed film as well the 'grain' or 'film noise' would also increase. But of course they are completely different technologies. :eek:
  19. Learning

    Learning Ethelred the Ill-Named

    Yes the sensitivity of the sensor is fixed as is the maximum number of electrons that can be liberated in each pixel before it is saturated. Have I got this right? Ideally we need enough bits in the A/D conversion to cope with a number of electrons at the pixel when that number is at the noise level and also when it is full. In the case of the D7000 sensor and 14 bit conversion that condition is satisfied. Often this condition is not satisfied and amplification is applied so that a sensible use of the available levels can be obtained. The sensitivity setting is also used to help the 8 bit JPEG processor adjust to the relevant part of the characteristic. I think that I understand why that characteristic is linear in the digital case and have a gut feeling as to why it isn't in the chemical case. Don't remind me that we should think with our brain rather than our guts; I know.
    I have to disagree with the first sentence of the quote. The sensitivity of the film is also fixed. Pushing or pulling the development of film is just like fiddling with the amplifiers of the digital sensor. There was an ideal development time and temperature to extract as much detail as possible from the linear part of the film sensitivity curve.
    I am not stating any of this as fact. I am trying to get my head round it. Hopefully someone who understands the issue better than me will put me right or say I am on the right track.
  20. Terrywoodenpic

    Terrywoodenpic Well-Known Member

    A question or two here...
    A pixel point in a sensor effectively counts the Photons that strike it. during exposure.
    So... Is clipping because that pixel is saturated ( full) ?
    Or,.... because the Counter is limited by the maths of the converter scale. ?

    Eg. does a 16 bit converter clip at a higher point than a 14 bit. on the same sensor ?

    At the other end of the scale A pixel recieves so few photons that the noise of the circuit acts much like fog level in a film? and detail becomes lost or spoilt by random background signal.

Share This Page