BTZS 0,5 ZS 0,57 Why?

Discussion in 'Exposure Discussion' started by AndreasT, Feb 11, 2013.

  1. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    Ok, here the issue because I was reading another thread and got very, very confused. When I look at the BTZS way of working an average scene of SBR 7 the avg. gradient is developed to a contrst of 0,5 and in most other publications I read to 0,57.
    Why?
    When pushing or pulling the values are very different as well.
    What am I missing?
     
  2. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    All the various forms of average gradient, including CI, are about determining the slope of the curve, Rise / Run. The value for the average gradient depends on a combination of three major variables: Log subject luminance range, paper LER, and flare. The average subject luminance range is 2.20 logs. Grade 2 paper printed using a diffusion enlarger has around a 1.05 LER. Average flare for a statistically average scene is considered to be 0.34 for large format and 0.40 for smaller formats. Flare reduces the apparent subject luminance range - 2.20 - 0.40 = 1.80.

    1.05 / 1.80 = 0.58

    This is what Kodak considers normal processing for a average scene.

    When it comes to compensating for longer or shorter luminance ranges there are two basic ways to deal with flare. One is using the same value of flare and the other is to use a variable flare model. The amount of flare does decrease with shorter luminance ranges and increases with longer luminance ranges, but as flare also can vary greatly within a given luminance range, there's a question whether the extra effort of the variable method is worth it.

    Here is a comparison of developmental models differing only in the application of flare.

    CI Chart of Developmental Models.jpg

    Almost all methodologies are going to give you a workable normal. Scene luminance ranges fall within a bell curve which means the statistically average scenes occur in the majority of situations. What distinguishes a method as effective is how successfully it applies to more extreme conditions. One of the difficulties with extreme conditions is in determining operator error from failed methodology.
     
    Last edited by a moderator: Feb 11, 2013
  3. flash26c

    flash26c Member

    Messages:
    122
    Joined:
    Mar 1, 2009
    Location:
    Oregon
    Shooter:
    Medium Format
    Great desertratt or should I call you Benjaman?
     
  4. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    Andreas, you might be misinterpreting Davis. Where did you see the 0.50 in BTZS?

    Frequently the ZS variables are 1.20 NDR and 2.10 subject luminance range which is 1.20 / 2.10 = 0.57. While this version of the ZS average gradient is practically identical to Kodak's (and the standard model), the justification for it doesn't reflect what is actually occurring.

    Here's a comparison of developmental models from Way Beyond Monochrome, the Zone System, no flare, and my Practical Flare model.

    CI Development Model Comparison.jpg
     
    Last edited by a moderator: Feb 11, 2013
  5. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    OK basically he say a normal scene has a contrast of 2,10 right. He uses a log of 1,05 for paper. His equation is G= DR/SBR, so G=1,05/2,10 which equals 0,5.
     
  6. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    I mentioned this to Fred Newman and he wrote me G=0,57 is too contrasty.
     
  7. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    Give me a page. Actually, this is one of the reasons why I find systems so problematic. The very nature of a system is to simplify the process in order to make it more manageable. Things will naturally tend to get lost in the process. On page 95 of the fourth edition, Davis discusses what he calls the effective DR. From page 95, "because you will want to produce negatives that print comfortably on your chosen paper, use the SI value you calculated in you paper test to locate IDmax (I hate his abbreviations), but do so only after making one further adjustment. This final adjustment is necessary because your curve data are based on a no-flare test condition, but you'll be applying the data to camera exposures that invariably involve significant amounts of flare."

    Davis then applies a compensation to the effective paper LER instead of simply adjusting the LSLR, which is what actually happens. If someone skims past these two paragraphs on page 95 that explains how flare changes the aims, then their results will be off. Why not just have the equation NDR / (LSLR - Flare)? It's impossible to miss flare as a variable when it's part of the equation.

    Did Fred say why? Don't accept unsupported statements.

    Here's a four quadrant reproduction curve with a LSLR of 2.20, flare of 2.0, and CI of 0.56. The NDR and LER match.

    Normal four quad.jpg
     
    Last edited by a moderator: Feb 11, 2013
  8. michael_r

    michael_r Subscriber

    Messages:
    6,568
    Joined:
    Feb 2, 2010
    Location:
    Montreal, Canada
    Shooter:
    Multi Format
    I really have to try the quadrants/"windmill" representation.
     
  9. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    Working with them and writing the program made the theory tangible. It really clarified things.
     
  10. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    OK, ok, I am going to work through this. Concerning abbreviations what do you mean by LSLR? LER is the paper contrast value?
    I hate all these terms, they are confusing since there are so many please bear with me.
    I am slow in grasping.
    (In short I grew up Bi-lingual (German/English) have been living in Germany for over 15 years, learnt Photography in German and learnt more complex photographic things in English, my brain needs a long time to process all this.)
    Have to run along and will return to this this evening.
    In short if I add the variable of flare will I get similar results to the other ZS, more or less.
    He mentions a small flare amount of 0,02 as a starting standard. What you (Stephen) say is much more.
     
  11. markbarendt

    markbarendt Subscriber

    Messages:
    7,589
    Joined:
    May 18, 2008
    Location:
    Beaverton, OR
    Shooter:
    Multi Format
    I've had that experience with various subjects, one of the most fun was programming a TI calculator to do spherical trigonometry for flight planning.

    I'm actually trying to write a concise piece on using light meters, even though I understand the use quite well, trying to explain it clearly to a general audience is a a challenge and helping me refine my understanding even more.
     
  12. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    LSLR and LER are the correct terms. Davis uses IDmax and SI. Scale is no longer used. Davis only kept these terms because he didn't want to make changes to the plotting program. There's even a disclaimer in the fourth edition. Some of the terms are dated and some he made up. It's frustrating when referencing BTZS with other material.

    Another problem with systems. If I remember correctly, Davis is referring to the density difference to the film curve as caused by flare. One, he applies the effects of flare to the film curve. This is an old approach which is more of a construct than reality. Flare affects the optical image and not the film curve. Two, his calculation is more of an approximation of the influence of flare. He doesn't use the flare equation. Or in other words a system.

    If you really want to understand sensitometry, you need to read something more academic. Photographic Materials and Processes would be an excellent start. Theory of the Photographic Process 3rd edition is a little more advanced.
     
    Last edited by a moderator: Feb 12, 2013
  13. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    From what I understand, this is what teachers go through. It's also one of the reasons why I participate in these forums.
     
  14. Sponsored Ad
  15. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    What programm?
     
  16. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    What bothers me about a lot of this is the explanation in many books and articles. They are sometimes difficult to follow. Ansel Adams I found labourous especially when trying it out for the first time. Personally for me the best method of explanation I have read is "Way Beyond Monochrom"
    Phil Davis I read after WBN and he jumped around too much with his explanations.
    I have been reading up a bit today of Phil Davis and I think some of the fog is starting to lift understanding his method. I have been using the "Way Beyond Monochrom" method for years and it worked, resulting in me reading BTZS less thorough.
    Anyway Stephen I will be reading up on some of yours posts as well, you cover this subjegt a bit.
     
  17. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    The four quadrant reproduction curve example from a few posts back is from a program I wrote.
     
  18. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    I envy people who can do that.
     
  19. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    Even though it is a construct won`t it work. At least in part. I see your Practical Flare Model is very close to the Way Beyond Monochrom numbers as well as The Zone System number.
    Flare always seems to just get a footnote.
    I have the BTZS PLotter programm by the by. Have been playing around with it today, with the flare part and I think I am getting the hang of it.
    However those flare numbers are difficult to swallow seeing how high they are.
     
  20. markbarendt

    markbarendt Subscriber

    Messages:
    7,589
    Joined:
    May 18, 2008
    Location:
    Beaverton, OR
    Shooter:
    Multi Format
    Interesting. So thinking out loud here... If I get the concept here, flare actually compresses the shadows on the original line of the curve. Even though we are moving into steeper parts of the curve flare is still compressing the tones.

    It seems to me then that essentially the construct you speak of is that the normal log exposure vs density curve is replaced by a scene zone placement vs density scale. Even if not perfect is that a reasonable understanding?

    If true that almost seems like a better way to visualize what is going to print. It becomes a labeling issue instead of a technical flaw. Seems to me that the log exposure vs density model hides flare's effect.
     
  21. Stephen Benskin

    Stephen Benskin Member

    Messages:
    1,646
    Joined:
    Jan 7, 2005
    Location:
    Los Angeles
    Shooter:
    4x5 Format
    Of course if works. That's not what is in question. It can lead to a misinterpretation of how the process works. Anyone viewing such a curve is required to be aware what it represents.

    The results are close, but the variables are different. Only one of the three models use variables that describe reality. Both the Zone System and WBM basically use what Davis calls the effective density range. Problem is, neither explains this is what they are doing. My Practical Flare Model is derived by averaging the values from the fixed flare model and variable flare model. While the variable flare model has the potential to be the most accurate. Matching the NDR to the paper LER is less an indication of quality as the luminance ranges progressively increase or decrease. I believe my approach tempers the extreme ranges and tends to fit the preferred NDR to LER relationships.

    That's the funny thing about facts, they don't care if you believe them or not. Seriously, before you judge, study up on the subject. What I believe you are referring to when you said 0.02 density increase is from the black box test described in BTZS. You appear to be confusing density increase from flare and the amount of flare. Also, I can't see the black box test as more of an exercise to create an impression of flare's effects and not an actual measurement. I did a similar black box test. Around the opening of the box, I place various targets, black, gray, checkerboard, and white. Below are the results. While there is a difference in density of 0.09 between the black target and the white target, the difference in exposure is 0.39 of 1 1/3 stops.

    Flare target comparisons.jpg

    On page 94 of BTZS, Davis recommends adjusting the SI by 10% for the effective DR in order to compensate for flare. For a NDR of 1.05, 1.05 * 1.10 = 1.15. The gradient for a log 2.20 luminance range would then be 1.15 / 2.20 = 0.52. If we take the unadjusted NDR of 1.05 and apply it to a log 2.20 luminance range minus flare, how much flare would it take to to equal 0.52? 1.05 / (2.20 - 0.20) = 0.52 or 2/3 stop flare.

    This is how flare really works.

    2 Quad flare example.jpg

    And for the Zone inclined:

    Zone Exposure - 2 Quad 1.jpg

    Here's an example of the effect on both the shadow exposure and the effective luminance range over a range of flare. In this example the no flare curve produces an effective luminance range of 2.20.

    Two Quad - Multi flare.jpg
     
    Last edited by a moderator: Feb 12, 2013
  22. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    Yes I relied on this quote. I never did this test myself and never really considered flare much. Well I never really had a problem with it because I always used the "corrected" methods by others. It only became an issue because of getting involved with BTZS.
    And because I started reading the posts of MR Benskin. Grrr...
    I need time for this to sink in.
     
  23. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    I see your graphs, I understand your saying about flare happening in the camera.
    Flare affects the optical image and not the film curve.
    How do I apply flare to my plotted curves. I am having a hard time getting there. Basically my pure clean curves are lying to me.
     
  24. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    Do you mean Photographic Materials and Processes by Leslie Stroebel? It has 608 pages!!
     
  25. Bill Burk

    Bill Burk Subscriber

    Messages:
    5,023
    Joined:
    Feb 9, 2010
    Shooter:
    4x5 Format
    Well you could do what I do. Pick a flare value that you believe applies to your system. I say 0.4 Log Exposure Units.

    At an appropriate part of the curve (and this is where I just throw darts and start where the curve crosses 0.1 density on the non-flare test plots), identify the exposure. Now identify the exposure at 0.4 Log Exposure Units to the right.

    Take the difference*. Add that amount (call it flare exposure) to the exposure for each plotted point. Look at the original graph to see what density you get for the summed exposure (original plus flare exposure*). Plot that density point directly above the original plotted point.

    As you move to the right, the flare exposure becomes logarithmically less significant, so the new curve will taper.

    Sounds awful, maybe someone can explain it better than me.

    *Get the antilogs so you are summing arithmetic values and then get the log of that sum
     
  26. AndreasT

    AndreasT Member

    Messages:
    355
    Joined:
    Sep 15, 2006
    Location:
    Berlin
    Shooter:
    Multi Format
    Thanks all, it is 3:21 AM, time to stop for today. Will be getting back to this.