Switch to English Language Passer en langue franšaise Omschakelen naar Nederlandse Taal Wechseln Sie zu deutschen Sprache Passa alla lingua italiana
Members: 68,685   Posts: 1,482,280   Online: 926
      
Page 6 of 8 FirstFirst 12345678 LastLast
Results 51 to 60 of 77
  1. #51
    Ian Grant's Avatar
    Join Date
    Aug 2004
    Location
    West Midlands, UK, and Turkey
    Shooter
    Multi Format
    Posts
    15,943
    Images
    148
    Quote Originally Posted by Rudeofus View Post
    If lowering Sulfite would have improved all key parameters of D76, one would wonder how Kodak could have missed that when they formulated it. Something is likely missing in this analysis.

    None of the developers you list here are D76 with reduced Sulfite, they all have different compositions, and Microphen is not even an MQ type developer. Their low Sulfite content proves nothing, and looking at Promicrol and various Crawley developers tells me that 100 g/l Sulfite or even more never really went out of fashion.
    A major problem was that D76 had become a standard developer and it was difficult to change an existing formula, it was made by a large number of companies, and results had to be consistent regardless of where in the world film was processed. So Ilford ID-11, Agfa 19, Defender 6, Foma FV3, Forte FD20, Foton N12 and others were all the same formula (Agfa added Bromide as a starter). Companies might buy what ever was on the shelf and repelenish with a competitors product. Ilford brought out ID-11plus in the US it was withdrawn because of cpmpatabilty issues

    Actually lowering Sulphite is imortant, you need to also take into account dilution after all D76 @ 1+1 is 50g/l working dev Sulphite.

    While Microphen/ID-68 is a PQ developer not MQ it comes from research by Ilford into a PQ variant of ID-11/D76 which went on sale as Autophen in the late 1950's. As Phenidone is far less sensitive to bromide build up this could be replenished more economically and wasn't prone to collapse like D76/ID-11. In many US publications the formula for Autophen is mistakenly assumed to be Microphen.

    With Microphen/ID-68 Ilford's dropping the Sulphite level is similar to the MQ developers from Agfa (& Agfa-Ansco), Adox etc.

    It's generally forgotten that in commercial/professional use Developers like D76/ID-11, Agfa 44 (Agfa Ansco 17), etc were always used replenished and that once seasoned gave far better results than fresh developer.

    Crawley recommends FX3 with it's 75g/litre sulphite where less loss of film speed is important.

    Ian
    Last edited by Ian Grant; 12-22-2012 at 04:06 AM. Click to view previous post history.

  2. #52
    Rudeofus's Avatar
    Join Date
    Aug 2009
    Shooter
    Medium Format
    Posts
    1,424
    Images
    10
    As soon as one accepts the noisy nature of these transitions one should start wondering why so many smooth curves were published. In order to measure such a smooth transition from dark to light you need to average over many silver particles, which also alters the result one measures that way. These measurements were invented by chemists and the telecommunications engineer in me cringes when it sees them.

    To give you a hint: Kodak used an 48Ám diameter aperture to measure noise of a system with a few microns of granularity. Just think how this measurement becomes meaningless if you enlarge a negative twentyfold! That measurement may correlate somewhat with real granularity for coarse grained film but inevitably breaks down with TMAX in Microdol X. All hell breaks lose once people throw in their own flawed measurements for grain and sharpness.

    To sum it up: not only is it incredibly difficult to establish a number for perceived sharpness (as others have already noted), all these published measurements won't hold much water when exposed to scientific scrutiny and mathematical rigor. Don't get hung up on published numbers.
    Trying to be the best of whatever I am, even if what I am is no good.

  3. #53
    Rudeofus's Avatar
    Join Date
    Aug 2009
    Shooter
    Medium Format
    Posts
    1,424
    Images
    10
    Quote Originally Posted by Ian Grant View Post
    Actually lowering Sulphite is imortant, you need to also take into account dilution after all D76 @ 1+1 is 50g/l working dev Sulphite.
    Again, D76 1+1 is a different animal than D76 since more than just Sulfite concentration changes.
    Quote Originally Posted by Ian Grant View Post
    Crawley recommends FX3 with it's 75g/litre sulphite where less loss of film speed is important.
    There are still 100 g/l or more in Promicrol, Crawley's FX-4 and FX-11, and they are all speed enhancing developers. One would assume that FX-4 and FX-11 were formulated after FX-3. I would say that for just about any reasonable combination of Metol, HQ, Sulfite and Borax one can find a published developer recipe with a matching formula, so posting a few recipes with 80 g/l Sulfite proves nothing.

    The film developing cookbook makes a strong claim that Sulfite should be lowered to 30 g/l for T-grain type films, unfortunately without explanation. It still doesn't state "use the same recipe just lower Sulfite", though.
    Trying to be the best of whatever I am, even if what I am is no good.

  4. #54

    Join Date
    Aug 2006
    Shooter
    Multi Format
    Posts
    297
    Quote Originally Posted by Rudeofus View Post
    These measurements were invented by chemists and the telecommunications engineer in me cringes when it sees them.
    So how would a telecom engineer look at this situation? Would they just concede that it is too difficult?

  5. #55
    Rudeofus's Avatar
    Join Date
    Aug 2009
    Shooter
    Medium Format
    Posts
    1,424
    Images
    10
    Quote Originally Posted by Mr Bill View Post
    So how would a telecom engineer look at this situation? Would they just concede that it is too difficult?
    Of course it is nontrivial to model grain accurately, since at the size range of typical silver particles all kinds of effects including diffraction take place. But it's still worth knowing where ones simple models fail. Allow me to illustrate this with 3 graphs to show where and how a 48Ám aperture fails to measure grain and sharpness. The attached graphs simulate the following:
    1. A linear density transition over 40Ám from 1 to 2 was assumed. One sample represents 1Ám which comes close to real film silver particles. Note that a 40Ám transition does not look all that sharp if enlarged 15 times.
    2. On top of that, 3 different Gaussian white noise patterns were superimposed, with standard deviation proportional to density. These are the blue curves.
    3. These noisy transitions were convolved with a rectangular window 48Ám=48 samples in width. These are the green curves. I use this as simple model what a 48Ám aperture does to the measurement.
    4. Three such plots were made for standard deviation of 0.1*D, 0.25*D and 0.5*D. Names indicate which one is which in case the order gets messed up by the forum software. If you look at a heavily enlarged print, the 0.25*D noise version looks quite realistic.

    The nice smooth green curves are what gets published, the blue curves are likely closer to reality and, more importantly, what our eyes and brains get to see and interpret as sharpness.
    Attached Thumbnails Attached Thumbnails Stdnoise_40um_0.50.gif   Stdnoise_40um_0.25.gif   Stdnoise_40um_0.10.gif  
    Trying to be the best of whatever I am, even if what I am is no good.

  6. #56

    Join Date
    Nov 2004
    Posts
    1,091
    I looked at one of Crawley's film reviews in which he scanned negatives of bars of ever decreasing thickness with a microdensitometer.
    He says that main subject outlines are sharper with greater height or amplitude (perhaps that would correspond to adjacency effects) but fine detail definition gives greater amplitude at higher lppm (might correspond to solvent developer where the grains are smaller).
    Effect of developer composition would then be shown,not as a number but as a series of charts similar to my sketch,possibly.
    But microdensitometer costs >$100k ?
    Attached Thumbnails Attached Thumbnails Microdensitometer scans (hypothetical).jpg  

  7. #57

    Join Date
    Aug 2006
    Shooter
    Multi Format
    Posts
    297

    How a telecom engineer sees it...

    Thanks Rodeofus. Let me see if I understand. The numbered steps below are my take on what you did.

    1) You started with a hypothetical ideal response, made of 3 straight line segments.

    2) Then you invented examples of how the ideal response might appear if it were contaminated with random noise, using three different "rates of noise." These are the blue curves.

    I presume that these don't represent actual grain. Rather they represent how actual grain might appear to a perfect instrument reading point samples? (Ideally the dots would not be connected as there is no clear relationship between them?)

    The more I think about this, the less clear I am on what the blue curve is intended to represent.

    3) Finally you run a convolution across the blue-curve data points to simulate what a real instrument, equipped with a 48 micron aperture, might see (the green curve).

    If I am understanding this right, the green curves represent the data which would be used for the acutance calculation? If so, I can see that this data has problems. I appreciate that you did quite a bit of work putting this together, but I'm still not sure exactly what it means. If the real data is obtained through a circular 48 micron aperture, the sampling zone would have width in additon to length, whereas it looks like (to me) you have simulated data collection along a line, with no width.

    Am I following this right?

  8. #58
    Photo Engineer's Avatar
    Join Date
    Apr 2005
    Location
    Rochester, NY
    Shooter
    Multi Format
    Posts
    22,709
    Images
    65
    Excellent article Alan. I have 2 sets of the charts shown on page 129. There is also a good discussion of this in Anchell and Troop.

    The method you describe that was used by Crawley is similar to what is posted on the Kodak web site for every film.

    PE

  9. #59
    Rudeofus's Avatar
    Join Date
    Aug 2009
    Shooter
    Medium Format
    Posts
    1,424
    Images
    10
    Quote Originally Posted by Mr Bill View Post
    Then you invented examples of how the ideal response might appear if it were contaminated with random noise, using three different "rates of noise." These are the blue curves.
    I read a couple of papers on how grain is modeled and Gaussian distribution was commonly used. We can safely assume that this model is not based on physical properties of film grain but on mathematical convenience (if your best tool is a hammer you model everything as nail). I think I can still justify my model because I don't claim any results which depend on the actual random distribution of the noise and because the blue curves look similar to what I see when I enlarge my negatives. I certainly do not see smooth density like what Alan sketched.
    Quote Originally Posted by Mr Bill View Post
    I presume that these don't represent actual grain. Rather they represent how actual grain might appear to a perfect instrument reading point samples? (Ideally the dots would not be connected as there is no clear relationship between them?)
    Yes, that comes close to it. The blue curve indicates what a densitometer with 1Ám aperture would likely measure.
    Quote Originally Posted by Mr Bill View Post
    Finally you run a convolution across the blue-curve data points to simulate what a real instrument, equipped with a 48 micron aperture, might see (the green curve).
    Correct. A 48Ám aperture sees something completely different than what is really there since it is much coarser than the actual granularity of B&W film. The nice thing about the green curve is that it looks so amazingly smooth, and I begin to wonder why all these plotted step response curves look so nice and smooth. Which aperture did they actually measure with? If their aperture is a few microns, where did the grain go? If it is much wider, how credible are their step responses?


    So why do I have an issue with this? All these measurements don't measure the real thing but something that can be measured with reasonable effort and hope that the result somehow correlates with the real thing or at least our visual perception of it. What might have been a valuable estimate for telling roughly whether an emulsion is finer grained or not is suddenly is used to distinguish between 100% and 102% granularity/sharpness just because a machine spits out these two numbers.
    Trying to be the best of whatever I am, even if what I am is no good.

  10. #60

    Join Date
    Aug 2005
    Location
    Los Alamos, NM
    Shooter
    Multi Format
    Posts
    2,016
    Quote Originally Posted by Rudeofus View Post
    Of course it is nontrivial to model grain accurately, since at the size range of typical silver particles all kinds of effects including diffraction take place. But it's still worth knowing where ones simple models fail. Allow me to illustrate this with 3 graphs to show where and how a 48Ám aperture fails to measure grain and sharpness. The attached graphs simulate the following: ....
    Excellent points. One thing about the green curves is that the roughness of them still reflects the amount of noise. A question is how well they reflect the amount. I suspect the reflection is less reliable as the noise goes down, but we could take the RMS deviation from the average curve and get a decent idea. It is worth noting that Kodak gave up on RMS granularity as a measure of film grain for regular still film image products and went to a subjective measurement system which, they say, better reflects actual conditions.

Page 6 of 8 FirstFirst 12345678 LastLast


 

APUG PARTNERS EQUALLY FUNDING OUR COMMUNITY:



Contact Us  |  Support Us!  |  Advertise  |  Site Terms  |  Archive  Ś   Search  |  Mobile Device Access  |  RSS  |  Facebook  |  Linkedin