Film curve plotting and fitting

[Kirk Keyes]

No matter how valid the studies from the 1960s on the subjective evaluation of tonal values were in their day it is extremely unlikely that people today making the same evaluation would come to the same conclusion for the very simple reason that photography today, and the way people look at photographs, is dramatically different from what it was a half century ago.

That's an interesting point. I assume that no one has reproduced Loyd Jones' work with materials from the last 25 years.

And since you mentioned Adams, Sandy, I'd like to point to "Moonrise, Hernandez" as a possible example of this effect. Adams printed it with less contrast in earlier years than in later years. Certainly some if it is due to Adams view as an artist changing over the eyars, but perhaps it was partially due to a shift in the way people's personal print viewing preferences had shifted over the years (and Adams own view in particular).

 

[Photo Engineer]

It is true that tastes vary with time.

Every time we designed a new family of film, or a new paper, the negatives had to be reshot so that they would reflect current tastes and so if you look for example at photos of the Kodak Girls over the years, they change artistic content reflecting this change.

Every few years, we reshot the test photos, and I remember many of the "evolutions" that we went through during my years designing products.

Yes, Sandy is correct.

And, with each new product we had panels to review results.

If I went into a meeting before the President, the CEO, and the CTO of EK (which I have) and spouted equations I would have been dead in an instant! I had to answer "is it on speed, on grain, on sharpness and if so show us the data and the photos to support it!"

That's about it!

PE

 

[Stephen Benskin]

The thing about input/output values is that they they are objective and easily mapped. Subjective tone reproduction is another issue, and one that falls into the area of art and taste. No matter how valid the studies from the 1960s on the subjective evaluation of tonal values were in their day it is extremely unlikely that people today making the same evaluation would come to the same conclusion for the very simple reason that photography today, and the way people look at photographs, is dramatically different from what it was a half century ago. You see, the plain fact of the matter about art is that it is always subjective, local and transitory.

To give you an example, I was recently reading Ansel Adam's book where he describes the making of forty photographs, covering a period from the 1920s until the early 1980s. What struck me in reading his tex and looking at the images was that his vision and the way he used the technique is specific to the way people viewed photographs in another period. In looking at one example where he explained how he dealt with a high contrast scene I realized that his interpretation would not be acceptable today because we have come to accept the vast potential of image process software to manipulate tonal values.

Sandy King

Sandy,

You are absolutely right that tastes change. That's why there are directions for the observers. For the first excellent print test they were instructed to pick the prints that gave the best impression of what they thought the real subject would look like. I've frequently said that if you consider speeds are based on those parameters, the film speed wouldn't really apply when the intent is not to reproduce the scene as it looks. What if the judgment prints where portraits? People liked darker tones in the 20s-40s. Since then, preferred skin tone has been getting lighter and lighter. What would the film speed be then?

As an individual, my taste in printing tends to swing back and forth between darker and lighter and more and less contrasty. But like any carefully prepared scientific test, I think they can determine the statistically average response which should eliminate trends and focus on the psychophysical phenomenon of basic human vision and perception.

Another question to ask is about the materials. Let's say you are asked to place the best to worse looking print in order. Easy enough for the materials you have on hand, but what if you aren't aware of some other materials or future materials. It's like looking for the best standard TV image quality before HD. Once HD comes long, how will that change the parameters of what is acceptable quality? - What are acceptable highlights and shadows, or contrast range etc.?

 

[Photo Engineer]

Steve;

The film speed could remain constant and the printing time or aperture could change during enlargement, so therefore your argument falls apart. In addition, this change with old films would emphasize either toe or shoulder information but with modern films only the density would change overall, all things being equal.

PE

 

[sanking]

To add just a bit to my previous comments about changes in artistic tastes we should look at what is being done today in the world of inkjet printing. As much as some people dislike digital photography it must be recognized that inkjet printing with pigment inks has become the ubiquitous method of photographic printmaking. Probably the best system being used today is JonCone’s Piezogaphy system. The profilers for this system distribute seven densities of carbon ink in multiple printing curves that produce a near perfect Gamma 2.2 linearization of input/output values from Dmin to Dmax. You could see it as the S curve of a silver print transformed to a perfectly straight line output.

Printing styles are highly personal and reflect an overall artistic vision that may or may not reflect the norm, but I think it fairly safe to say that the norm today in the highly specialized world of professional level inkjet printing is a perfectly linear curve, which in analog photography could only be made with a perfectly linear negative and a perfectly linear positive process. Shadow values with no separation, which often result from exposures that leave important shadows on the toe of the curve, are not be acceptable to me, and IMO, to the artistic sensitivities of most our contemporaries.

My own work, about which PE has commented, tends to also be highly linear from DMin to Dmax, and it results both from my use of a highly linear printing process, carbon transfer, and the creation of negatives that place all of the important value on the straight line section of the curve. I do this in analog work by choice of a film that gives a long straight-line curve, and by applying process curves and profiles to my digital negatives so that input/output values are equal.

Sandy King

 

[Stephen Benskin]

Steven;

The eye integrates an image and supplies a subjective contrast of unity when the mid tone is 1.5 and the toe and shoulder are optimum. A print with a contrast of 1.0 in the mid tones is flat-flat-flat!
PE

I didn't say the print contrast shoud be 1. I said the resulting contrast which means the contrast of the reproduction or in simple terms the relationship of the contrast of the subject to contrast of the print. The simple equation I was referring to was negative contrast x print contrast should equal 1 or unity. Since the basic assumption is to reproduce the scene, the resulting reproduction gradient should be unity.

According to tone reproduction theory, prints with compression of the highlights and shadows to varying degrees are not the determining factor as to what is considered a quality reproduction. What decides the perception of quality is the mid-tone gradient which must exceed ~1.12.

Perhaps we are just misreading each others posts, but Ron I'm getting extremely tired of your condescension. Either stop doing it or stop commenting on my posts. "I guess that is the only proof I can offer if the simple equation does not satisfy you. I am very sorry."

 

[Stephen Benskin]

Steve;

The film speed could remain constant and the printing time or aperture could change during enlargement, so therefore your argument falls apart. In addition, this change with old films would emphasize either toe or shoulder information but with modern films only the density would change overall, all things being equal.

PE

What argument? I've had many posts in this thread.

 

[Photo Engineer]

What argument? I've had many posts in this thread.

All right, statement then! The one in the post referred to by me!.

As for my "data", I have over 100 reports on this and similar subjects published internally at EK. I had to hand them in when I retired. I cannot produce them to justify 32 years of experience as a Photo Engineer.

I make no comments on your qualifications as I really don't know them. Frankly, they went under our radar so AFAIK I have never heard of you before I joined APUG and saw your posts. Sorry, but true.

My instructors in these matters were all members of the SPSE and ISO standards committees. That is about all I can add. Oh, I can add that I was an SPSE member too! It is now called ICIS, and my membership has lapsed. Are you an ICIS member? Did you attend the meeting last month in Peking? Do you know what the latest trends are in this subject? I have read through the huge 2" ICIS Peking meeting journal and am updating myself on the latest. If I run across anything new, I'll let you know!

The point is, that Sandy is right! This field is a moving target.

My apologies. Some of your comments are very good, but nearly 40 years out of date if you use studies from the 60s. They can help, but are not modern enough.

PE

 

[Stephen Benskin]

1. The pre-1960 ASA method was based on a speed point where fg=0.3, and the film speed was calculated such that this point corresponded to an exposure about 5-steps below that indicated by a calibrated meter. (Fig. 8, Nelson, C.N., Safety Factors in Camera Exposures, 1960). The useful luminance range was assumed to go down four stops from the indicated exposure, and the extra stop was considered a safety factor.

2. The revised ASA method included two changes. First, the fg=0.3 criterion was replaced with D=0.1. Testing showed that there was an approximately constant difference between these to points, especially when the negatives were developed to a constant contrast. With an average gradient of ~0.6, the fg=0.3 pt is about 1 stop lower than the D=0.1 point. Simply replacing the fg point with the D point (without changing the calculated ASA speeds) would have meant that the new speed point should be 4 stops below the metered point. At the same time, however, the ASA speeds were doubled, so that the D=0.1 point was now three stops below the indicated exposure. This was done in the belief that the 1-stop safety factor was no longer needed. The net result was that the low end of luminance range (-4 stops) was placed at the old speed point, where fg=0.3 and D is significantly less than 0.1.

David,

I want to get home and take a look at the paper again. I had a similar conclusion when I first saw figure 8. It appeared to go against what I thought I knew about the relationship between the meter and shadow exposure. If you look carefully, the Shadows fall at a Δ0.35 log-H units to the right of point s which represents the fractional gradient speed point. The Average luminance (meter) point is Δ1.57 log-H units from point s which would seem to indicate that the range between the metered exposure and the shadow is just over five stops. But the Shadow point doesn't reach down to point s, and the relationship between the Meter point and Shadow point is 1.22 which is four stops. I believe the figure illustrates the shifting of the exposure toward the right which is the result of flare.

Today's standard exposure model has the Δlog-H range between the speed point and metered exposure at 10X, log 1.0, or 3 1/3 stops while the luminance range between the metered exposure and shadow is Δ1.25 or 4 1/3 stops. Flare brings it back up. This one to 1 1/3 stop bump from flare can be considered part of a safety factor. As flare is variable and considering that you have the standard 2.2 log luminance range, flare can equal the unreal value of 0 and you will still have an excellent negative. As flare is never zero, any value above zero together with addition of a 1/3 of a stop exposure created by the constant 0.8 in the speed equation can be considered a safety factor.

Not considering how flare takes the luminance range and reduces the exposure range at the film plane has cause much confusion in interpreting exposure placement and film speed. That's really the reason why Zone System testing produces slower speeds that the ISO method. They are stopping down the approximately 4 stop difference between the metered exposure and the shadows while the difference at the film plane is only three 1/3 stops. I'll tell you more about the connection between the film speed number and mid-tone exposure sometime.

I also find Fig 1 and 5 to be of some interest as they address print quality vs log camera exposure. In Fig 1, you can see how quickly quality drops with increased exposure when the images are enlarged and that contacting shows a very broad range. As the smaller formats tend to be enlarged to a greater degree, this would indicate that smaller formats have a more limited exposure range than larger format films.

Did you find the first paragraph as ironic as me? "During the past three or four years, much criticism has been aimed at the safety factor involved in the use of American Standard exposure indexes' with exposure meters calibrated in accordance with American Standard procedure A number of articles in photographic magazines have pointed out the penalties and disadvantages resulting from the use of too large a safety factor and have urged that a smaller safety factor be introduced by means of a revision in the American Standard for determining ASA exposure indexes for black-and-white negative films. The general spirit of these articles is illustrated by the following title of one of them: "ASA Exposure Index: Dangerously Safe.""

Steve

 

[alanrockwood]

I would be interested to know if there is a functional form that reasonably accurately fits the asymptotic behavior of the exposure curve in the toe region. (The shoulder region would also be of some interest.)

Some possible candidates include the Logistic function (which has exponential tailing), the arctan function (which is the integral of a Lorentzian function and therefore a more gradual shoulder), the arctanh function, and the integral of the "normal" error curve (which as a fairly abrupt shoulder).

One advantage of using functions such as listed above is that they are sigmoid in shape, and therefore look roughly like the full exposure curve. It would seem reasonable that one could use a sum of two or three such functions (with translations in the x-axis, and appropriate y-scaling) to give a reasonably good fit to the exposure function over a very wide exposure range, including at least part of the shoulder. This could be done with relatively few adjustable parameters. However, to work in the best way this scheme would require that the sigmoid functions are reasonably good at fitting the tails of the exposure function.

Any thoughts?

By the way, some quantum chemistry programs use superpositions of Gaussian functions to represent approximate wave functions for atomic or molecular orbitals. This sort of scheme is somewhat close in spirit to what I am suggesting here. However, it differs in that the gaussian functions are not particularly good at matching the features of molecular orbitals, so it takes a lot of Gaussian functions to build a reasonably good molecular orbital.

 

[Stephen Benskin]

All right, statement then! The one in the post referred to by me!.

My apologies. Some of your comments are very good, but nearly 40 years out of date if you use studies from the 60s. They can help, but are not modern enough.

PE

No, I was asking you to quote my statement so that we both would know what you were talking about and then I could address it, but forget about it now.

I've read pretty much every book that has come out in the last fifty years on photography and I've read as many scientific papers on exposure and tone reproduction that I could get my hands on. Most of the material has been consistent over the years with little change. As far as I know, there just hasn't been anything new lately. I'd love to know about it if there were.

The 1993 ISO speed standard with all it's tweaks had no fundamental changes; however, the reasoning and theory behind the standard can be found in papers published from the 1940s and 1950s. That's where you can find the explanation for the ISO contrast parameters. The standard itself doesn't even hint at it. If you didn't read those dusty old sources, you wouldn't have a clue. I can't tell you how many authors have erroneously accused the standard of having too high an average gradient which consequently over rates film speeds.

The principles of psychophysical testing are still valid. Human physiology hasn't changed that much in the last few years. There hasn't been much in the way of advances in tone reproduction theory in the last 60 years. They figured it out. One of two papers I have available to anyone who wants one (I've noticed you're not one of them) is from 1994 and it contains the same concepts as from the 1960s and with some small changes in variables, the 1940s. As far as I know, most of the changes with traditional photography over the last 40 years has been in manufacturing, not in exposure theory or tone reproduction theory and those are the subjects I like to cover. If you remember, I've asked you for more modern references so that I could do additional research but you had none to offer.

Not modern enough is quite a statement to make without evidence. You might have noticed that when I state something as factual, I will include a reference source and often times a quote, otherwise it's just like an opinion. And everyone has one of those. I like the saying that extraordinary statements require extraordinary proof. You might want to consider giving that a try.

Oh, the "I'm from Kodak, who the hell are you?" act is desperate and despicable. We're done here.

 

[RalphLambrecht]

That's an interesting point. I assume that no one has reproduced Loyd Jones' work with materials from the last 25 years. ...

I have done it for myself to find out what people like- with one scene and 32 observers.
The results were very similar in some important areas:

1. high preference for increased midtone contrast
2. lesser of an issue with low shadow and highlight contrast
3. shadow and highlight detail important, but...
4. contrast more important than detail

 

[Mahler_one]

Steve...I am so sorry to say that the "We're done here" is good news. Frankly, I found PE to be tolerant to a fault. The expertise and bona- fides of PE, Ralph, and Sandy are well known. I have appreciated your posts. However, the current thread has denigrated into personal "attacks" that have become tiresome. Steve-best of luck nevertheless. I hope the storm passes.

Ed

 

[Photo Engineer]

No, I was asking you to quote my statement so that we both would know what you were talking about and then I could address it, but forget about it now.

Well, it was several posts earlier and I'm sorry I did not quote it. Several quick posts by others seemed to have intervened.

I've read pretty much every book that has come out in the last fifty years on photography and I've read as many scientific papers on exposure and tone reproduction that I could get my hands on. Most of the material has been consistent over the years with little change. As far as I know, there just hasn't been anything new lately. I'd love to know about it if there were.

You see, this is the problem. I have "DONE" and you have "READ". That is what I am trying to explain as the difference between our approaches. It is not so much a set of publsihed text books as it is the methodology taught to us at Kodak.

The principles of psychophysical testing are still valid. Human physiology hasn't changed that much in the last few years. There hasn't been much in the way of advances in tone reproduction theory in the last 60 years. They figured it out. One of two papers I have available to anyone who wants one (I've noticed you're not one of them) is from 1994 and it contains the same concepts as from the 1960s and with some small changes in variables, the 1940s. As far as I know, most of the changes with traditional photography over the last 40 years has been in manufacturing, not in exposure theory or tone reproduction theory and those are the subjects I like to cover. If you remember, I've asked you for more modern references so that I could do additional research but you had none to offer.

This is not entirely so if you read what Sandy King and other say, and the evidence is in the work of AA which is cited. Even fashions change.

Not modern enough is quite a statement to make without evidence. You might have noticed that when I state something as factual, I will include a reference source and often times a quote, otherwise it's just like an opinion. And everyone has one of those. I like the saying that extraordinary statements require extraordinary proof. You might want to consider giving that a try.

What will you do when a new author publishes, or an old author publishes here? Ask him/her to cite something? The new author, if going against conventional wisdom cannot cite anything and the old author falls back on old arguments adding new where they can with no citations. The work you have cited or posted here, I was doing in 1965 in the lab, often with my own coatings. I was learning by doing. I said that much of what you posted was good and correct, but did not account for the changes in films and papers in the last 40 years, and others have pointed out that the style of photography has changed as well, a factor I had forgotten to cite here. Even so, you seem to want to dismiss Sandy King and the others with less than a wave of the hand.

Oh, the "I'm from Kodak, who the hell are you?" act is desperate and despicable. We're done here.

I was citing my credentials only to show that I have had some experience in this field by actually doing the design work on such products. I hasten to add though that it is true that unfortunately Stephen Benskin was totally under our radar. We had working with us many members of the ANSI committees. That was a fact of life at Kodak, Ilford, Agfa and Fuji and I have met many of those very good scientists and engineers as well. Those are the people that are important to me, from all companies.

I have had access to them at many SPSE meetings or in most cases, with the Kodak people, I could walk down the hall and spend an hour with one of our experts on a thorny problem. I had no need, nor did any of us have a need to read literature which fell far outside the normal journals. See here:

http://facweb.cs.depaul.edu/sgrais/color_context.htm

or here:

http://books.google.com/books?id=8_...Ag#v=onepage&q=bartleson and breneman&f=false

or here:

http://www.scribd.com/doc/15758809/Daan-Zwick-The-Effects-of-Viewer-Involvement-and-Experience-on-

or here:

http://ieeexplore.ieee.org/Xplore/l...042940.pdf?arnumber=4042940&authDecision=-203

These are some of the experts with whom I consulted even though the articles are a bit off topic. They are what these individuals have produced. In particular, Jim Bartleson was a colleague of mine and we often consulted on thorny imaging problems and programming problems until his sudden untimely death.

You have stated that you have read, and I have said above that I have done. I have also written! Unfortunately, those reports are in the private files of Eastman Kodak along with other reports by those very individuals who taught me much of what I have stated here. I'm sorry. Perhaps you should request permission to have access to the EK-KRL Library and see what you can find. Or, the RIT or GEH Libraries which are quite complete. Even so, confidential reports are not available.

PE

 

[dpgoldenberg]

Printing styles are highly personal and reflect an overall artistic vision that may or may not reflect the norm, but I think it fairly safe to say that the norm today in the highly specialized world of professional level inkjet printing is a perfectly linear curve, which in analog photography could only be made with a perfectly linear negative and a perfectly linear positive process. Shadow values with no separation, which often result from exposures that leave important shadows on the toe of the curve, are not be acceptable to me, and IMO, to the artistic sensitivities of most our contemporaries.

Sandy,
I don't think that I understand your point here. Given that the range of scene luminance often greatly exceeds the range of reflectance from a print, how can the input/output relationship be linear over the entire range (and still maintain reasonable contrast in the mid-tones)? As I understand it, that has always been a fundamental challenge in photography, and I don't see how digital or ink jet techniques could have changed that very much. What is true, I think, is that these methods now allow tremendous flexibility to control the curve throughout the dynamic range, so that variations (contrast) can be detected very close to both ends. But my bet is that the overall response curve is very non linear (at least in my hands).

Am I missing something here?

David

 

[tlitody]

Don't move !

 

[ic-racer]

I want some software that will calculate fractional gradients.
Is that 'Python' script something that runs on the Macintosh?
How does it work. Do you feed it a dataset?

Just something to save time. So, for a 'first run' try with an unknown film that does not turn out to have my 'usual gradient' I'd still like to extract speed information before repeating the test with a different development time.

 

[alanrockwood]

View attachment 19778
Don't move !

I like this cartoon. I have used it in talks about mass spectrometry, with attributes of course.

 

[dpgoldenberg]

Ok, I have not read this thread, but I want some software that will calculate fractional gradients.
Is that 'Python' script something that runs on the Macintosh?

Sort of! I can send you an application package that should run on a Mac (OS X) without any other special stuff installed. It's pretty primitive, though. You will have to create an input file with a text editor, and then "drop" this file on the application. Windows will pop up with the text and graphical output.

I'm presently tweaking with the script to incorporate some of the ideas that have arisen here. But, I will be happy to share it shortly.

David

 

[tlitody]

I want some software that will calculate fractional gradients.
Is that 'Python' script something that runs on the Macintosh?
How does it work. Do you feed it a dataset?

Just something to save time. So, for a 'first run' try with an unknown film that does not turn out to have my 'usual gradient' I'd still like to extract speed information before repeating the test with a different development time.

I really don't undersstand why people don't just use excel charts. All you need to do is enter your zone densities once you have set up a chart. Very easy to copy for each test you do and reading off numbers to calculate slope at any two points on a curve is pretty simple.

 

[sanking]

Sandy,
I don't think that I understand your point here. Given that the range of scene luminance often greatly exceeds the range of reflectance from a print, how can the input/output relationship be linear over the entire range (and still maintain reasonable contrast in the mid-tones)? As I understand it, that has always been a fundamental challenge in photography, and I don't see how digital or ink jet techniques could have changed that very much. What is true, I think, is that these methods now allow tremendous flexibility to control the curve throughout the dynamic range, so that variations (contrast) can be detected very close to both ends. But my bet is that the overall response curve is very non linear (at least in my hands).

Am I missing something here?

David

David,

Not sure why you have a problem with what I wrote. We expose film to place it at some desired point on the film curve, and we then develop it so that the DR (density range) matches the ES (exposure scale) of our printing process. The printing process may be pure palladium, which has an ES of close to log 3.0, or a high contrast silver paper with an ES of log 0.8 or less. Regardless, in order to print in pure palladium we will need a negative with a DR of log 3.0 or about if we want to reproduce all of the tones, and we will need a negative with a DR of about log 0.8 to reproduce all of the tones with a high contrast silver paper. The prints will of course have a different look but which one is better is a subjective judgement.

With most wet processes (silver, vandyke, palladium, cyanotype) it is not possible to completely match input values of the negative with output values on the print, even if you have a perfectly linear negative, because there is usually some toe and shoulder in the print process. So it is often necessary to make compromises in printing by cheating either the highlights or the shadows, or by bumping up mid-tone contrast.

That is not true in inkjet printing because there you can exactly match input and output values. It is also not true with the wet process I use, carbon transfer, since the process curve is very straight line.

Sandy King

 

[Bill Burk]

This is a brilliant discussion, don't shut it down.

I respect Stephen for his depth of interest in this topic and his historical perspective. PE deserves recognition for a lifetime contribution to this field. Sandy brings up the fact that modern tools give a linear print and discussed changing tastes over time. PE, corroborates with direct boardroom experience - tastes have changed. That really illustrates the story!

David, if you are still working on a curve-fitting tool, I want it to have options to illustrate both historical and currently-accepted methods.

Suppose now that one scans a negative and prints inkjet. What happens when you shape the toe with curves in Photoshop? I'm not going to try because I tossed out my inkjet printers, but Sandy what if?

I use TMY-2 and was about to say I get a straight line from zero density to 2.0 density over a 3.1 exposure range. Then I laid a ruler on the graphs and found my straight line really begins at at around 0.4 density. But there is plenty of information all the way down to 0.06 density!

I also have to chuckle because in practice, I place my shadow on Zone II. Guess where that is? You guessed it - 0.4 density. And that's the negatives that make my best prints.

So we're getting somewhere.

 

[Stephen Benskin]

I want some software that will calculate fractional gradients.
Is that 'Python' script something that runs on the Macintosh?
How does it work. Do you feed it a dataset?

Just something to save time. So, for a 'first run' try with an unknown film that does not turn out to have my 'usual gradient' I'd still like to extract speed information before repeating the test with a different development time.

The Delta-X equations give fractional gradient speeds using only the 0.10 fixed density and the difference in density at 1.30 log-H units over from the fixed density speed point. I've written a paper on this that includes the equations if you're interested

 

[sanking]

Suppose now that one scans a negative and prints inkjet. What happens when you shape the toe with curves in Photoshop? I'm not going to try because I tossed out my inkjet printers, but Sandy what if?

Within limits you can decrease shadow contrast or increase shadow contrast by shaping the toe. Same is true for any other point on the curve. You can also easily cause posterization if the manipulation is too extreme.

I would add that in the event one plans to scan there is no reason to expose to place threshold shadow values on the straight line part of the curve. Since they can be manipulated after scanning it is possible to place them even lower than log 0.1. But this being APUG I won't pursue the point.

Sandy King

 

[Photo Engineer]

Suppose now that one scans a negative and prints inkjet. What happens when you shape the toe with curves in Photoshop? I'm not going to try because I tossed out my inkjet printers, but Sandy what if?

A digital print, massaged to give an analog curve looks very close to an analog print but with the typical digital defects. It is, however, very good and close to what can be done with the best darkroom equipment. And, it is getting better as time goes by. The digital prints however, still suffer from smear with time due to dye migration and to fade due to poorer dye stability.

I use TMY-2 and was about to say I get a straight line from zero density to 2.0 density over a 3.1 exposure range. Then I laid a ruler on the graphs and found my straight line really begins at at around 0.4 density. But there is plenty of information all the way down to 0.06 density!

Yes, there is, and there was more toe in older films, but to get the result you have to take the slope at any given point and figure out the density difference in the print. The problem, as you will see, is that the slope is very small and the density differences are hard to see. The overall print quality will suffer as a result of this "underexposure".

I also have to chuckle because in practice, I place my shadow on Zone II. Guess where that is? You guessed it - 0.4 density. And that's the negatives that make my best prints.

And this is exactly how it should work. Slight "overexposure" is best.

So we're getting somewhere.

I sincerely hope so.

Thanks for an interesting post. I'm glad that you do not dismiss my practical experience out of hand. Nor have you denigrated the fact that I worked for Kodak. After all, they did have a few good people there and a few of them were on these ANSI committees. And, some of them taught me!

PE