speed point, 0.1 over base plus fog or 1/3 of gamma?

[alanrockwood]

Enjoy.

The internet these days. You know how long it took me to track down all these papers? Searching through the stacks at USC and UCLA, and all those hours standing over a hot copier machine.

Yes, reminds me of my graduate school days, back in the paleolithic era.

 

[alanrockwood]

The vertical axis is log-E or today log-H which I have to believe is the difference in where the film speed fell from the aim of 0. 0.1 would be 1/3 stop. The plus and minus would be considered faster or slower speeds than the judged speed.

How about the x-axis? Does each point represent a photo that was evaluated?

 

[Stephen Benskin]

How about the x-axis? Does each point represent a photo that was evaluated?

At the bottom of the ΔX chart, there are number codes that represent the print or corresponding negative from the tests. There is a lot more detail on the materials and how they were handled in the papers associated with the first excellent print test and it's conclusions. Like The Evaluation of Negative Film Speeds in Terms of Print Quality, parts 1 and 2. One is more on the conditions of the test and two on the results. And The Brightness Scale of Exterior Scenes and the Computation of Correct Photographic Exposure.

Fun thought, the judges for the first excellent print test were instructed to pick the best print that most closely represented what the original scene would have looked like in their mind's eye. What would the results have been if the instructions were different? What does that mean to a photographer, in regards to film speed, who isn't intending to do an accurate rendition of a scene?

 

[Stephen Benskin]

I decided to skim through Jones' paper The Evaluation of Negative Film Speeds in Terms of Print Quality as it's been many years since I last read it. I found a number of really good quotes that might be of interest.

Print Judgements
" The long and laborious computation of tone reproduction quality and the attempt to establish a satisfactory correlation between sensitometric results and the statistical results derived from the print judgements was begun."

Evaluation of Speed
"The speed of a photographic negative material should be evaluated in terms of the minimum exposure which will yield a negative from which an excellent print can be made."

"Statistical Evaluation. - The most direct method making the evaluation as defined above appears to be by the direct judgement of print quality by a sufficiently large group of observers so that the average may be relied upon to agree with the average opinion of the photographic public at large."

"While there is no question that the statistical method already described yields results of satisfactory significance, the operation of such a method of speed determination in practice is quite impossible. The amount of labor involved in making a series of negatives and prints therefrom and in having them judged by a sufficient number of observers to establish the desired certainty in the statistical average is prohibitive. It is essential; therefore, that some sensitometric method be found which will yield results in close agreement with those obtained by the statistical method. As already pointed out, speed values based upon inertia are not proportional to those given by the statistical method of direct print judgement, nor will the use of a low fixed value of density accomplish the desired end."

Conclusions
"It seems impossible to find in the theory of tone reproduction a justifiable reason for the use of any fixed value of density or the use of the inertia value as a criterion of effective camera speeds."

 

[Stephen Benskin]

Some results from the Print Judgement method:

The number of the negative that yielded the print is used in the following evaluation.

Here are the negatives that produced the four examples. I emphasized the inertia point which was one of the speed methods used. Depending on the curve shape, it's relationship with M (approximate 0.3x) varies considerably. It's also interesting to note the negative preferred for print D has higher exposure than the others and that it belongs to a long toed curve.

 

[Bill Burk]

I never knew there was a second peak!

 

[ic-racer]

This shows that over-exposure and over-development are horrible for small format negatives.

I just printed some HP5 Minox negatives. Exposed at EI 200 and inadvertently processed for 6 min, instead of my normal 5 min for that film.

The image is in focus and the shutter speed is 1/2000, yet the resolution has diminished quite substantially from what would be achieved under ideal conditions.

 

[MattKing]

I admire the work Jones and his group did. However it is difficult to know if the results correlate with the way I print, so all I could ever really say about negative film speeds is that they are a good standard.

Maybe I would have considered the first excellent prints excellent. Maybe not. One thing I have noticed over the years is that the way in which “the photographic public at large” (quoting Jones) looks at prints is somewhat at odds with my own preferences. Nothing snobby or right/wrong meant by this. Just a difference.

The biggest advantage of the "first excellent print" approach is that it references how a large percentage of viewers see things.
As photography is a communicative art, it is really useful to have a sense of what communicates well.
It is always open to a photographer to present their work in a way that is different than the norm. But to do so, it is important to have a sense what that "norm" is.
It also supplies a target for mass production of prints - the high volume commercial lab environment.

 

[ic-racer]

The way I interpret the work in the literature is that print evaluation is supreme, and most advanced darkroom workers will know what negative will make a good print and have determined how to expose to get there.

From that point as a reference, one can work backwards to find your speed point and use that for an unknown film. Using a sensitometer and densitometer, without having to make a year's worth of prints to 'learn' the film, one can start using the film right away. One can even do this speed determination at night, when there is no good light for any in camera testing of the unknown film.

https://www.photrio.com/forum/threads/testing-some-new-film-shanghai-8x10-how-i-do-it.171836/

 

[Stephen Benskin]

In any case, it is the derivative of density with respect to exposure (ie contrast) that matters. It seems like that should be trivially obvious (and I suspect it always was), but empirical evidence is good.

One of the reasons why I include the historical view with my answers is to show there are no absolutes. Science has caveats. And that systems, such as the Zone System, tend to oversimplify things.

All the speed criterion have dealt with gradient in one way or another. It's always been pretty obvious as to it's importance. According to Jones, "Hurter and Driffield defined a perfect negative as one in which the opacities are directly proportional to the brightness of the object photographed." Their inertia speed method used Gamma to define the speed point and then used a constant as the numerator for the film speed value. The value of the constant changed over time, but the concept that the straight-line portion was instrumental to reproduction quality remained.

The various fixed density methods are similar as they assume their position will fall somewhere in the toe with a enough of a gradient to produce a good print. In the mid-1930s, just prior to the first excellent print testing, Jones considered a minimum useful gradient as the speed point. I believe he considered something like 0.35.

What makes the first excellent print test and the fractional gradient method different is that it is based on psychophysical tests, not just assumptions base on interpretation of sensitometric data. Munsell's gray scale wasn't just about equal steps between black and white. It was about perceived equal steps which wasn't simply evenly dividing the reflection range (Note for Zone System practitioners). It's pretty obvious to most people what the color blue is, but it took until the 1940s to define color as a psychophysical phenomenon and not something inherent in the object or as psychological (also Jones).

 

[alanrockwood]

This shows that over-exposure and over-development are horrible for small format negatives.

I just printed some HP5 Minox negatives. Exposed at EI 200 and inadvertently processed for 6 min, instead of my normal 5 min for that film.

The image is in focus and the shutter speed is 1/2000, yet the resolution has diminished quite substantially from what would be achieved under ideal conditions.
View attachment 271411

View attachment 271412

Is it decreased resolution you are seeing or increased grain or both?

 

[RalphLambrecht]

I am not sure if this post goes best in the exposure section of the darkroom section.

I understand that early-on Kodak folks considered the best choice for speed point would be where the slope of the characteristic curve is 1/3 of gamma. Later they joined the rest of the world and went with a density of 0.1 over base plus fog as choice of speed point.

I suppose for a short toe film there's probably not much difference, but for a long toe film it might make a difference. If so, and if one could do accurate measurements of the slope of the characteristic curve (admittedly a challenge), for determining one's personal film speed (EI), would 1/3 gamma be a better choice than 0.1 over base plus fog? Intuitively it seems to me that it would be a better choice.

When I mentioned a short toe film it is based on the idea that if the curve was a straight line with a perfect break to zero slope at the toe (i.e. no toe) there would be no difference between the two methods. That's why I speculate that for a short toe film the two choices are probably not much different.

Opinions?

a density of 0.1 over base plus fog as choice of speed point or 2/3 below box speed has always served me well in my film testing.

 

[Stephen Benskin]

Excerpts from The Theory of the Photographic Process, 3rd edition.

"In a practical sense, however, the fractional gradient criterion has certain weaknesses. The method requires the determination of two gradient values, and unless great care is taken in making these measurements, significant errors can arise in the determination of speed. For this reason primarily, the fractional gradient criterion has not received universal acceptance. Experience with its use has shown that the errors arising from the measurement of the two gradients more than offset the theoretical advantage of the fractional gradient criterion. The fixed-density criterion for determining speed has found strong support on the grounds that it is more convenient to use and is subject to smaller experimental errors than the fractional gradient criterion.

The possibility of effecting a compromise between the fractional gradient method and the fixed density method for determining speeds was suggested in a paper by Nelson and Simonds. They studied the sensitometric characteristics of hundreds of different kinds of negative materials and found that the relation between fractional gradient speeds and speeds based on a density of 0.10 above fog is systematically related to the average gradient of the D-log E curve on which the measurements are made."

"In 1957, the responsible committees of the American Standards Association undertook the drafting of a revised American Standard Method for Determining Speed of Photographic Negative Materials. One of the primary objectives in writing the new standard was to specify a method of testing which would permit the use of a fixed density speed criterion. The view was held that by abandoning the fractional gradient criterion, which had been specified in previous American Standards, and adopting a fixed density criterion, a serious obstacle in the way of international agreement on the method for determining speed would be removed. Because, in addition, the use of a 0.1 fixed density criterion in consideration with a suitable development specification offers the convenience and precision characteristic of this criterion and retains the practical significance of the fractional gradient criterion, this method was adopted in the new American Standard."

- This refers to the Delta-X Criterion for the American Standard.

 

[alanrockwood]

Excerpts from The Theory of the Photographic Process, 3rd edition.

"In a practical sense, however, the fractional gradient criterion has certain weaknesses. The method requires the determination of two gradient values, and unless great care is taken in making these measurements, significant errors can arise in the determination of speed. For this reason primarily, the fractional gradient criterion has not received universal acceptance. Experience with its use has shown that the errors arising from the measurement of the two gradients more than offset the theoretical advantage of the fractional gradient criterion. The fixed-density criterion for determining speed has found strong support on the grounds that it is more convenient to use and is subject to smaller experimental errors than the fractional gradient criterion.

The possibility of effecting a compromise between the fractional gradient method and the fixed density method for determining speeds was suggested in a paper by Nelson and Simonds. They studied the sensitometric characteristics of hundreds of different kinds of negative materials and found that the relation between fractional gradient speeds and speeds based on a density of 0.10 above fog is systematically related to the average gradient of the D-log E curve on which the measurements are made."

"In 1957, the responsible committees of the American Standards Association undertook the drafting of a revised American Standard Method for Determining Speed of Photographic Negative Materials. One of the primary objectives in writing the new standard was to specify a method of testing which would permit the use of a fixed density speed criterion. The view was held that by abandoning the fractional gradient criterion, which had been specified in previous American Standards, and adopting a fixed density criterion, a serious obstacle in the way of international agreement on the method for determining speed would be removed. Because, in addition, the use of a 0.1 fixed density criterion in consideration with a suitable development specification offers the convenience and precision characteristic of this criterion and retains the practical significance of the fractional gradient criterion, this method was adopted in the new American Standard."

It is true that any method based on slopes extracted from experimental data is going to be subject to error. In essence, when you take derivatives of noisy data the process ends up amplifying noise, and experimental data always contains noise to some extent.

In principle this problem of sensitivity to noise can be reduced but not eliminated by curve fitting, but for that to work the functional form of the fit function should be a reasonably similar to the underlying (noise-free) data. For example, a fit function that is too flexible will typically not satisfy this criterion because it will be more likely to try to follow the noise at the expense of the underlying data than would be a function that less flexible but more similar to the underlying data.

For example, if the underlying data should fit a roughly bell-shaped function then a three-parameter Gaussian function (mean, standard deviation, and amplitude) is more likely to give a good fit to the data than a three-parameter polynomial. This would be true even if the underlying data is roughly bell-shaped but not a true gaussian function.

It was interesting that in one of the papers previously cited the fractional gradient method produced better agreement with the panel of photo judges than did the 0.1 over base plus fog method.

 

[Bill Burk]

(Tongue in Cheek but sort of chuckling what if) The way things are these days, maybe we should all participate in crowd source based print judgement speed tests whenever a new film, developer or technique comes out.

 

[ic-racer]

(Tongue in Cheek but sort of chuckling what if) The way things are these days, maybe we should all participate in crowd source based print judgement speed tests whenever a new film, developer or technique comes out.

Actually not that crazy. We could to it here on phototrio. The most difficult part will be making the prints and ensuring a traceable path of data back to the original exposure. Can you think of any other way to make the negatives other than photographing a transparency? Maybe with a camera with interchangeable film backs and motor drive, one could rapidly get the negatives in a short period of time for a real-life scene that is very stable in lighting.

 

[Stephen Benskin]

Can you think of any other way to make the negatives other than photographing a transparency?

That was Jones' conclusion. "In making several series of exposures it is extremely difficult to be sure that the lighting conditions on an outdoor scene will remain sufficiently constant over an adequate period of time. Moreover, it may be desirable to make these series on different days and here the case of controlled and reproducible lighting is even more hopeless. With a living model...experience has shown that it is practically impossible for a living model to maintain the same pose over a sufficient period of time and even more difficult fo9r the model to assume exactly the same pose on different occasions."

If I am reading the paper correctly, the transparencies are made from contacting original negatives. The negatives for the judgement prints were made from using a copy camera to shoot the transparencies with a range of exposures on multiple film stocks. So they are copy negatives from interpositives.

 

[Stephen Benskin]

I never knew there was a second peak!

Found an answer to that. "In the region to the right of A, there is considerable disagreement among observers as to whether the general tendency is somewhat upward or somewhat downward, or possibley oscillation about horizontal line, as indicated by the solid curve. While there may be some doubt as to the exact shape of this function, there is no doubt whatever as to its general shape."

I believe this indicates there is little noticeable difference in quality beyond a certain point.

 

[Bill Burk]

Thanks, that's likely contact prints, The other chart that show degradation for 10x enlargement is good to keep in mind. For 35mm you will want to hit close to the sweet spot.