Box ISO rate and Real ISO

Photo Engineer · May 2, 2014

It should be noted here that both "first acceptable print" and the "zone system" are less than quantitative measures due to their very nature. They are "shadows" of real sensitometric tests based on the ANSI standards. And combining the sensitometry with "first acceptable print" (as is done in Mees) one has a really effective method of examination of this problem.

Having done this sort of thing dozens of times with hundreds and hundreds of "users" and "customers", I can say that "when things are said and done, more things are said than done".

BTDT. Guys, go out and do about 1000 or so exposures and 3000 or so print tests with about 100 people.

Sorry for the rant.

PE

Bill Burk · May 2, 2014

Haa, Every once in a while I knuckle down and get something done.

I have a roll of TMAX 100 to develop that I shot with the new (to me) Kodak Retina I where I thought I calibrated the shutter speeds. But I have my doubts about the shutter test I performed. I'm a little worried. Seems the responsiveness of the sensor in my tester has a bit of lag, which over-estimates the time. So where I thought 1/100 was 1/40, it might really be 1/100. I don't like making mistakes in the underexposure direction so I am going to shoot a test roll - and I rarely do that - just to validate the shutter speeds. I'll include sensitometry exposure and develop the two rolls together so I'll soon find out whether I knew what I was doing all along or if I just thought so.

Stephen Benskin · May 2, 2014

Doremus Scudder said:
But, for the sake of clarification, when I develop, say N-2 in order to accommodate a rather high subject brightness range, am I not changing the gradient and therefore the effective speed of the film? Or would you say that this is outside the parameters of ISO testing standards and simply does not apply?

You are asking the wrong question. I’ve already covered it to a degree in other posts so you may already know what I’m about to write. The effective film speed changes with changes in development. The question is how to measure it. As I’ve pointed out, a fixed density method lacks good correlation with the print judgment speeds. The fractional gradient method and the subsequent Delta-X method uses the point on the curve that is 0.3x of the average gradient. This point tends to shift in a inverse relationship to a fixed density point which causes a slower change in the effective film speed than with the fixed density method.

This example from the Simple Methods for Approx Fractional Gradient Speeds paper shows the inverse relationship between ΔD (contrast) and ΔX (difference between the 0.10 fixed density point and the fractional gradient point). As ΔD decreases, ΔX increase and visa-verse effectively tempering changes in effective film speed with changes in development.

The chart shows three sensitometric exposures from the same film type developed to different contrast indexes. One corresponds to the ISO contrast parameters and the others are developed to a less and greater degree. An effective film speed was determined from each of the samples using both the 0.10 fixed density method and the Delta-X method.

Should I really be questioning whether my testing method "is yielding reliable results"? Are the Zone System and BTZS tests unreliable?

Technically, yes to both questions. Practically, no to both. You don't need this stuff to shoot. I think it's good to know if only to understand the lack of precision with the type of equipment generally available and concern yourself with what's really necessary to produce good images.

Doremus Scudder · May 3, 2014

Stephen Benskin said:
You are asking the wrong question. ...

The chart shows three sensitometric exposures from the same film type developed to different contrast indexes. One corresponds to the ISO contrast parameters and the others are developed to a less and greater degree. An effective film speed was determined from each of the samples using both the 0.10 fixed density method and the Delta-X method.

View attachment 87261

... You don't need this stuff to shoot. I think it's good to know if only to understand the lack of precision with the type of equipment generally available and concern yourself with what's really necessary to produce good images.

Stephen,

First, thanks for taking the time to explain in terms I can deal with. I'm busy digging through the papers you posted. That said, the excursion is a bit on the superfluous side, since my primary concern/goal is to improve my technique in the field. And, although you maintain that, "you don't need that stuff to shoot," I'm finding that much of it is useful, if only in clearing the clutter of accumulated misconceptions and Zone System inaccuracies. I'll continue to use the Zone System, though, since I find it such a wonderful visualization tool, but I'll be trying to refine exposure some now.

I mentioned earlier a disconnect (for me, anyway) between ISO and personal E.I. This connects somewhat with your discussion on gradient speeds vs. fixed point speed. First, let me say that I'm now convinced that the Delta X or fractional gradient speed based on first excellent print testing is likely a better method than the Zone System fixed-point speed determination. But, if I accept that, then it would seem that much of the exposure compensation that I have been doing for expansions and contractions would be unnecessary. This confuses me a bit, for it really seems to me that when I make a negative of a very contrasty scene that requires, say, N-2 development, that, unless I give substantially more exposure (along the lines of that suggested by the fixed-point speed) to compensate for the reduced development, my shadow values suffer. Conversely, expansion negatives are often overexposed (which doesn't bother me much) if I don't compensate by reducing exposure.

This is the disconnect I am addressing. And, I don't even really know where to start asking informed questions about it except to fire away and hope you will indulge me yet again. So, did the first excellent print tests include lots of expansion and contraction subjects, or were they more or less concerned with prints of subjects with more "normal" SBRs? If so, then should we expect much different results for similar print tests with subjects with significantly higher or lower SBRs? If not, then why does it seem that in practice I have to compensate exposure for contractions and expansions than the Delta X or fractional gradient speeds would suggest?

I'll leave it at that for now, as I'm probably asking the wrong questions again

Looking forward to your answer,

Doremus

markbarendt · May 3, 2014

Stephen Benskin said:
You don't need this stuff to shoot. I think it's good to know if only to understand the lack of precision with the type of equipment generally available and concern yourself with what's really necessary to produce good images.

I agree that we don't "need" this stuff to shoot.

The value I find in building my understanding of these concepts though, and in getting the terminology/language right, is in taking the mis-understandings, myths, urban legends, and magic bullet chasing out of my shooting and processing.

Stephen Benskin · May 3, 2014

I think this example from Simple Methods for Approximating the Fractional Gradient Speeds of Photographic Materials is a rather good illustration of the difference between the print judgment speeds, "S", 0.10 fixed density, "S`", and inertia speed thrown in for good measure, under different curve shapes and changes in gradient.

I should point out that the graphs are not implying that exposure should fall on S` or S. Only that it they are points where film speed can be determined. More on this later.

For the clincher on Zone System speed testing is Calibration Levels of Films and Exposure Devices by D. Connelly. Two important take aways are the value of "P", photographic constant, and "k1", ratio between P and the speed constant (also difference between Hg and Hm for B&W).

View attachment Calibration Levels of Films and Exposure Devices, Connelly.pdf

Bill Burk · May 3, 2014

Stephen Benskin said:
Basically, the print judgment speed is the most accurate method of determining film speed. It is based directly on the finished prints; however, it is incredibly laborious and therefore impractical to do on a regular basis. The next best method is to find a mechanical approach that will yield speeds similar to the print judgment speeds in the majority of cases. The fractional gradient method is that method. Fixed density methods were evaluated but weren't as consistently accurate. The fractional gradient method was determined by finding the spread between the proposed method and the speeds resulting from the print judgement speeds. Just like the spread example with testing alternative methods to the fractional gradient method. Of the three methods shown, the fixed density 0.10 method had the highest degree of spread and therefore the least consistently accurate to the print judgment speeds.

View attachment 87258

I want to point out a quick personal interpretation of this chart... It shows the relative ability of a few of the better speed determining methods... to predict whether the speed used for taking a picture... to predict if that picture will be excellent.

Notice the spread of the three better methods... A spread of 0.2 density units, or two-thirds of a stop. A way to guarantee a picture will be excellent... is to rate your film using any of the three good methods illustrated... AND THEN set your Exposure Index two-thirds of a stop lower than the speed you determined.

PE, I think you said you do that. I know I do that.

Xmas, I tend to always talk about Black and White negative... slides would of course require a different method to guarantee an excellent picture.

Bill Burk · May 3, 2014

Doremus Scudder said:
... N-2 development, that, unless I give substantially more exposure (along the lines of that suggested by the fixed-point speed) to compensate for the reduced development, my shadow values suffer.

Hi Doremus,

When you use N-2 to accommodate a long scale subject, your aim is still to print on Grade 2 paper. The shadows are suffering not for lack of any detail, but for lacking enough detail to be revealed on Grade 2 paper. So increasing exposure is a good idea for this case.

The Delta-X criterion justifies using the same speed according to the fractional gradient method, for a relatively normal scene where the underdevelopment is done for other purposes, say... to maintain high resolution and fine grain. Then the aim would be to print on Grade 4 paper, and the speed could be same and give a similar print as a normally developed negative of the same scene printed on Grade 2.

Stephen Benskin · May 3, 2014

Doremus Scudder said:
I mentioned earlier a disconnect (for me, anyway) between ISO and personal E.I. But, if I accept that, then it would seem that much of the exposure compensation that I have been doing for expansions and contractions would be unnecessary. This confuses me a bit, for it really seems to me that when I make a negative of a very contrasty scene that requires, say, N-2 development, that, unless I give substantially more exposure (along the lines of that suggested by the fixed-point speed) to compensate for the reduced development, my shadow values suffer. Conversely, expansion negatives are often overexposed (which doesn't bother me much) if I don't compensate by reducing exposure.

This is the disconnect I am addressing. And, I don't even really know where to start asking informed questions about it except to fire away and hope you will indulge me yet again. So, did the first excellent print tests include lots of expansion and contraction subjects, or were they more or less concerned with prints of subjects with more "normal" SBRs? If so, then should we expect much different results for similar print tests with subjects with significantly higher or lower SBRs? If not, then why does it seem that in practice I have to compensate exposure for contractions and expansions than the Delta X or fractional gradient speeds would suggest?

Since the effective film speed is based on 0.3x average gradient, you shouldn't expect to match the over all density of negatives with expansion and contraction development to that of a normal negative. The overall negative density will change with development, but the shadow gradient at the speed point remains at 0.3x average gradient (this is why the just black proof is valid only with normal development). Things tend to get more complicated at the extremes. The degree of flare is more of an uncertainty, as is preferred tone reproduction. We haven't even discussed developmental models yet. I'd like to show you a four quadrant example, but my program is down because of a conflict with Internet Explorer and I don't have an example of a N-2 development. I have one of a large subject luminance range with 2 stops flare and compensation with the paper grade, but that might just muddy the waters. Anecdotally, when I shoot scenes that I intend on progressing less than -1, I tend to error on the safe side.

One of my favorite Jones quotes comes from A Study of Various Sensitometric Criteria of Negative Film Speeds. This is the paper comparing various speed methods to the print judgement speeds. Jones writes, "From the standpoint of tone reproduction theory there seems to be no justification for the adoption of any value of density as a significant criterion of the speed of a photographic negative material. The primary function of the negative material is to record brightness differences existing in a scene. Density, per se, has no significance as an indication of the ability of the photographic material to perform this function. The value of negative density by which any particular object brightness is rendered, as, for instance, the deepest shadow, is of no consequence except insofar as it may have some bearing on the exposure time required to make a print from a negative.

Tone reproduction theory indicates that there is only one characteristic of the negative curve that is significant in expressing the capacity of the material to reproduce brightness (luminance) differences, and it is upon the way in which brightness differences are reproduced that the quality of the film positive must depend. This characteristic of the D-log E relationship to the gradient or slope, since this determines the magnitude of the density differences by which brightness differences in the object will be rendered in the negative and eventually in the positive made therefrom."

More specifically, I do have a development chart that uses both fixed density and Delta-X speed determination. In this example, I used a fixed flare model to determine aim contrast index. Fixed flare tends to higher and lower CIs faster as the luminance range moves out from the statistical normal then variable flare.

For a reference, here is a data from one of the images used in the first excellent print test.

To answer your question about terminology. EI or effective film speed seems to be commonly used for any method of determining speed outside of the ISO standard. I prefer effective film speed for sensitometrically tested film because EI has other uses.

Stephen Benskin · May 7, 2014

What's important to remember about speed points is that they don't necessarily represent where the exposure is supposed to fall. For the fractional gradient point, it is the 0.3x average gradient point and can be considered the minimum useful gradient point. The Delta-X criterion's and the ISO standard's use of the 0.10 fixed density point is in it's correlation to the fractional gradient point. The testing establishes the lower limitations of the film. Based on this point, any adjustment would be relative to this known quality limitation. Once the speed point is determined, the exposure required to produce that point is entered into an equation that contains a speed constant. The film speed value is part of the exposure meter's equation to determine exposure placement. There is a known relationship between the exposure at the speed point and the exposure at the metered exposure point. Adjusting the speed constant, changes the film speed and consequently changes the relationship between the speed point and the metered exposure point.

The speed equation for reversal film used to be 8 / HR, which is the metered exposure point. Now its 10 / HR. Thats a 1/3 stop adjustment without changing how the speed point is determined. The same could have been said for b&w film speed. They could have simply adjusted the film speed constant to eliminate the safety factor, but many people didnt like how difficult it was to determine the fractional gradient speed point. The only way to create a universal standard was to simplify the process. Thats where Delta-X comes in. It has good correlation with the fractional gradient method while using an easy to find fixed density.

Heres an example of the math. This is all built into the ISO standard when the ΔD is 0.80 at Δ 1.30 log-H, but if you want effective film speeds for any other development conditions, youll need to use the equations. Anyone recognize the ISO standard's contrast parameters in this example?

This chart gives the ΔX value for a given ΔD.

Stephen Benskin · May 7, 2014

The value of Delta-X is the difference between the fixed density point of 0.10 and the fractional gradient point which is the speed point. Referring to the Delta-X Table in the previous post, as the contrast (ΔD) is reduced, the value of ΔX increases and as the contrast is increased, the value of ΔX is decreased. The further to the left of the speed point, the faster the film is in relation to what it would be using only the fixed density method. What this means is that the film doesn't lose speed as quickly with reduced development and doesn't increase as significantly with expanded development compared to the fixed density method. With the Delta-X method, the speed point will remain at the same 0.3x average gradient at all levels of processing. It's an apples to apples comparison.

pharmboycu · May 8, 2014

This is a great, informative thread. I've read it and reread it a couple of times to try and follow it all. Now please bear in mind I'm not very well versed in the development procedures. That said, have I got the (admittedly oversimplified) generalizations correct?

1) E.I. is a personal choice based upon how the developed images correlate with the desired intent of the photographer whereas the ISO is the standard given as a starting reference by the manucaturer.

2) The photographer sets the film speed on the camera *at the ISO* and develops as though it were shot *at the E.I.*

Knowing that the E.I. is a personal choice of the photographer, for those of you who have gone through the process of determining your own E.I., may I ask what specific criteria you found useful in this process? At what point in your photography did you feel you'd learned enough to evaluate those criteria?

Thanks!

John

markbarendt · May 8, 2014

John, three criteria.

Can I get the print I want?

How easy is it to get that print?

How big is my safety factor? (How far can I over or under expose and get what I want?)

ic-racer · May 8, 2014

pharmboycu said:
Knowing that the E.I. is a personal choice of the photographer, for those of you who have gone through the process of determining your own E.I., may I ask what specific criteria you found useful in this process? At what point in your photography did you feel you'd learned enough to evaluate those criteria?

Thanks!

John

The work of Jones et al provides me with a good minimum point for the EI and I have no qualms his findings. I found the "W speed" as an estimation of 0.3G easier to calculate with a spreadsheet than either directly finding 0.3G or claculating Delta-X. So I use that [W-speed] when I do the math on my sensitometric data.

To link my sensitometric findings with a lens/shutter/camera/exposure meter system I use the 0.1log d ASA method with an 'in camera' Zone 1 exposure.

I also, have been using a 'minimum flare density of 0.1 log d' by placing a box or can with a hole in it in a typical composition. This provides a simple and practacle empiric solution to the issue of flare and its effect on film speed.

For my large format work I tend to use a generous safety factor to minimize exposure errors in the field. This is possible because there is very little or no image deterioration of the 8x10 negative image due to over exposure. Especially when considering the largest prints I make with that format are only 2.4x magnification ( 20x24").

I control contrast when I print (dichroic head) so that simplifies negative processing a thousand-fold

alanrockwood · Dec 25, 2014

Is there a place where I can get a pdf of the paper "Study of Various Sensitometric Criteria of Negative Film Speeds" by Jones and Nelson?

NickLimegrove · Dec 25, 2014

If you have internet access via a university network, you can download it directly (http://dx.doi.org/10.1364/JOSA.30.000093), or if you happen to live near an (analog) library, you'll probably be able to get a (digital) copy via their computers.

Bill Burk · Dec 25, 2014

alanrockwood said:
Is there a place where I can get a pdf of the paper "Study of Various Sensitometric Criteria of Negative Film Speeds" by Jones and Nelson?

I'd like to read it too, but I see that it takes a bit of cash to read that article...

What are you curious about? Maybe we've already discussed it or can get answers from someone on APUG (Stephen Benskin or PE may know the answer)...

Sirius Glass · Dec 25, 2014

alanrockwood said:
Is there a place where I can get a pdf of the paper "Study of Various Sensitometric Criteria of Negative Film Speeds" by Jones and Nelson?

I though about it too, but then I realized that I like what I get by using black & white and color at box speed, so why bother? Maybe if a copy were sitting around for me to read ...

alanrockwood · Dec 25, 2014

Bill Burk said:
I'd like to read it too, but I see that it takes a bit of cash to read that article...

What are you curious about? Maybe we've already discussed it or can get answers from someone on APUG (Stephen Benskin or PE may know the answer)...

It's not so much a single question as a desire to study the article.

I don't think our local university has the Journal of the American Optical Society, so that makes it harder to get a copy.

RattyMouse · Dec 25, 2014

alanrockwood said:
It's not so much a single question as a desire to study the article.

I don't think our local university has the Journal of the American Optical Society, so that makes it harder to get a copy.

What about inter-library loan?

Photo Engineer · Dec 25, 2014

There is a generic description of curve shape and speed in Haist.

PE

alanrockwood · Dec 26, 2014

alanrockwood said:
I don't think our local university has the Journal of the American Optical Society, so that makes it harder to get a copy.

I was wrong. The local University library had the article. I was able to log on and download it. It is a long article, and I have only skimmed through and not digested it yet, but one conclusion, which as also been mentioned in this thread, is that the authors recommend basing exposure on some form of gradient rather than some value of base above fog.

This sounds reasonable, but there is also one potential difficulty with this method. The gradient (sometimes called the first derivative) can be difficult to measure accurately because it is very sensitive to the quality of the data. Specifically, the first derivative is very sensitive to noise in the data. Noisy data (i.e. if the data points do not fall directly on a smooth curve, but instead bounce around a smooth curve in an irregular fashion) causes the calculated derivative to be very uncertain.

You can reduce noise by doing "signal averaging", which in this case means you would repeat the experiment many times and then take an average of all of the repeats. However, this is very costly because the improvement in the data varies with the square root of the number of repetitions. For example, if you want the uncertainty in the data to improve by a factor of two you need to repeat the experiment four times, and if you want to improve the data by a factor of ten then you need to repeat the experiment one hundred times.

Sirius Glass · Dec 26, 2014

alanrockwood said:
I was wrong. The local University library had the article. I was able to log on and download it. It is a long article, and I have only skimmed through and not digested it yet, but one conclusion, which as also been mentioned in this thread, is that the authors recommend basing exposure on some form of gradient rather than some value of base above fog.

This sounds reasonable, but there is also one potential difficulty with this method. The gradient (sometimes called the first derivative) can be difficult to measure accurately because it is very sensitive to the quality of the data. Specifically, the first derivative is very sensitive to noise in the data. Noisy data (i.e. if the data points do not fall directly on a smooth curve, but instead bounce around a smooth curve in an irregular fashion) causes the calculated derivative to be very uncertain.

You can reduce noise by doing "signal averaging", which in this case means you would repeat the experiment many times and then take an average of all of the repeats. However, this is very costly because the improvement in the data varies with the square root of the number of repetitions. For example, if you want the uncertainty in the data to improve by a factor of two you need to repeat the experiment four times, and if you want to improve the data by a factor of ten then you need to repeat the experiment one hundred times.

So the rest of use will use box speed which has worked in the past. I understand that photons have not changed the way they work since the start of the digital age.

Photo Engineer · Dec 26, 2014

Alan, fortunately negative films are built to give an average curve of gradient 0.6 in the mid scale when properly developed. At the same time, grade 2 papers have a mid scale gradient of 2.5 when properly developed.

Since prints are best at a gradient of about 1.5, and since print gradient = paper gradient X film gradient, then 0.6 x 2.5 = 1.5. Taking into account the flare of consumer camera lenses, the film is actually built to a gradient of 0.63 or so and to account for the flare in enlarging it is suggested that you use a grade 3 paper. Same thing holds true.

Now, while all of the above is true, the eye perceives the world at a gradient of 1.0 by definition since we are the observers. And, since there is a toe and shoulder in the print, this means that a gradient of 1.0 in the mid scale would be integrated over the entire curve, then we would perceive a 1.0 contrast print as being flat and so to compensate for the toe and shoulder, the overall contrast is best at 1.5. Some claim a contrast of 1.7 is better.

Please do not confuse contrast with speed in these arguments. Optimum true speed is reached at 0.6 mid scale slope and anything above that is just an increase in contrast, although many here would interpret that as being a speed increase.

One additional item. If you draw a straight line along the Dmin (fog) line of the X axis of an H&D curve and then one along this mid scale line of slope 0.6, then where they meet is often called the speed point. However, the true speed, the threshold speed is where the straight line following the X axis deviates from the real curve of the film. This point is where the grains see the first light and generally cannot be changed.

This was a quick summary of the article, or what I suspect was in the article having read many such over the years.

PE

Bill Burk · Dec 26, 2014

Thanks PE,

This is the kind of summary that is easy to understand.

Box ISO rate and Real ISO

Privacy & Transparency

Privacy & Transparency