Critical Thinking 2 - Normal Negative Density Range - ZS vs Tone Reproduction

[squinonescolon]

I think Nicholas has it right. A couple of weeks ago I went to see the Henri Cartier-Bresson show at SFMOMA. If you look at the images for their technical aspects—and I know this is a sacrilege to say—they're not that good. Most of the pictures are out of focus, and tones are very compressed for most of the images. But when you look at the images from a pictorial frame of mind, which is how everyone should be looking at them, of course they come alive.

The composition is always superb; and an easy overlooked aspect is that they're printed at the right size. Most images are of varied sizes. Not 8x10 and 11x14. Which to me means that either him or his printer actively sought the "right" size for each image. Which is easily overlooked. The size of the image—I think—tells you the context of the subject and helps to reinforce the composition and the scope of the image. Is this an intimate portrait, a grand landscape, a frenetic snapshot? Size and composition inform that greatly, and few knew how to exploit that like HCB.

 

[ic-racer]

Sounds like you're right on track now. You scared me for a moment. I knew you knew this stuff.

Ok, I was getting confused with the other thead on multigrade printing.

 

[ic-racer]

One also needs to define what one is testng. That is, 'system testing' (that would incorporate the flare) vs 'material testing.' And to not confuse them or us the wrong one. For example it would be incorrect to state some unique property of a paper or film based on system testing. If one wants to test materials (ie for side-by-side comparison) one needs to eliminate the sytem variables (flare etc).

 

[ic-racer]

Lots of people. Lots and lots of people.

Get thee an enlarging meter - a proper enlarging meter [cough] (can't believe I haven't converted you yet, Mr. Racer):.

I agree, that method has to be the gold standard for analyzing a negative for an a priori paper match.

 

[Nicholas Lindan]

... an a priori paper match.

As opposed to the norm on APUG, the a posteriori match ...

 

[Chuck_P]

We don't disagree on this. I can't see how you think we do.

This makes me very happy, seriously.

 

[Chuck_P]

The shape of HD curves vary all over the place. To try and characterize a given curve with one number may have been state-of-the-art in 1929 but doesn't, to my mind, cut it anymore.

The HD curve is a continuum of points. It can not be mathematically defined, nor should it be. It is best to work with the raw curve as it lays.

I agree with this completely. ZS methodology (agree with it or not) does not consider CI, gamma, or mean gradient when evaluating characteristic curves. It considers the totality of the curve at all points. But when the curve is very linear then gamma would probably be a meaningful number. My TMX and d-76 1:1 curves are very linear, but I don't need to calculate slope for it to be meaningful to me in any way.

 

[Stephen Benskin]

The first problem with all this methodology is that it tries to boil tone reproduction down to one number - ISO contrast/LER/CI/...

Well, that's kind of an overstatement and oversimplification. The numbers aren't intended to be all encompassing explanations and it would be disingenuous to suggest otherwise. But they do carry with them information beyond just the numbers. There's the testing conditions under which they were determined as well as the supporting theory behind the methodology. Something like ISO speed carries with it the entire testing parameters defined in the standard, which are supported by numerous papers and studies. Even with all of what's behind the numbers, they just represent one or two tools in the toolbox.

Is anyone claiming the methods are perfect? Of course not. The short comings as well as the strengths and reasons for one method over another are covered in various papers. In "The Control of Photographic Printing by Measured Characteristics of the Negative" by Loyd Jones for example, Jones found there wasn't a single criteria that gives a high correlation in all cases between the optimum quality obtainable in a print and negative parameters. In an uncharacteristic manor he laments that "it might be felt at this point that the situation is hopeless and that the negative density range criterion for choosing the paper contrast grade should be abandoned. However, no other unique and measurable negative characteristic has been discovered which is useful as a substitute criteria." He goes on to say that it's not really as bad as it first sounds.

The shape of HD curves vary all over the place. To try and characterize a given curve with one number may have been state-of-the-art in 1929 but doesn't, to my mind, cut it anymore.

More of the perfect being the enemy of the good argument. If not, then what?

The HD curve is a continuum of points. It can not be mathematically defined, nor should it be. It is best to work with the raw curve as it lays.

The curve itself is a mathematical construct. Ultimately there will always be a gap between the science and the art, but there's psychophysics in subjective tone reproduction which helps define how people will generally respond to given conditions.

The second problem is the insistence on using the full range of tones on either the negative or paper. The pictures I have that use the full tonal and informational range look just like drugstore prints, high quality drugstore prints I like to think, but boring dull snapshots all the same unless you are interested in Great Aunt Edna or what Timmy looked like at age 5.

The insistence is strictly from pop culture and not from photographic theory. There's a very interesting explanation in Theory of the Photographic Process 3rd edition on and around page 489. It basically supports your point, but without the strawman.

There's a lot of information that can be gained from what ic-racer calls "system testing" (nice term BTW), especially when it's about the interaction between the various elements (original subject, camera image, negative, print, reproduction). Will it tell me if my photograph will likely hang in a gallery? No, but it will help to better explain how the photographic process works if that's what you're interested in, but it's just one aspect of photography. It's not a panacea.

Using the four quadrant reproduction curve, I'm able to explore many different scenarios and test different points of theory such as 18% or 12%, speed point and flare, and Delta-X speed. No more "light meters read 18%" crap for me. Curves are basically meaningless if you don't interpret them.

I'm also working on different ways to compare and evaluate the data. The different attachments show the different gradients for each of the steps in the guidelines, a more detailed breakdown for the film quadrant, and a breakdown of the values in the camera image quadrant. Bit by bit it all adds up to a better understanding of the photographic process which might just make it slightly easier to obtain the creative results one desires.

 

[Stephen Benskin]

Getting back to the topic of the thread which is about a specific point of theory - correctly defining and interpreting the testing procedures and results using as an example the seeming discrepancy of negative density aim values between the Zone System and Tone Reproduction methodologies. While the two testing methods described in the first message produces the same results and are both relatively valid, because the ZS method doesn't properly define the conditions of the test, it can lead to misinterpretations. The best testing is worthless if the analysis is bad.

Attached are five examples from five different books: The Negative, Ansel Adam's Guide: The Techniques of Photography, Beyond the Zone System, Way Beyond Monochrome, and Photographic Materials and Processes. Only one correctly illustrates the distribution of illuminance in relation to the film curve - Photographic Materials and Processes. All the other examples have the exposure along the x-axis at equal intervals. For a one stop flare factor, this means that they are incorrectly indicating a negative density one stop beyond what would actually result in usage. To me, an important aspect to testing and evaluation is to assure it reflects reality as closely as possible. In four out of these five examples, it does not.

 

[Stephen Benskin]

It's possible to reconcile the numeric discrepancy between the two methods using nothing more than Zone System information from The Negative and a little analysis.

 

[Usagi]

How about doing system calibration by photographing test subject - step tablet, gray cards or something.
That should include the camera flare already and thus camera flare could be practically ignored (ofcourse different lenses makes also difference).

 

[Stephen Benskin]

How about doing system calibration by photographing test subject - step tablet, gray cards or something.
That should include the camera flare already and thus camera flare could be practically ignored (ofcourse different lenses makes also difference).

Too many variables. You couldn't be sure what to attribute the results to and that isn't good testing.

Flare is difficult to control. Shooting a simple one toned test target like with the Zone System won't factor in flare. Eighty percent of veiling flare comes from the subject, the range of tones, and their distribution. A single toned test subject has almost zero flare.

Even if a multi-tonal test subject is chosen, flare is difficult to measure. You basically have to measure the subject luminance range and the subject illuminance range at the film plane. Not too many people have the required equipment. Without knowing the input and output values, what can you know? Many who do ZS testing with a camera erroneously believe their testing incorporates flare simply because they are using an optical system. They don't know the difference because they can't test for it, which will adversely effect the interpretation of the aim film contrast and film speed.

It's best to do a test for the film, then a separate test for flare. Phil Davis has a black box method in BTZS. You make a box with a snoot around an opening and place black velvet inside. Meter a gray card and shoot a scene or test target that includes the box. I used test targets that had the center cut out which were placed over the box opening. The targets were white, gray, black, and checkerboard. A sensitometric exposure was made and everything was processed together. I've attached the results. BTW, it's also a good example of flare and film speed (both examples are actually), but that's a topic for a different thread.

There's another factor to consider with single toned tests. Even if you stipulate a normal flare range for the test, the exposure is made at the point on the image curve where little flare exists. For average flare conditions, the flare at the metered exposure point is only 3% of that in the shadows. For ZS type testing, the target is metered and then the exposure is stopped down. The tone is still from the metered position. It's just placed lower on the film curve, not lower on the camera image curve. I've attached an example to illustrate this.

 

[Chuck_P]

There's another factor to consider with single toned tests. Even if you stipulate a normal flare range for the test, the exposure is made at the point on the image curve where little flare exists..............For ZS type testing, the target is metered and then the exposure is stopped down. The tone is still from the metered position. It's just placed lower on the film curve, not lower on the camera image curve. I've attached an example to illustrate this.

Doing the testing as described in The Negative instructs one to stop down 4 stops.

When exposing the step wedge in the film holder to a single toned target, the exposure is not stoped down 4 stops, its exposed at Zone X. Most people that test this way, do not use the actual method in the AA's text--it's more time cosuming and uses a lot more film and chems.

Are those areas of the step wedge that produce those low density values still not being influenced by any flare given the exposure is at Zone X and not Zone I? Wouldn't using a white target (even though it is single toned) produce flare and if so, wouldn't the flare then be factored into the speed point once the EI that produces a 0.1 density at Zone I is determined?

 

[Stephen Benskin]

Exposing the step tablet in camera won't introduce any flare to the test because the step tablet is contacted. The interpreting of the data obtained from a contact by definition excludes flare. Flare, therefore, must be factored in.

There also seems to be some confusion between exposure and the camera image. Flare is create from the interaction of the subject and the camera. It is independent of exposure. As you can see from the examples, changes in exposure simply moves the whole curve up and down. The relationship of the camera image to the original subject doesn't change. I've also included an example using scene reflection density for better clarification.

 

[Stephen Benskin]

Chuck,

I know you use the Schaefer testing method, so you might find this interesting. The exposure value from the above example for Zone X and/or Zone VIII + two stops exposure is the key to evaluating its accuracy. The exposure value represents the illuminance falling on the step tablet. You already know the density of the steps on the tablet. All you need to do is calculate the transmitted light at each step in the tablet. We know that film speed is calculated based on the exposure where the film curve density is 0.10 above Fb+f - 0.8 / Hm. Schaefer says that the Zone I exposure will fall 9 stops down from the Zone X exposure which correlates at 2.70 step tablet density. The example above of 2.032 is the exposure for a 125 speed film. That means you are looking for an Hm to equal 0.0064 (0.8 /125 = 0.0064). If that happens at 2.70, his exposure technique is accurate.

 

[Bill Burk]

Is it possible that there would be "no" flare in an evenly illuminated single-toned scene? If so, and if you normally assume all camera tests include flare, then Flare as an "uncontrollable variable" gets even more irritating.

In a camera test with a taped-on step tablet there would be no flare but there could be light fall-off at the corners.

Kodak Papers data book from 1955 bases a discussion on the density of the negative at the exposure plane. Readings at the exposure plane would include both the Callier and Flare effects. Ansel Adams specified 1.2 but his description makes me believe he used a standalone densitometer which would NOT include Flare. The Kodak data book says Flare could reduce the effective scale range 10%. It also recommends selecting a paper with about 0.2 greater scale range than the negative because in practice you will aim for 90% of the paper's dMax.

Thus, while your earlier conclusion that Flare is the difference between Ansel Adams and Sensitometry may be correct, but I think you were drawing in camera flare, while it looks to me that enlarger flare is where the difference creeps in.

 

[Stephen Benskin]

Is it possible that there would be "no" flare in an evenly illuminated single-toned scene? If so, and if you normally assume all camera tests include flare, then Flare as an "uncontrollable variable" gets even more irritating.

Bingo. What is worse, knowing a difficult to control variable exists or not knowing it exists?

In a camera test with a taped-on step tablet there would be no flare but there could be light fall-off at the corners.

Totally agree. I've made this very point before. It's one of my primary concerns with this method. Part of the equation for exposure meters factors in a 12% angle of image point from axis of lens. I'm not into optics, but I'm sure someone out there knows the average light loss as you move from the center of the image circle outward toward the edge.

The Kodak data book says Flare could reduce the effective scale range 10%. It also recommends selecting a paper with about 0.2 greater scale range than the negative because in practice you will aim for 90% of the paper's dMax.

What Kodak data book is this from? This is something I have to check out.

Thus, while your earlier conclusion that Flare is the difference between Ansel Adams and Sensitometry may be correct, but I think you were drawing in camera flare, while it looks to me that enlarger flare is where the difference creeps in.

I have to disagree with you on this one. It's less of a variable because of the control and consistency of the processes. Negative to negative will introduce slight differences, but the working conditions are usually consistent, and because of controlled processing of the negatives, you simply don't encounter the range of variance with the negative as you would when shooting exterior scenes. However, I'd like to hear more of your thoughts on this.

I'm curious about the 10% change you reference. That has to refer to enlarging? Average camera flare is one stop or 100% increase in the shadows and uncoated lenses are two stops.

Flare makes it impossible to be precise about exposure or effective luminance range. Really all you can do is use an average and adjust in printing. Except even that introduces difficulties.

Take for instance determining a model for development. Average flare increases as the luminance range increases and decreases as the luminance range decreases. Off the top of my head, the CI for a fixed flare model at +2 is the same as the CI for a variable flare model at +3. How flare is handled and interpreted makes a difference at this point.

 

[Bill Burk]

I think you'd be able to see my point if I was a little more clear. Where you thought Flare could explain why the different aims between the Zone System and Sensitometry arrive at the same results. I wanted to agree but point out that maybe it is Enlarging Flare, not Camera Flare. Then I gave a few reasons to illustrate.

Enlarging flare's effect on density scale from Kodak Data Book, Kodak Papers, Sixth Edition, First 1955 Printing.

Relation of Negative to Printing Paper.

"In a diffuse enlarger, flare light will lower the effective density scale by 10 percent or more."

While flare impacts shadows on the negative, in the print you lose highlights.

One of Nicholas' meters would determine the density scale including Enlarger Flare, giving you directly applicable results. Your full four-quadrant system would also include Enlarger Flare (by reading the print).

But a direct-reading densitometer won't include Enlarger Flare. You would need to pick different aim points to arrive at the same results. And you would need those different aim points because of Enlarger Flare.

 

[Stephen Benskin]

I agree with everything except how it applies to the discrepancy problem. Both the Zone System and Tone Reproduction numbers are for a diffusion enlarger. It works the same for a condenser. The end points are the same and the negative is the same. The key difference is the interpretation of the negative data.

Now, if you want to talk about another discrepancy between the two systems and how it too can also be resolved with the inclusion of flare? Take a quick look where the shadow exposure of the non-flare curve falls compared to the flare curve (and the 2.10 luminance range in the example is shorter than the 2.20 of the standard model).

Anyone notice Zone V (Four Quad example post #33) is at a RD of 0.92 or 12%? And the metered exposure value for the 125 speed film is 0.068? And where the metered exposure (Zone V) falls on the film curve is 1.00 log-H units (3 1/3 stops) above 0.10 and not 1.20 log-H units (4 stops)? And did anyone notice that even though Zone V = 12%, in the print it is close to 18%? Okay, the last one isn't about flare but it is interesting with interesting implications.

 

[Bill Burk]

So your comparison includes a Zone System camera test that includes flare, which is common enough.

I also wonder about the 2/3 stop discrepancy you hinted at. One of the thoughts I have is that some Zone System methods (maybe not the one you illustrated) have you find the first noticeable differences at Zone I and the last noticeable difference at Zone VIII. I noticed your "Zones" are centered on the endpoints, but maybe the endpoints that people find are the beginning of Zone I to the end of Zone VIII.

 

[Stephen Benskin]

I also wonder about the 2/3 stop discrepancy you hinted at. One of the thoughts I have is that some Zone System methods (maybe not the one you illustrated) have you find the first noticeable differences at Zone I and the last noticeable difference at Zone VIII. I noticed your "Zones" are centered on the endpoints, but maybe the endpoints that people find are the beginning of Zone I to the end of Zone VIII.

I gotta admit I don't like talking in Zones because they are arbitrary. I usually use reflection density for the scene. The four quadrant examples below use the version I personally like to use. It tells you more about the subject and it has a direct relation to the exposure equation. But to answer your question, the reference guidelines are in the center of the Zone range. Beginning or end, it's about range. Take a look at the Guideline Data chart which has the values of all the points and the resulting reproduction gradients.

The other three graphs illustrate flare and speed testing. The two quadrant example was for an article on the subject. The other two are quick examples from my new program. Both have the metered exposure point fall at the same place on the film curve. The difference between where the shadows fall is dependent on the lack or presence of flare. I want to add, that these two examples have been simplified slightly with a luminance range of 2.10 and not the standard model's 2.20.

As you know, the meter doesn't read percentages. It measures luminance and wants to produce an exposure of 8 mcs (that's the E in E * t = H equation). E * the shutter speed equals the exposure at the film plane or H. I like to use 125 speeds because it is the only film speed that has an equivalent shutter speed. So for a 125 speed film, the metered exposure (Hg) is 8 * 1/125 = 0.064. As you can see that is the value in the non-flare curve. Black and white film speed is calculated using the equation 0.8 / Hm. To find what Hm should be for any given film speed 0.8 / EI or 0.8 / 125 = .0064. As you can see the difference between these two points is 1.0 log-H or 3 1/3 stops. These more on this in the Sunny 16 attachment.

BTW, the whole middle gray dilemma can also be solved by factoring in flare.

The average luminance range falls 4 1/3 stops below the metered exposure and with the Zone System, Zone I falls 3 stops below. So why is the film speed measured 3 1/3 stops down? Simple answer - flare. Flare brings the shadow exposure up. The standard model uses one stop flare, but in practice is more likely to be around 1 1/4 to 1 1/3 stops.

So if you do a Zone System test, which doesn't incorporate much flare into the test, and stop down four stops, the point where the exposure falls is 2/3 stops below the speed point exposure. That is why so many ZS practitioners find their "personal" film speeds to be 1/2 to 1 stop slower than the film's ISO speed.

 

[Chuck_P]

The average luminance range falls 4 1/3 stops below the metered exposure and with the Zone System, Zone I falls 3 stops below. So why is the film speed measured 3 1/3 stops down? Simple answer - flare. Flare brings the shadow exposure up. The standard model uses one stop flare, but in practice is more likely to be around 1 1/4 to 1 1/3 stops.

The first curve is a speed test that produced a log exposure of .11 (the range of Zone I densities in the ZS is 0.09 - 0.11) at Zone I. Since it is a speed test, Kodak's ISO speed of 100 was used. I'm getting full box speed with TMX and D-76 1:1. You can see the curve reaches a 0.1 negative density at Zone I, not at Zone I 2/3 or Zone II. Flare? Perhaps it is incorporated into the speed point, IDK, but I know what you think .

However, the second curve is a TMX and HC-110 1:63 speed test that produced a log exposure of .07 at Zone I. But it can be seen that the curve required 2/3 stop additional log exposure to reach the 0.1 neg density line at Zone I 2/3. You can see that after drawing the curve, the line segment between Zone I and Zone II crosses the 0.1 negative density line at Zone I 2/3. So the EI for this film/dev combo is shaved down to EI64, which is 2/3 stop off the box speed. Is it flare? Or is it just the response of TMX to HC-110 diluted to 1:63? I know what I think.

These were exposed in the camera on the same day, same lighting conditions (on the shady side of my house on a cloudless sky), same target, same lens, same spot meter about 15 minutes apart.

AA contends that flare is accounted for, IDK, really. Am I drinking the cool-aid, I'm sure many will say yes. Perhaps the flare is just not as much of factor in my system as some may think it should be, IDK.

P.S.
To anyong interested:
The ultimate "N" development time for the D-76 curve ended up being reduced to 8:00 to get the curve to reach Zone VIII at a neg density of 1.3. I made one additional exposure of the step wedge at my tested speed (ISO 100) and reduced the development by 1.5 minutes, redrew the curve for this new sheet and it reached my target density for Zone VIII-----this established my "N" dev time for this combination of film/developer.


The "N" development time for the H-110 curve ended up remaining the same at 11:00----this is becuause after shifing the density point at each zone 2/3 of a stop to the left on the log E scale (to account for the reduction in speed to EI 64) and then re-drawing the curve, it then reached Zone VIII nicely at a density of 1.3-----In instances like this, there is no need to redo the test, it's a simple matter of shifting each point the apppropriate distance on the Log Exp scale to reflect the curve at the new EI. This is the method taught by Schaefer and one that is under the gun here in that it is not believed that it represents "reality". I contend that the "reality" is simply not that great of a difference.

 

[ic-racer]

Am I drinking the cool-aid, I'm sure many will say yes. Perhaps the flare is just not as much of factor in my system as some may think it should be, IDK.

If you want, you can compare your film curves from the photographed reflection wedge with a curve generated by a transmission step wedge taped on the film in the camera. The amount of flare in your system, under those conditions, will then be apparent.

 

[ic-racer]

Flare is not image forming light. Its presence is irrelevant to the separation of values in the image. Increases film density of the low values, but it DOES NOT alter the separation of low scene values.*

If you "account" for flare in an exposure index test and then there is no flare in your scene, nothing bad happens. Likewise, if you don't account for flare in your exposure index test and flare is present on the film plane, nothing bad happens.

(*The potential of flare to act as 'pre sensitization' light and affect the reciprocity characteristics of the film are being ignored in that statement)

 

[Chuck_P]

If you want, you can compare your film curves from the photographed reflection wedge with a curve generated by a transmission step wedge taped on the film in the camera. The amount of flare in your system, under those conditions, will then be apparent.

These curves are generated from a step wedge on the film in the camera. I was responding to Mr. Benskin's claim that film that ZS practitioners film speeds are 1/2 to 1 stop slower than the stated ISO speed because of flare interaction between the subject and the camera that he believes is not accounted for on the film image in ZS testing.