Separation

[Michel Hardy-Vallée]

It's one of these terms that seem to be seldom defined, either because it's so simple that it's taken for granted or because it means something different for different people.

In AA's The Negative, there is a pair of photos of a lit lightbulb. The film for the first one was developed with a standard MQ-style developer; for the other, it was developed with pyrocatechin.

In the first one, the lightbulb is just textureless white; in the second one you can see clearly the details of the bulb, almost the filament. In visual terms, separation seem therefore to mean that you can distinguish different shades of gray in a specific area.

But what is "separation" in "highlight/shadow separation" when understood in terms of characteristic curve? Does it mean a flatter slope or a steeper slope?

When people speak of films with long toe (e.g. 320TXP), they often say that these films do not have good shadow separation, or muddy shadows. However, long-toe films are recommended for flash photography because of their highlight separation. From the Kodak documentation, 320TXP has a long toe with an upswept curve.

Straight-line films like 400TX are said to have better shadow separation, i.e. a steeper slope in the shadows region.

On the other hand, highlight separation, as exemplified by the above lightbulb case seem to mean a lower characteristic curve slope: if the curve were too steep, the density would rise too high too quickly, and then become unprintable.

Is "separation" then just a functional term relative to the context, visually meaningful only, but that can be realized differently according to circumstances?

 

[Ole]

In AA's The Negative, there is a pair of photos of a lit lightbulb. The film for the first one was developed with a standard MQ-style developer; for the other, it was developed with pyrocatechin.

The "other one" was developed in Windisch' Extreme Compensating Pyrocatechin developer - and unless I'm mistaken, the illustration was lifted directly from Hans Windisch' book.

It's not really a very good example of "separation", since the "first one" shows completely blocked-up highlights, and the other one shows the effects of an extreme compensating developer which gives very low contrast (and thus separation) throughout the range. I've used that developer myself for an extreme scene, a multiple exposure of a partial solar eclipse.


The way I understand it is that "separation" is directly related to the slope of the curve, so that the steeper the curve the better the separation. A typical "S-shaped" curve will give good midtone separation and less good highlight and shadow separation. But a shallow long-shoulder curve (like XP-2, or APX100 in Windisch) will give printable highlights where a film/developer combination with a steeper highlight curve would be completely unprintable - or at least a lot more difficult to print.

 

[Roger Hicks]

But what is "separation" in "highlight/shadow separation" when understood in terms of characteristic curve? Does it mean a flatter slope or a steeper slope?


A print on good paper has a log brightness range of about 2.2 (seven and a third stops). Of this, about 2.0 (six and two thirds stops) is 'dynamic' range with texture and detail; in Zonespeak, Zones 3 to 7 of the original 9-Zone range.

Subject brightness range can vary from 1:1 (plain tone) to 1,000,000:1 or more (landscape, sun in shot). Obviously 1:1 is irrelevant but below about 100:1 you can either 'stretch' the tonal range so that it fills the paper brightness range (by increasing development time) or just record it 'straight'. 'Stretching' is known as N+ development in Zonespeak.

At around 100:1 you can match the original subject brightness range, but beyond that you have a choice. You can discard highlight detail ('blown' highlights) or more usually shadow detail ('blocked' shadows); you can dodge and burn; or you can reduce development to reduce contrast and compress the representation of the subject brightness range in the negative and hence on the print. This is N- development in Zonespeak.

Modest compressions are scarcely noticeable, but the greater the compression for a pictorial subject, the greater the chance of its looking muddy and peculiar. Yes, compressions as far as N-6 can work; but equally, even N-2 can look flat and muddy, depending on the subject.

Now, as the film's recording curve is typically more or less S-shaped, separation is not constant. In the shadows (the thinnest parts of the negative, the toe of the D/log E curve) even quite large variations in subject brightness will lead to quite small variations in neg density and hence print density. An easy way around this is to give extra exposure so that you are not relying on the toe portion of the curve.

In the highlights, you eventually reach the shoulder where quite large variations in subject brightness once again produce minor variations in density. In normal use, you seldom use that part of the image brightness that lies on the shoulder, but you can force the shoulder down via compensating development.

To confuse matters, the eye's response to brightness is closer to logarithmic than linear. Whereas quite small variations in highlight detail are perceptible -- less than can be measured by most densitometers, if the variations meet at a line -- much bigger variations are needed in the shadows if they are to be apparent. A compensating developer may therefore give a more natural-looking highlight separation.

Depending on the subject brightness range, the use of a compensating developer may however result in compression of the mid-tones, i.e. less separation there. This is a matter for each photographer to choose for himself or herself.

You can choose where you want maximum separation, by choice of exposure and developer, but you can't have maximum separation everywhere unless the subject brightness range is either very small (under 2.0) or very uniform -- in other words, with no need for significant extra separation in any one of the shadows, mid-tones or highlights.

'Separation' is, therefore, an extremely complex interplay of several fairly simple ground rules, and I for one would not care to lay down any universal rules other than those based on first principles, as above.

Cheers,

R. (and you might care to look at the Photo School in www.rogerandfrances.com where we try to explain a number of fundamental concepts in as plain English as possible).

 

[noseoil]

Not to put too fine a point on it or muddy the waters, but to add to Roger and Ole's posts there is a variable which has nothing to do with the film. The paper's curve needs to "match" the film's curve. Since the paper's curve has a "toe" (it is the opposite of the film, or highlight values) like a film's curve, the two can be played against each other to make a given negative look different. By changing the paper type, a negative can be made to look entirely different in the print in its tonal range. tim

 

[Donald Miller]

Not to put too fine a point on it or muddy the waters, but to add to Roger and Ole's posts there is a variable which has nothing to do with the film. The paper's curve needs to "match" the film's curve. Since the paper's curve has a "toe" (it is the opposite of the film, or highlight values) like a film's curve, the two can be played against each other to make a given negative look different. By changing the paper type, a negative can be made to look entirely different in the print in its tonal range. tim

Tim raises a very valid point since separation is only partially derived from the film characteristics. The paper and it's characteristics are also contributory. Beyond that the process by which the paper is exposed is also contributory.

Contact prints typically exhibit greater separation than the same negative and paper when projection printed. I have found that a well designed optical system (read condenser) will also contribute to enhanced tonal separation.

The reason that I believe this to be true is that any means that will allow more perpendicular presentation of the light beams to the negative interface will provide better adjacency effects in the print. A point light source, in my personal experience, will contribute still more than a diffusion/condenser system.

Now before someone tries to interpert what I have said as being not clear enough, it might be that I mean just exactly what I have said. I have no interest in parsing terms in an attempt to elevate someones ego.

 

[Roger Hicks]

Not to put too fine a point on it or muddy the waters, but to add to Roger and Ole's posts there is a variable which has nothing to do with the film. The paper's curve needs to "match" the film's curve. Since the paper's curve has a "toe" (it is the opposite of the film, or highlight values) like a film's curve, the two can be played against each other to make a given negative look different. By changing the paper type, a negative can be made to look entirely different in the print in its tonal range. tim

Dear Tim,

Absolutely true! Thanks for the addition. This accounts (I am sure) for many 'magic' combinations of film and paper, either of which does not look as good in some other combination.

Cheers,

R.

 

[Roger Hicks]

Contact prints typically exhibit greater separation than the same negative and paper when projection printed. I have found that a well designed optical system (read condenser) will also contribute to enhanced tonal separation.

Dear Don,

I am inclined to agree with this, though in a sense it runs contrary to something else that seems also to be true, namely, that a diffuse source (especially cold cathode) seems to give tonality closer to that of a contact print.

I rather wonder how far this is related to the degree of enlargement, i.e, you get more separation at larger degrees with condensers, amd more separation at lesser magnifications with diffuse sources.

We may have an interesting point for discussion here, the difference between tonality and separation. Perhaps this is separation at boundaries (hard-edge separation) versus overall separation.

One small argument, though: surely you didn't mean to imply that well designed systems necessarily use condensers, and that there are no well designed cold cathode or other dffusion systems? Your post could be read that way.

If I'm still on your ignore list, perhaps someone else would care to respond to or copy this.

Cheers,

R.

 

[Michel Hardy-Vallée]

Alright, those were very substantial answer to a vague question, so thank you all! and I'll add a few comments.

Ole: That's what I wondered on second thought. Windisch's developer is about about controlling extremes of contrast in order to have an interval of printable densities. It would be then equivalent to reducing the slope of the highlights portion of the curve so that it fit on paper's brightness range. So, to echo noiseoil, separation is relative to the whole printing system.

Roger: thank you so much for your patient answer, and I've also read all the free modules on your site some time ago. . What you're saying is consistent with what I've seen and read, and puts it all together cogently.

I think one could sum up a "separation" principle like this: a given portion of a film curve, if it is contained on the straight line of a paper curve, will generally show more tonal separation if its slope is steeper, and vice versa. Corollary: increased separation in one portion of the curve can bring decreased separation in another portion of the curve, depending on the nature of the film curve.

I think somebody ought to write a quadrant tracing software because I realize that all these explanations depend on the ability to manipulate matched paper/curves.

 

[Jean Noire]

Alright, those were very substantial answer to a vague question, so thank you all! and I'll add a few comments.



I think somebody ought to write a quadrant tracing software because I realize that all these explanations depend on the ability to manipulate matched paper/curves.

If I remember rightly there used to be something called "windmill diagrams" that did as you asked. Not software though.
Can't remember too much about them but I will look it up if I can find the book, or maybe Roger can remember these?

Regards
John

 

[Michel Hardy-Vallée]

If I remember rightly there used to be something called "windmill diagrams" that did as you asked. Not software though.
Can't remember too much about them but I will look it up if I can find the book, or maybe Roger can remember these?

Regards
John

I know Kodak calls them quadrant diagrams, and so does Mees, IIRC. You split up a spreadsheet in four quadrants: lens flare curve; film curve; paper curve; and tonal rendition curve. The last one is essentially the difference between an "ideal" rendition (no compression/expansion, just all straight 45 deg lines) and actual rendition. With that method and statistical subjective evaluation tests, it was found that the most pleasing prints were not those with the most perfectly equal tonal rendition (100% straight line), but those that show selective compression and expansion at key positions.

If you can find a copy of the Kodak Encyclopedia of Practical Photography, there's a great article called "Tone reproduction" that explains in details what I've outlined. It's so good I think I'll go photocopy it some day for my archives.

 

[Jean Noire]

I know Kodak calls them quadrant diagrams, and so does Mees, IIRC. You split up a spreadsheet in four quadrants: lens flare curve; film curve; paper curve; and tonal rendition curve. The last one is essentially the difference between an "ideal" rendition (no compression/expansion, just all straight 45 deg lines) and actual rendition. With that method and statistical subjective evaluation tests, it was found that the most pleasing prints were not those with the most perfectly equal tonal rendition (100% straight line), but those that show selective compression and expansion at key positions.

If you can find a copy of the Kodak Encyclopedia of Practical Photography, there's a great article called "Tone reproduction" that explains in details what I've outlined. It's so good I think I'll go photocopy it some day for my archives.

This is the same thing, as you know. This is in line with my book.
Can't add anymore to it.

Regards
John

 

[Roger Hicks]

'Windmill diagrams' appear in numerous books from the 50s in particular. The one I found quickest was Dunn, Exposure Meters and Practical Exposure Control, 1952 The Fountain Press, pp. 42-43. The first quadrant is subject brightness > negative curve; the second, a transfer quadrant (straight line); the third, the print curve; the fourth, the overall reproduction produced by quadrants 1+3.

They are not particularly easy to understand intuitively, and they are even harder to follow visually, but they demonstrate the whole concept very well indeed.

Cheers,

R.

 

[Jean Noire]

'Windmill diagrams' appear in numerous books from the 50s in particular. The one I found quickest was Dunn, Exposure Meters and Practical Exposure Control, 1952 The Fountain Press, pp. 42-43. The first quadrant is subject brightness > negative curve; the second, a transfer quadrant (straight line); the third, the print curve; the fourth, the overall reproduction produced by quadrants 1+3.

They are not particularly easy to understand intuitively, and they are even harder to follow visually, but they demonstrate the whole concept very well indeed.

Cheers,

R.

I did not think that anyone else could have this book!
I have the Dunn and Wakefield 1981ed. My bible!
Regards
John

 

[Roger Hicks]

I did not think that anyone else could have this book!
I have the Dunn and Wakefield 1981ed. My bible!
Regards
John

Dear John,

Dunn and Wakefield is even better than Dunn alone. This was the inspiration for my 'Perfect Exposure' (details on www.rogerandfrances.com, of course). Since D&W, as far as I could see, no-one had done a serious exposure book aimed at a popular market. In fact, astonishingly few books take a close look at the theory behind exposure. Usually, they are partisans for a particular System, or frankly fluffy. Some are both!

Cheers,

R.