Probably a dumb question about graphs

[warden]

Why does Ilford graph their exposure from 0 to 4.5, and Kodak uses -3.5 to +1.0?

They seem to be showing similar information with differing zero points on the horizontal axis, but why? Is one "correct"? I'm trying to compare films and this isn't helping.

 

[AgX]

Because one uses a relative scale and the other an absolute scale (w. unit).

 

[Stephen Benskin]

log exposure vs relative exposure or actual vs arbitrary.

 

[RalphLambrecht]

Why does Ilford graph their exposure from 0 to 4.5, and Kodak uses -3.5 to +1.0?

They seem to be showing similar information with differing zero points on the horizontal axis, but why? Is one "correct"? I'm trying to compare films and this isn't helping.





View attachment 296424 View attachment 296425

Both are correct on different scales. Ultimatum noticed that they both plot density over 15-stop range. So they can be compared with admittedly it's a bit confusing at first.

 

[warden]

Because one uses a relative scale and the other an absolute scale (w. unit).

Ok that helps, but Ilford's graph is "relative" to what? I'm confused.

 

[warden]

Here's what I'm doing. I took three graphs from the film manufacturer's data sheets for films that I like, and superimposed them, just to see the differences. (They each have 4.5 units on the X axis and 3.0 units on the Y axis.) The result is this image:

One of those three lines is Tmax 400. One is Plus-X. The final one is Delta 400. I aligned the scales so that they match, but it brought to light the question of the units of the horizontal scale. I'm not sure if combining these graphs shows anything of value, because of the horizontal scale.

 

[warden]

Here's another way to look at it, highliting the matching Y axis and possible mismatch of the X axis. This is Plus-X and Delta 400 superimposed. Delta 400 is red. Plus X is greyed. Because the manufacturers graph differently I'm not sure how much relevance there is in comparing brand to brand.

.

 

[reddesert]

Kodak's graph is in actual units of log exposure in lux-seconds, and has minus signs (because one lux second is a significant amount of light). One unit in log space means 10x as much exposure.

Ilford thought the minus signs might be confusing so they added a constant, meaning their graph is in log(relative exposure), not in a specific unit like log (lux-seconds).

For amateur photographic purposes they're the same thing, just shifted by a constant. You only have to worry about the relative/absolute difference if you're trying to do calibrated absolute photometry.

Different films, in different developers, will have slightly different curves. To align the curves you would need to pick a standard density to align them (like "0.1 above film base + fog"). Then you would have to address the question of whether these curves are useful to you, or should you get a densitometer and measure your film, the way you process it, because your curve may be a little different from the manufacturer's standard.

 

[warden]

Kodak's graph is in actual units of log exposure in lux-seconds, and has minus signs (because one lux second is a significant amount of light). One unit in log space means 10x as much exposure.

Ilford thought the minus signs might be confusing so they added a constant, meaning their graph is in log(relative exposure), not in a specific unit like log (lux-seconds).

For amateur photographic purposes they're the same thing, just shifted by a constant. You only have to worry about the relative/absolute difference if you're trying to do calibrated absolute photometry.

Different films, in different developers, will have slightly different curves. To align the curves you would need to pick a standard density to align them (like "0.1 above film base + fog"). Then you would have to address the question of whether these curves are useful to you, or should you get a densitometer and measure your film, the way you process it, because your curve may be a little different from the manufacturer's standard.

Thank you, I appreciate the explanation. It sounds like there isn't much to be gained from my little experiment.

Ilford was right about the confusion issue, at least for me! ;-)

 

[AgX]

Ok that helps, but Ilford's graph is "relative" to what? I'm confused.

Each mark relative to the others, all differing by same value.

To make things more easy to read, Ilford deliberately put their mark "1" at the exposure that lifts the curve just by the decisive level above minimum density. And from there you go with the other marks.

 

[warden]

As long as the scaling on both axes are aligned (ie the intervals are the same size), and the absolute numbers on the y-axis (density) are the same, then you can compare the shapes of the curves of different films developed in different developers, which is useful. What you can’t technically compare, are the emulsion speeds. In order to do that, the absolute log exposure numbers have to be the same.

So, this means since Kodak tends to plot absolute log exposure on the x-axis, you can superimpose the curves for Kodak films and compare curve shapes and emulsion speeds. Since Ilford only plots relative log exposure on the x-axis, superimposing curves for Ilford films, or superimposing Ilford and Kodak curves allows you to compare curve shapes but not emulsion speeds.

In terms of practical utility, the ability to compare emulsion speeds is less important in the case of both Ilford and Kodak curves since we know the ISO speeds of the films. In other words you don’t learn a whole lot about the emulsion speeds of these films by looking at the graphs anyway. With films of known/given emulsion speed, comparing the shapes of the curves is the useful part of superimposing them, so as long as the scaling (interval size) is the same on both axes, and the absolute density numbers are the same on the y-axis, you’re good. You can do what you did, and slide the curves back and forth along the x-axis to align them closely enough to get a visual sense of differences in curve shape. For example, does one film have a longer toe than another. Is one film more s-shaped, etc.

That was very helpful Michael, thank you. And you are describing exactly what I am trying to accomplish: make the scales match, superimpose two characteristic curves, and see which is steeper in the midtones, which has the longer toe, which has no shoulder, etc. That's easy to do if you stick to one brand only, but I've read comments like, "Delta 400 is a good substitute for Plus-X", and to check that out it would be nice to see the curves for those two films on top of each other.

 

[mrosenlof]

Not a dumb question, a perfectly good one!

Ilford's graph doesn't state units like Kodak's does. So the exposures it relates are relative to the far left of the graph, which could be labelled '0' on the X (horizontal) axis. Ilford's graph is also logarithmic (but unstated), so if the far left of the graph is one unit the 1 tic is 10 units, the 2 tic is 100, and 1000, 10000 successively.

The origin of Kodak's graph is "ten to the minus 3.5 power" in lux seconds which might be a little different from Ilford's "one unit" of exposure. To best compare, you probably want to match up the middle straight line parts, or match the curves at a low exposure point where they give approximately the same density.

I hope that helps!

 

[warden]

Not a dumb question, a perfectly good one!

Ilford's graph doesn't state units like Kodak's does. So the exposures it relates are relative to the far left of the graph, which could be labelled '0' on the X (horizontal) axis. Ilford's graph is also logarithmic (but unstated), so if the far left of the graph is one unit the 1 tic is 10 units, the 2 tic is 100, and 1000, 10000 successively.

The origin of Kodak's graph is "ten to the minus 3.5 power" in lux seconds which might be a little different from Ilford's "one unit" of exposure. To best compare, you probably want to match up the middle straight line parts, or match the curves at a low exposure point where they give approximately the same density.

I hope that helps!

Thanks for the extra input, I appreciate it.

In this image I think I've done what you and Michael suggested, by moving the Plus-X curve to the left so it takes off at about the same point as the red Delta 400 line. And this does seem to show that the characteristic curves are pretty similar between the two, at least with the developers, temperatures, and times that the respective companies were using to make these curves..

 

[radiant]

I still rant about the same thing; I cannot understand why they do not use stops for units on both axis. An universal unit and more understandable..

Here is one I made using stops as both units: https://www.photrio.com/forum/attachments/trix-png.268932/ - from this you can see fast that the data is from 14 stops and linear-ish part is 10 stops. And that the density on this grap is about 8 stops, so pretty doable on normal grade paper.

 

[radiant]

On the x-axis each 0.3 log exposure is 1 stop.

Yup, I know. But why the trouble of multiplying?

Putting stops on the y-axis doesn't make sense. It is density we are interested in on the y-axis..

Yes, if you know that what the density scale means you can "get" it. But if you don't know, it is easier to say "hey this produces this many stops of density range" and you can compare it to your output medium.

But yeah, I know this is controversial and I know it is done this way because it has been always done this way. For newcomers it seems a bit strange. If you ask what these values mean they are usually explained with stops. So why not directly put .. you get my point

 

[radiant]

The confusion there is the concept of "stops of density range". There is no fixed density/stop per se, even for the straight line portion of the curve since that depends on the gradient (slope).

Not in any standard of course, because I'm here redefining the standard (very smart..).

But what comes to printing I think it can still be used to figure out can you reproduce the amount of density on for example darkroom paper. That I believe you can find out from the curve. Density and f-stop have mathematical correlation for sure?

 

[Stephen Benskin]

Using Stops along the X-axis isn't much different than relative log-H. Using anything other than Density along the Y-axis can be problematic because while a Density of 0.30 is equal to one stop, the average gradient for a normal
Subject Luminance Range is about 0.58. A stop exposure for a film developed to an average gradient of 0.58 will have a Density difference of 0.16 or 0.17 for each stop exposure.

There's no rule against stacking different references. Here's a Kodak diagram that does just that.

f/Stops along the X-axis would be confusing. F/Stops are part of an equation. Log-H is how much light strikes the film. For this to happen with f/Stops, you will need to know the Luminance, camera shutter speed , and the constant q for thespecific optical conditions. This means you will need to apply the camera exposure equation to each value along the X-axis. Since sensitometry uses sensitometer and not an optical system, you don't need to incorporate any amera variable thus eliminating many variables that can unknowingly influence the results.

Finally, logs work for all areas. An example of a four quadrant reproduction diagram:

 

[radiant]

It sounds to me like what you are really talking about is tone reproduction. In addition to the film curve you need the paper curve. Then you connect them. There are various graphical ways of doing this. The Lloyd Jones "windmill" diagram is popular but Kodak later had a different way of showing the same thing.

Yes I know that but I'm not diving so deeply into the data; I mean I can quickly check that how the "normal" darkroom papers 6 stops of dynamic range sets into the curve. Nothing else. Just as overview. I know it is 2.0 in log density, but still .. Why the conversion.

Using anything other than Density along the Y-axis can be problematic because while a Density of 0.30 is equal to one stop, the average gradient for a normal Subject Luminance Range is about 0.58. A stop exposure for a film developed to an average gradient of 0.58 will have a Density difference of .16 or .07 for each stop exposure.

Hmm, interesting. But isn't this unnecessary in this use? If I look from X-axis that linear part of the curve is for example 10 stops and look where this sets into Y-axis; I see that it translates to density range of for example 2.0 in logarithmic scale, I know that it just fits the paper density range at grade 2? Or didn't I understand that, hmm. I know that the scene in reality is compressed etc. in the film density, but that is what the characteristic curve shows. And the graph helps me to understand what it means in terms of density, how the film represents the exposure stops. Or did I miss something? I still believe that stop and density has linear mathematical equation?

f/Stops along the X-axis would be confusing. F/Stops are part of an equation.

That is why I talk about stops, _not_ f/stops - just not to mess with exposure calculation. A stop is just double the amount of light. For sure aperture _stops_ are working the same way, for some pretty good reason cameras have pre-set aperture stops, but a stop is still just a double the amount of light. It works great in darkroom printing, my timer + analyzer both work on stops and it is just a joy to work with stops as unit. It is so logical and universal unit.

Thanks for being patient

 

[radiant]

And of course, the obligatory xkcd - strip (a bit worn out, I know):

 

[Stephen Benskin]

That is why I talk about stops, _not_ f/stops - just not to mess with exposure calculation. A stop is just double the amount of light. For sure aperture _stops_ are working the same way, for some pretty good reason cameras have pre-set aperture stops, but a stop is still just a double the amount of light. It works great in darkroom printing, my timer + analyzer both work on stops and it is just a joy to work with stops as unit. It is so logical and universal unit.

Sorry about that. I accidentally combined what Michael and you wrote. I need to take a closer look to what you wrote about film / paper matching before possibly responding.

 

[radiant]

Sorry about that. I accidentally combined what Michael and you wrote. I need to take a closer look to what you wrote about film / paper matching before possibly responding.

No worries. I'm keen to know if I'm totally lost or is my way technically correct but traditionally wrong

 

[DREW WILEY]

These published curves are so abbreviated anyway that one needs to either stare at them a long time to interpolate and extrapolate what is actually going on, or else preferably plot their own families of curves in greater detail using their own densitometer, like Michael has done, though I often debate with him about his interpretations of specific curves. He his own kind of intervals, while I still do it by hand on old-fashioned official translucent Kodak curve-plotting paper, which better represents their long-standing formal convention, yet in much greater detail and longer scale than the product tech sheets themselves do. Plotting paper is graphed LOGARITHMIC, just like density units themselves, and just like formal step tablets for sake of plotting specific films via contract exposure with these.

And generally being truncated at the top, one seldom sees what the shoulder is actually doing in those published film curves; one more reason that if you want something done right, do it yourself.

I mention "families" of curves because no single curve does it all. You need a different curve plotted not only for every different film-developer combination you plan to use, but also for every different relevant variable of time or temperature potentially involved.

 

[Bill Burk]

A pet peeve of mine is just as you say… the published curves have lines drawn at half units so it is virtually impossible to read them. All they give is evidence that manufacturers did their testing.

I prefer lines drawn at 0.02 units with bold lines at 0.10 units so that a point can be drawn in pencil that can be read back to 0.01 resolution, the same resolution that my densitometers display.

Here is my working lab sheet:
https://www.beefalobill.com/imgs/sensitometrymarkvi-6.pdf

The top scale is “value added” and so are the “green lines” these are both specific to my sensitometer. The top scale shows the calculator log meter candle seconds. In base 10 logs, -1 means 1/10th and -2 means 1/100th. Minus 3 means 1/1000th

An 800 speed film has a meaningful (0.01 density when developed) to 1/1000th of a meter candle second. 800 is a pretty decent speed and there aren’t many films faster. So if they make the zero about -4 that should cover everything.

Here is a generic graph anyone can use. The bottom scale is “attenuation”, the densities of the step wedge that you put on the film when you made the test exposures.

https://www.beefalobill.com/imgs/sensitometry.pdf

 

[Craig]

Density and f-stop have mathematical correlation for sure?

Not a constant one. If it was constant, the curve would be a straight line. At the top for example, additional light doesn't build any more density. I could add 2 or 3 more stops of light and it wouldn't build any more density. Similarly at the bottom, until the activation energy of the silver is reached, light and density have a non linear (or no!) relationship. I would think that defining an equation that accurately describes the relationship between stops of light and resultant density would be a very complex endevour.

Why try and correlate density to stops anyway? Density can be measured directly off the film or paper. If you measure density it's a direct measurement that can be compared to any other film. Stops are always relative to something else, and that would need to be defined to compare between films or papers.

 

[radiant]

Not a constant one. If it was constant, the curve would be a straight line. At the top for example, additional light doesn't build any more density. I could add 2 or 3 more stops of light and it wouldn't build any more density.

You are mixing the film characteristics here. I'm not talking about stops used in film exposure.

Stop is just doubling the amount of light. Nothing else. It is not related to anything else.

Why try and correlate density to stops anyway?

Because stop is more understandable (for me at least) and very universal thing in photography. It is easy unit to calculate because it works throughout the whole process. The log density reading doesn't translate to anything else known, it is its own scale.