Question About Film ISO and Exposure Index.

I was reading the technical data on Ilford Delta 3200 film. See: https://www.ilfordphoto.com/amfile/file/download/file/1913/product/682/

It says the film's EI is 3200, yet its ISO is 1000. Why the big difference? What the heck ISO should I use?

Forgive my ignorance, but I'm getting old and my brain doesn't work like it used to.

Hmm... I'll say...!

http://bayouline.com/o2.gif

In simple terms, ISO (international standards organization) comes up with standardized ways of measuring everything. Measuring contrast of an image is something they've standardized, look up "ISO contrast". Then "ISO speed" defines the amount of light you need to achieve the ISO contrast (film or digital).

Exposure index is simply whatever your camera is set to. You can give your film less light but develop longer (pushing). Delta 3200 is a misleading name, you should interpret it as "ISO 1000 film optimized for pushing to 3200". BTW you can "push" digitally too: you'll end up with a darker RAW file, but move "exposure" slider to the right (effectively multiplying all recorded values by some factor) and you'll end up with a brighter image but more contrast (because of the reduced range of useful values)

ISO has a very specific definition based a given exposure producing a given density. The ISO is "baked in" to the film so to speak, and can't be changed.

Film, however, CAN be intentionally underexposed and then overdeveloped. This does not change the ISO, but allows a reasonable density image to form under those conditions. This comes with a big downside, though, in that over-development increases contrast dramatically.

Superspeed B&W films(Delta 3200 and TMAX P3200) typically have low enough contrast at their nominal ISO that they handle pushing fairly gracefully to ~3200 or higher.

There is an additional term to throw into all of this, and that's the Exposure Index or EI. That's the speed at which a film is actually exposed. Delta 3200 and TMAX P3200 are DX coded(for modern cameras that can read DX codes) to set to an EI of 3200. EI is a common concept with really any B&W film-i.e. a person might expose Tri-X at an EI of 200(overexposing it per the ISO) and under-develop it to reduce the contrast of what is otherwise a decently contrasty film.

There is no film on sale that has a true ISO much greater than 1000. That's based on every published film curve I've ever seen. "Pushed" (LOL) or not.

Ilford's "3200" label is mainly marketing hype, and is about the maximum EI rating you can get away with, without having very noticeable loss of shadow detail.

Ahhhh! It's based on digital, not film! Ahhhhh...

Huh!?

I'm pretty sure that Delta 3200 was around well before digital cameras became a viable opposition to film.

Huh!?

I'm pretty sure that Delta 3200 was around well before digital cameras became a viable opposition to film.

I was talking about the article Sandy posted - the article relates to digital sensors and their ISOs. I couldn't relate to that...

ISO values for digital sensors are a strange and different question.

Giving ISO ratings to black and white films is complicated as there is no standard ISO developer.

ISO also makes some assumptions on the shape of the curve that may or may not apply.

(Ilford mentions this in some of their data sheets.)

For color films, the ISO values depend on the use of the process meant for the film.

Though for faster films, that isn't quite right. Faster films naturally need more development, such that the

times for standard color processes likely aren't optimal for faster films. The result is that some films do

better with push processing than one might normally expect.

In any case, you are supposed to say EI when exposing other than at the ISO or ASA value given.

The 1960 ASA/BS/DIN standard - later adopted wholesale as the ISO standard - for measuring black & white film speed, specifies the following:

A 'speed point' density of 0.1D above base + fog gives the effective speed. The exposure, in Lux-seconds, required to give this density is divided into 0.8 to directly give the arithmetic speed.

Or the log(base 10) of the reciprocal exposure is multiplied by 10 to give the logarithmic speed.

I.e.

Arithmetic speed = 0.8/E (at 0.1D over base + fog)

Log speed = 10 log(1/E)

As far as development goes, the specification is that an MQ developer will be used such that; at 20x (+1.3 log E) the exposure required for a density of 0.1D above base+fog there is an increase in density of 0.8D.

This = a C.I. of about 0.62.

Obviously, this is a bit of a 'Catch 22' juggling act to achieve with any high degree of accuracy, and if you look at any published H&D curves you'll rarely find one that exactly fits the above ISO specification. But close enough is good enough, since the speed point changes hardly at all with a change in development.

The ISO speed requirements for colour film are different and more complex.

For reversal (slide) films the minimum density (= maximum exposure) point is changed to 0.2D, and the density range is pegged at 2.0D, both over base+fog. These two points are then plugged into a calculation involving an undisclosed constant to give the speed number.

There was no internationally agreed standard for determining the speed rating of colour negative film when the above American, British and German national systems were agreed and amalgamated.

And I'm not going to pay some bunch of greedy Swiss gnomes 38CHF to find out any current ISO standard, thanks.

 

(snip)

 

Obviously, this is a bit of a 'Catch 22' juggling act to achieve with any high degree of accuracy, and if you look at any published H&D curves you'll rarely find one that exactly fits the above ISO specification. But close enough is good enough, since the speed point changes hardly at all with a change in development.

 

(snip)

Hardly at all could be a lot more than none.

Also, that only gives two points on the curve, but a lot could happen over the rest of the curve.

Ilford has a comment, something about "foot speed", and if you look there is no part that is really straight.

but a lot could happen over the rest of the curve.

Reminds me of the introduction to an old TV series - "Anything could happen in the next half-hour!"

It very rarely did.

if you look there is no part that is really straight.

Since it's a log, log curve 'straightness' is an illusion and difficult to assess.

Well, you can also talk about density for positives, such as reversal film.

But yes, it is mostly used for negatives.

One reason for being ''just dense enough'' is that you might not have enough light, depth of field, and shutter speed for more density. Often a little more than ''dense enough'' is best. You want the important parts to stay within the (more or less) straight part of the characteristic curve. Negative films have plenty of latitude for overexposure, resulting in more dense, harder to print, but otherwise fine negatives. (That is, don't waste film speed.)

Since the density can't go below zero, you want things that aren't completely black to be at least a little above 0.

The ISO values are supposed to have a one stop margin.

There are a lot of good reasons for having a margin, such as an averaging meter might miss the correct exposure,

there might be light sources in the frame confusing the meter, you might be using a camera without a

meter, and it might be darker than you though. (Humans aren't all that good at judging light levels, as our

eyes try to compensate.)

For many films, the recommended times for a one stop push are the same as no push.

That is, they are using the margin when you can't get to the right exposure.

(Already at maximum aperture, and your hands are only so still.)

when you guys talk about "density", you're talking about negative density yes?

Yes, but better to say film density (to include reversal films).

Because an ideally exposed negative would be just dense enough, not too thin and not too blacked out. Is that correct, or am I off base here?

For the most part, yes. Fwiw, though, a "film speed," by itself, is not enough to establish a "correct" exposure. Historically a film speed has been used to establish the point where a film is just beginning to show a useful response to light. But when you use an exposure meter you are generally measuring light levels that are much higher. So there are exposure meter standards that sort of set a relationship between the two light levels, etc., etc.

Like many other things, the closer you look the more complicated things become.

If all subjects were 18% gray cards, then it would be easier.

It seems that typical subjects are close enough for averaging meters

(the only kind in the early days of metering) to work.

Subjects will normally have a range of reflectances.

Yes the film speed is measured at a point a little above where the film starts responding, which should be

enough to allow reflectance above and below 18%, especially the below. (And especially as some parts

might be in an actual shadow.)

So, traditional meters average over the whole scene, which works surprisingly well, considering.

Sometime later, spot meters came along, which measure more toward the center, and hopefully

more important parts of the subject.

Better than averaging (arithmetic mean) would be geometric mean. That is, the value whose log equals

the average of the logs. With a spot meter on a manual camera, you can point different directions,

and feel (usually) where the exposure is on the aperture scale, then pick a point in between.

Or you can use matrix metering, where the camera does all this internally, to get the range of

reflectance within the range of the film.

when you guys talk about "density", you're talking about negative density yes?

Yes, but density is density. Whether it's a negative or reversal film.

Density, scientifically, is specifically measured as the logarithm of opacity, and opacity basically equals (light out/light in). In other words, opacity is the fraction of light that passes through whatever is being measured, and density is the log^base 10 of that fraction. Density has the symbol D after the figure.

For example: A one stop or x2 neutral-density filter passes half the amount of light incident upon it. As a decimal fraction this is 0.5, and the log^10 of 0.5 = -0.3. The negative sign is usually dropped, therefore that filter has a density of 0.3D.

Densities can be simply added. So a two stop ND filter has a density of 0.6D. Three stops 0.9D, etc.

For a coloured medium, the density refers to its opacity to light of the complementary colour (e.g. yellow/blue, magenta/green, red/cyan, etc.)

The ISO values are supposed to have a one stop margin.

That 'safety margin' was done away with in the 1960 ASA/DIN/BS (later ISO) specification that I quoted above, and hasn't been re-introduced.

There is no 'one stop margin' in the ISO speed specification. As can be seen by looking at any published H&D curve.

However, a one-stop change in exposure has very little effect on the visual appearance of a B&W negative, especially in the over-exposure direction. Causing only a 0.2D average shift in density. (0.1D is approximately the density of a 'transparent' sheet of thin glass or plastic.)

If all subjects were 18% gray cards, then it would be easier.

 

It seems that typical subjects are close enough for averaging meters....

The figure of 18% reflectance wasn't dreamed up on a whim. It was derived empirically by studying thousands of negatives that flowed through the hands of photo-finishers during the early days of mass market photography. This study discovered that the 'average' subject would reflect 18% of light incident upon the scene.

That figure works out very well in practise and nobody has proposed any good reason to change it. It encompasses a subject brightness range of more than 7 stops easily, and allows for a half-stop margin over a subject with 100% Lambertian reflectance (e.g. matt white paint, virgin snow, fluffy white clouds, etc.). While at the shadow end it takes a material like black velvet to get a reflectance of less than 2.25%.

If you choose to want to capture detail on the surface of the sun whilst simultaneously wanting to see detail 20ft inside of a shadowed cave mouth, then 18% reflectance probably isn't the standard for you!

As well as I know it, there is still supposed to be a one stop margin, so one less than before 1960.

Color filters should have an absorption spectrum, density as a function of wavelength

(or wave number). But yes, for CP filters it is the density somewhere around the

complementary color of the filter.

Yes, but density is density. Whether it's a negative or reversal film.

 

...

 

The figure of 18% reflectance wasn't dreamed up on a whim. It was derived empirically by studying thousands of negatives that flowed through the hands of photo-finishers during the early days of mass market photography. This study discovered that the 'average' subject would reflect 18% of light incident upon the scene.

...

I seem to recall reading that Ansel Adams played a part in selecting 18% rather than 12% or vice versa as middle gray as part of his Zone system theories. For some reason, reflective and incident readings also are off often by one stop. I forget the full stories on both these things. Maybe someone else can fill us in.

I seem to recall reading that Ansel Adams played a part in selecting 18% rather than 12% or vice versa as middle gray as part of his Zone system theories. For some reason, reflective and incident readings also are off often by one stop. I forget the full stories on both these things. Maybe someone else can fill us in.

I am not sure about that one, but some common skin colors are about 36%, so if you forget your gray card you

can meter off your hand, and add one stop.

For some reason, reflective and incident readings also are off often by one stop. I forget the full stories on both these things. Maybe someone else can fill us in.

I remember this post, see link - Shooting Film at Box Speed

Short answer: the metering standards allow the manufacturer to bias the reading over a limited range, by using different values for either a 'K' constant (reflective meters) or a 'C' constant (for incident meters). The stipulation is that they must also specify the value that they use. (See the user manual, in the "specifications" section.) One probably needs to see the metering equations from the standards to make sense of these values.

I think that Minolta used to go with the ~18% value for their incident vs reflective values, so likely the readings would mostly agree between the two using a frontally lit Kodak 18% gray card. Whereas some other common brands want to see a less reflective card.

I recently did the comparison with a Gossen meter. As I recall, I think this meter was looking for a nominal,14% reflectivity from the gray card. (The method I used was to rotate the card away from the light source until the readings matched. Then calculate, from the angle, the reduction in light striking the card.)

I seem to recall reading that Ansel Adams played a part in selecting 18% rather than 12% or vice versa

As I've commented before, the sainted Ansel didn't seem to be able to match the descriptions of his precious Zones to actual, measured reflectance values.

He states quite categorically that Zone V mid grey is equal to 18% reflectance, while also describing something like 'snow with texture, a white picket fence, etc.' as falling on Zone VIII, a full three stops away.

Now 3 stops more than (8 times) 18% reflectance takes us to a ridiculous 144% reflectance!

Something wrong somewhere.

FWIW, a figure of 12.5% reflectance for Zone V would resolve that particular issue, but that was never, apparently, AA's argument.

Whatever. I'm pretty sure that the 18% 'average' reflectance figure came from Kodak's research labs and not from Mr Adams.

Also FWIW, if you take a reading from a white card - a doubled up sheet of copier paper will do - and add 2.5 stops, then you get the equivalent of an 18% grey card reading. Doing this gets an exact match for placing the white card reading at the extreme right, but not saturated, of my digital camera histogram. Co-incidence? I don't think so.