Are the film specification sheets accurate?

retro film · Aug 15, 2013

I feel like theyre not accurate (but they should be, as with any technical data reference). For example, I was looking at the Kodachrome 25 data sheet, and the spectral sensitivity curve for the magenta forming layer seems too low. It seems like with sensitivity that weak for the green layer, any neutral grey objects shot under the correct lighting would appear pinkish due to access magenta dye.

Can anyone comment on this? are the sheets accurate? Is my suspicion correct?

Kevin Kehler · Aug 15, 2013

Well, there is two things going on here - one, film manufacturers are using $100k+ machines to make determinations as opposed to most of us who just eyeball it. Second, they generally are testing to a specific standard, one which most of us are not using. So, I have always found ISO speed ratings to be very liberal and found much better to use my own rating for films. I don't doubt the film meets the test criteria but the test criteria are not real world scenarios, they are manufactured scenarios that the real world does not replicate.

In short, trust the sheets for a start and then develop your own knowledge, based on your shooting style, location in the world, development method and printing style.

DREW WILEY · Aug 15, 2013

The official sheets are very accurate, but determined according to company and industry standards, so you have to learn how to read them.
and what the implications are for your personal use. You are only partially correct. The greens of earlier Ektachromes were in particular infected with quite a bit of magenta, and Kodachrome was engineered in an entirely different manner; but if you were an advanced color printmaker in something like dye transfer during that era, you knew that you had to do certain things differently with Kodachrome than Ektachrome. Looking at a slide on a proper lightbox or projector, you wouldn't spot anything off. Kodachrome 25 was famous for neutral grays, but sometimes had minor issues with warmer neutrals or warmer "spring" greens. Dye sensitivity in these curves is not synomymous with how that respective hue will actually look in the end result. It's quite a bit more complicated than that. You also have to note at what
color temperature standard they made these charts to begin with.

AgX · Aug 15, 2013

I know several sheets that a wrong. At one company I assume malafide intention (or blatant ignorance) at another ompany I assume negligence.

Furthermore there are no fixed rules how to make up such data sheets. As result it is often difficult to compare films from different manufacturers.

Photo Engineer · Aug 15, 2013

I know that Kodak attempts to make their data sheets accurate. It does take some skill in reading spectral sensitivity data, as it depends on the light source used for exposure.

If the film works though, it works!

PE

clayne · Aug 15, 2013

Also, who's to say that just because sensitivity is lower in one dye vs another that the dye coupling chemistry won't account for that? Kodak knows what they're doing.

Europan · Aug 16, 2013

Eastman-Kodak had false curve toes with their black-and-white reversal films PXR and TXR. Actually, the entire curve was simply shifted out of the base density of log 0.23 in order to pretend clear lights.

Photo Engineer · Aug 16, 2013

Europan said:
Eastman-Kodak had false curve toes with their black-and-white reversal films PXR and TXR. Actually, the entire curve was simply shifted out of the base density of log 0.23 in order to pretend clear lights.

How interesting. Can you show us a reference?

Thanks.

PE

falotico · Aug 16, 2013

EK has a reputation for maintaining the highest technical standards and I would hesitate to suggest that they published false data. They have far more to risk than to gain. If I found an inconsistency I would study it further to see if there was some factor which I missed.

Tom1956 · Aug 16, 2013

Can you name another company that has worked harder on accuracy than EK? I don't think there ever were any. When they went under, a world standard of quality was lost.

Photo Engineer · Aug 16, 2013

Tom, they (EK) are still here and still making film to the same high standards, just at a reduced rate of output.

PE

retro film · Aug 16, 2013

Well the issue I am having with the Kodachrome 25 sheet is this:

If you look at the H-D curve on the sheet where the density=1, you see that the required exposure to get density=1 for both blue and green layers is pretty much exactly the same. Both blue and green layers as well as red all must have equal densities in order for the developed film to look neutral in color if a neutral object was shot under the correct lighting. So from this we know at the very least the blue and green layers must absorb very closely equal amounts of exposure to produce a neutral color in the developed film.

But then when you look at the sensitivity curves, you see the blue layer has a big peak at about 420-430 nm, that peak will cause it to absorb a lot more exposure than the green layer, a lot more. So really it doesnt look like there is anyway blue and green will absorb the same amount of exposure if we are shooting a neutral object. but that is what must happen if the two layers are to develop equal densities. This is the problem.

But maybe the color temperature of the illuminant compensates for this? No, the correct color temperature for this film is 5500K-5600K, which while slightly lopsided giving more weight to the green and red side of the spectrum, it is still approximately balanced. The green and red range of the spectrum hardly are getting much more exposure than the blue range. The big peak in the blue sensitivity curve at 420-430nm will cause it overcome the subtle imbalance of the spectral power distribution favoring the green and red range, as a result the blue layer still absorbs more exposure and so the two layers blue and green are not equal.

So if Kodak is releasing accurate data sheets, then where is the flaw in my logic? What am I failing to account for?

Tom1956 · Aug 16, 2013

With a Kodachrome 25 sheet, what difference does it make now?

retro film · Aug 16, 2013

not much, it just bugs me. And makes me wonder if this sheet is wrong, how many other sheets are also wrong including the ones for films Kodak still sells.

Or it could mean that there is something wrong with my understanding of the how to interpret data sheets. I thought I knew how to read them, now Im questioning myself.

Tom1956 · Aug 16, 2013

Typos are always possible, but speaking for myself I always figured the technicians at Kodak knew something I just wasn't considering. I mean, obviously Kodachrome was the best color film ever made, so that speaks for itself. Just my opinion, mind you. What you have read in the tech data in itself explains a lot when the results are self-evident. It may have been a simple reciprocity-effect in that layer or dye, or another variable the technicians compensated for. In my 40 some years in photography, I don't recall ever seeing any typo errors in a Kodak publication, though, that I could ever tell.

Photo Engineer · Aug 16, 2013

Ok, lets try this approach for you guys.

The human eye integrates the area under the curve of any dye. The broader it is, the lower the peak must be and the narrower it is, the higher the peak must be. The same thing holds for the spectral sensitivity. So, basically, you have to make a "cutout" of the total dye curve all by itself and "weigh" it to see how much it is "worth" in the imaging process.

If you look at the dye curves (spectral dye density curves), you will see that the cyan is very narrow, and the magenta and yellow are very broad. Thus, to make a neutral and the rest of the colors as well, you need more cyan. See the sensitometric curves just to the left of the dye density curves. The cyan is higher in contrast to the densitometer.

So, when you view it, the Kodachrome appears neutral even though to the instruments there is apparently more cyan (or less magenta and yellow).

This is both good and bad.

But, it works!

PE

clayne · Aug 17, 2013

I'm so truly sad we don't have Kodachrome anymore.

retro film · Aug 18, 2013

Alright, so Ive concluded that the specification sheets ARE accurate as Ive been seeing the same issue in specification sheets for other films and from other brands. Theres no way they can all be wrong.

So the problem here is that I am missing something when interpreting the spectral sensitivity curves. I thought I know how to read them, I guess not. Can anyone help me out by explaining how I would use the log sensitivity value at a wavelength to determine how much light a layer is absorbing from light at that wavelength? Theres not a lot of information out there explaining how to read these data sheets, in fact theres really none.

AgX · Aug 18, 2013

That is why I question the value of all those sheets. The majority of readers does not understand them, or at least not correctly.

They are written by engineers for engineers.

Concerning your curves-issue I posted on that very issue in detail some time ago.

Photo Engineer · Aug 18, 2013

Retro...

The sheets can be used that way. The vertical height of the curve is the response in LogE. The "Y" axis is thus the speed at that wavelength. However, the speed is the integral of the entire curve. And so the speed that you measure at one wavelength is infinitely small regardless of the "Y" value unless combined with all of the other values.

This all is related to calculus, I'm afraid. And, the curves have to be smoothed out because they are bumpy due to the dyes themselves.

It is easier for a B&W film than for a color film, due to overlap. Thus, some green light produces red exposure. How do you estimate that? You need to do spectral curves of each dye formed at each wavelength and then begin to calculate what must be done to improve color reproduction.

To do that, you must also include a factor for the human eye vs the measuring system, and you must therefore consider the half band width of the imaging dyes.

If I can find one, I might post a typical picture used for this type of measurement.

PE

MartinP · Aug 18, 2013

May I ask, to what practical application will the OP put a perfect interpretation of these spec-sheet curves?

It's not meant to be a silly question, as all that consistency necessary thirty years ago - when film was the starting point for many analogue image-production processes - seems to have been removed. Now aren't we are using consistency in our own personal processes to achieve what we want at the level of an individual print (or small, personally repeatable, batches), not as the start of a long chain of analogue processes where the published result depends on perfect global consistency at every step?

retro film · Aug 18, 2013

Photo Engineer said:
Retro...

The sheets can be used that way. The vertical height of the curve is the response in LogE. The "Y" axis is thus the speed at that wavelength. However, the speed is the integral of the entire curve. And so the speed that you measure at one wavelength is infinitely small regardless of the "Y" value unless combined with all of the other values.

This all is related to calculus, I'm afraid. And, the curves have to be smoothed out because they are bumpy due to the dyes themselves.

It is easier for a B&W film than for a color film, due to overlap. Thus, some green light produces red exposure. How do you estimate that? You need to do spectral curves of each dye formed at each wavelength and then begin to calculate what must be done to improve color reproduction.

To do that, you must also include a factor for the human eye vs the measuring system, and you must therefore consider the half band width of the imaging dyes.

If I can find one, I might post a typical picture used for this type of measurement.

PE

Photo Engineer, thanks for the response,

I actually am familiar with calculus, so feel free to discuss more in terms of it, I actually prefer more technical explanations.

Could you work out an example? Lets say I have a spectral power distribution where the spectral irradiance is as follows: between 450 and 451 nm, the spectral irradiance is 7 erg cm^-2 s^-1 nm^-1 and at every other wavelength the spectral irradiance is zero. What would the exposure be for the blue layer in lux-seconds for Kodachrome 25? I would like to see an example worked out so I can find the error I think is present in my calculations. You can just give me the approximate value (no need to actually integrate over that 1nm range, just approximate using a riemann sum of 1 rectangle to keep calculations simple)

I noticed that the exposure in the spectral sensitivity curves for kodachrome 25 is a radiant exposure
but the exposure on the H-D curve is a luminous exposure. Ive accounted for that in my conversion by involving the luminous efficiency function, I think this may be where my mistake lies, theres a detail(s) Im failing to account for.

Truzi · Aug 18, 2013

MartinP said:
May I ask, to what practical application will the OP put a perfect interpretation of these spec-sheet curves?

It's not meant to be a silly question, as all that consistency necessary thirty years ago - when film was the starting point for many analogue image-production processes - seems to have been removed. Now aren't we are using consistency in our own personal processes to achieve what we want at the level of an individual print (or small, personally repeatable, batches), not as the start of a long chain of analogue processes where the published result depends on perfect global consistency at every step?

Some end-users can interpret, to some degree, in a manner meaningful to them. If given enough time I could probably understand the specs, but nothing I do (or want) requires the vast majority that information. Still, it is nice to have, so long as the person can interpret, and it can be interesting just to read and learn.

Along those lines, consider ingredients listed on food. Most do not read them. Some do, but only look for certain things and do not take full advantage of the information (which is not terribly useful descriptive in the first place).

However, I am lactose-intolerant, and have learned there are many words that basically boil-down to "dairy" product/lactose. Just because a label does not say "milk" or "dairy" does not mean it is not in there in some capacity; the laws in the U.S. don't require that in the way many believe. Fortunately, disclosure is required for ingredients that can be deadly.

Only by reading and interpreting can I discern what I need to know, even if it is of no interest to someone else.

Admittedly, this does not affect the majority of analog food users, but I am glad the information is there and that I can understand it well enough for my purposes.

MartinP · Aug 18, 2013

In practical use, one can assume that, under the relevant light-source and with any "Professional" film, you will end up with a neutral grey and optimum colour representation.

One then measures the light you are actually working in (assuming that it is a continuous spectrum source, else you'll need to test, spec-sheet or no), goes to the filter diagram and check how many mireds you need to shift and choose the appropriate filtration.

Working out the modification from first-principles, based on wavelengths and amplitude of radiation, seems somewhat inconvenient.

Photo Engineer · Aug 18, 2013

Retro...

What you ask is a rather large problem to me at present. At EK, it was all done by computer. And, our work was directed to getting the total speed which was reported back to us. And so, in the third image, this is a material with only blue speed, so the green speed would be reported as zero. In the second image, the green speed would be reported in some given speed value related to the illuminant used for exposure (thus for your example, the illuminant must be specified). You did this, but we used a different energy standard.

In the first image I show 6 types of spectral sensitization. In the final graph I show a concentration series of one dye on a film. In all of these cases, the energy was held constant so that the output was the same on the "X" at each wavelength. Speed can be calculated for each dye at each concentration, and by varying the distribution of energy, one can then calculate the speed under daylight or tungsten (and etc.).

The next to last scan is an exposure to light of different wavelengths and overlapping, thus giving us a human eye view which is compared to the densitometry to give us values that relate color reproduction to dye set to spectral sensitivity to illuminant (working backwards from the actual image). There are about 25 or so of these exposures in a set.

In any event, although I have examples, I have no computer program to do this for me. I have given some real data but that is as far as I can go.

Sorry. And sorry for the scrambled order of the uploads. APUG did that!

PE

Are the film specification sheets accurate?

Attachments

Privacy & Transparency

Privacy & Transparency