Spectral sensitivity and its possible consequences

snusmumriken · May 14, 2024

In the last year or so I've been experimenting with Kodak/Eastman Double-X. I like it in many ways, but in some ways it frustrates me. So I fell to wondering how it compares with more familiar films in terms of spectral sensitivity. Here are the spectral sensitivity curves as presented by Kodak and Ilford in their technical data sheets.

My first question is whether the curves are even comparable between the two manufacturers? Do they use equivalent methodologies? I'm not interested in the vertical scale, of course, just in relative sensitivity of each film across the visible spectrum.

If they are comparable, it looks as though HP5+ and (especially) FP4+ are a lot less sensitive (relatively) to the violet/blue end of the spectrum than are Tri-X and Double-X. Is that correct?

And if that is the case, what are the practical consequences? Does it mean that the two Kodak films will make shadows more luminous and pick out more shadow detail in them? Will the Kodak films show a more marked response to filters than the Ilford films, or vice versa?

Rudeofus · May 14, 2024

The Kodak charts don't mention spectral distribution of the light source, whereas the Ilford charts explicitly mention a light source with a very strong emphasis on long red wavelengths. If Kodak compensated their sensitivity spectrum, i.e. recalibrate their measured curves against the incoming light spectrum, then Kodak's charts will appear more blue sensitive than Ilford's without actually being so.

Even if both makers use the same calibration and their films are indeed different, a suitable color correction filter should be able to balance out these differences.

Andrew O'Neill · May 14, 2024

The film's characteristic curves are more important to me, as it tells me more about how a film will render tones. But one cannot ignore the spectral graph, either. If you do a comparison between tri-x and hp5, you will find very little difference in spectral response. HP5 and say Delta 400, which has more red sensitivity, you will see a difference in how reds are rendered, although it can be subtle.

Mark J · May 14, 2024

I hadn't properly thought about that before Jonathan, but it's clear to me now that Ilford's test is a bit simpler and cheaper using the wedge spectrogram method with essentially a normal lightbulb. The tests Kodak have used would require a proper way of calibrating and equalising the light output across the visible spectrum in order to provide a more scientifically accurate view of the film sensitivity. I'm sure one of our test engineers at work could explain the detail of the kit for me.

There is quite a big difference across the 400-700nm bandwidth as to how a Tungsten 2850K source looks, compared to a normalised flat spectrum.
I would say, just by eyeballing the two sets of data, that the apparent differences between the Kodak and Ilford films could very well be fully explained just in terms of the different test method - they may in fact be very similar.

If the differences were in fact real and inherent to the emulsions, then I would expect to see very different filter factors quoted for the two filmstocks.

albireo · May 14, 2024

Andrew O'Neill said:
The film's characteristic curves are more important to me, as it tells me more about how a film will render tones. But one cannot ignore the spectral graph, either.

Interesting Andrew. I pick one film rather than another primarily because of any differences in spectral sensitivity.

Characteristic curves tell me more about how linearly or non linearly one film in a certain developer will map highlight and shadows (the so called 'toe' and 'shoulder') and I will learn something about contrast from the curve slope, but I'd never be able to guess from e.g. Foma 400's characteristic curves in D76 1:1 that Foma 400 would render Caucasian skin tones and lips so pale - I can only tell that from the spectral graph, or even better from a test shot of a colour chart such as the Kodak Gray Card plus.

Mark J · May 14, 2024

I somebody is keen and has the time and a bit of maths & Excel knowledge , it should be relatively easy to create proper objective spectral sensitivity curves for the Ilford materials by multiplying the Ilford graph data by the inverse of the 2850 K black-body curve, over a few points across the spectrum.
You can easily find a black-body curve generator on the 'web. I can get you a link if needed.

( ps. no .. I don't have time at the moment !! )

snusmumriken · May 14, 2024

Mark J said:
I somebody is keen and has the time and a bit of maths & Excel knowledge , it should be relatively easy to create proper objective spectral sensitivity curves for the Ilford materials by multiplying the Ilford graph data by the inverse of the 2850 K black-body curve, over a few points across the spectrum.
You can easily find a black-body curve generator on the 'web. I can get you a link if needed.

( ps. no .. I don't have time at the moment !! )

I could do that, Mark, but I don't understand the logic. Could you explain a little?

Andrew O'Neill · May 14, 2024

albireo said:
Interesting Andrew. I pick one film rather than another primarily because of any differences in spectral sensitivity.

Characteristic curves tell me more about how linearly or non linearly one film in a certain developer will map highlight and shadows (the so called 'toe' and 'shoulder') and I will learn something about contrast from the curve slope, but I'd never be able to guess from e.g. Foma 400's characteristic curves in D76 1:1 that Foma 400 would render Caucasian skin tones and lips so pale - I can only tell that from the spectral graph, or even better from a test shot of a colour chart such as the Kodak Gray Card plus.

You're correct of course with how skin will be rendered, relative to spectral response. I guess I never thought of that because I never photograph people.

Mark J · May 14, 2024

( Reply to Snusmumriken )

Installment 1 ( ! )

The reason the Ilford responses look biased towards the orange and red is that the output of a tungsten lamp at 2850K ( filament temp ) is a lot higher in the red than in the blue. In fact the emission hits the deck at somewhere under 390nm.

Here's one link :

Spectral Calculator-Hi-resolution gas spectra

Accurate, rapid online simulation of high-resolution molecular spectra, and other spectroscopy tools for researchers, teachers and students.

www.spectralcalc.com

I have just run it for 2850K and a band from 400nm to 700nm ( = 0.4 to 0.7µm ) -

Mark J · May 14, 2024

Installment 2 -

Unfortunately that mean little website expects you to start a paying subscription just to download a text file of that graph data !
( we can do better than that, give me a day or two ).

The idea is to take the Ilford graph, probably print it off big onto a piece of paper , draw more lines up from the axes to cross the curve eg. every 20nm, and collect a set of points & estimated values on the curve, that we can feed into some Excel cells.

Then you do the same every 20nm for the 2850K black body curve from 400 to 700nm. It doesn't matter what the units are , big or small - we can deal with that later.

Then we create a reciprocal ( 1/ ) of each data point from the black body curve .

Then we multiply each Ilford response point by the relevant reciprocal black-body data point, across the spectrum.

Finally, we can just adjust the resulting units to make the highest point on the graphed data be called '1.0' and scale the rest accordingly.

DREW WILEY · May 14, 2024

I think it's a little more complicated than that, because the panchromatic sensitizing dyes aren't necessarily linear responsive respective to each other. You'd have to ask an expert. It's a shame that only one type of spectrogram or another has been published, rather than both tungsten and daylight balanced, or at least a single common convention.

But if seemingly minor distinctions between at least matched light source spectrograms weren't important, why are the routinely included in technical spec sheets? For example, the difference between HP5 and FP4 is significant not only at the blue end, but in at the red end, if you notice that red sensitivity starts to crash at 650nm with HP5, versus nearly 700 for FP4. That might make a real difference in choice of specific red filters, if you're trying to push the envelope into sorta near-infrared.

Mark J · May 14, 2024

No, it's not more complicated than that. The Ilford one is done with 2850K, whereas the Kodak one is a true emulsion sensitivity referred to a flat ( normalised ) spectral output. There is no 'daylight' involved. The bit about "the panchromatic sensitizing dyes aren't necessarily linear responsive respective to each other" is just confused irrelevant words, I'm sorry.
I don't need to 'ask an expert' because I thoroughly understand what I'm explaining here - it's the sort of thing I regularly do in my day job when we're looking at sensor profiles and spectral weightings for optics in various different wavebands.

DREW WILEY · May 14, 2024

Well, you are plainly just guessing. Sensitizing dyes and the kind of sensors you might personally work with aren't the same thing. Your sensors must have an already known, calibrated, and programmable set of parameters. But in this case, changing a sum Kelvin response a certain amount doesn't mean every illuminance wavelength will be handled in the same proportion by the sensitizing dyes. That's film engineer territory. Time for a second opinion.

You refer to "normalized". Normalized to what? There has to be a standard. In one case, we already know it's 2850K tungsten. The other must be equivalent to the manufacturer's idea of daylight, because that is how films are used, or at least adjusted to using supplementary filters, and what filter factors are also compared to. They aren't "normalized" to honey bee vision.

I've spent thousands of hours with spectophotometers myself, of multiple designs, and know that dyes typically don't behave as neutral linear profiles. You merely have to look at the spectrograms for dyed gels like in a Wratten filter handbook to see that achieving certain changes can occur in various ways, often involving discontinuities. Combine that fact with potential idiosyncrasies of pan sensitizing dyes themselves - what is the net effect - specifically? Probably NOT a simple answer.

reddesert · May 14, 2024

The average photographic lens isn't going to image or transmit that well blueward of say 350 nm, and for some or many lenses the cutoff will be closer to 380-390 nm or so, depending on the types of glass used. If you have a UV or Skylight filter on the lens, virtually guaranteed to have low transmission shortward of say 390 nm ("UV filter," not just a sales term). For ex the common Nikon "L39" UV filter starts to cut off around 390 nm. (And a Nikon "Y48" yellow filter cuts off blueward of 480 nm. Nikon labeling is sometimes comprehensible.)

See for example: https://www.makarovns.com/INTERESTS/Photo_Filters/index.htm

So basically I think the bluest end of the Tri-X curve is irrelevant unless you are doing UV photography with film, and you should really just be concerned with the slope of the transmission curve between about 400-700 nm, and there the question is whether Kodak and Ilford have followed similar methodology to normalize the sensitivity curves to the light source used, which is what Mark is saying.

RalphLambrecht · May 14, 2024

snusmumriken said:
In the last year or so I've been experimenting with Kodak/Eastman Double-X. I like it in many ways, but in some ways it frustrates me. So I fell to wondering how it compares with more familiar films in terms of spectral sensitivity. Here are the spectral sensitivity curves as presented by Kodak and Ilford in their technical data sheets.
View attachment 370419
My first question is whether the curves are even comparable between the two manufacturers? Do they use equivalent methodologies? I'm not interested in the vertical scale, of course, just in relative sensitivity of each film across the visible spectrum.

If they are comparable, it looks as though HP5+ and (especially) FP4+ are a lot less sensitive (relatively) to the violet/blue end of the spectrum than are Tri-X and Double-X. Is that correct?

And if that is the case, what are the practical consequences? Does it mean that the two Kodak films will make shadows more luminous and pick out more shadow detail in them? Will the Kodak films show a more marked response to filters than the Ilford films, or vice versa?

the curves are comparable but your conclusions are a bit iffy. All I can see is that the films are very similar as far as spectral sensitivity goes and should respond similarly to contrast-enhancing filters.

DREW WILEY · May 14, 2024

reddesert - in the context of black and white films, we obviously often use rather strong "contrast filters". I have all sorts of UV, skylight, and warming filters relative to precise color film usage, but never personally use those in relation to black and white films.

But when it comes to contrast filters, one soon learns that black and white films are not all alike, even from the same manufacturer. And this fact has real implications in terms of final look in a print.
We all learn certain lessons the hard way. Often people get quite comfortable with a particular film,
but once that gets discontinued, or we try something new for its own sake, a new learning curve is encountered. That especially applies to factors like filtration. I certainly know how to read spectral graphs; but still, I do a lot of real world testing with any unfamiliar film before I commit to it on any important project.

koraks · May 15, 2024

DREW WILEY said:
Well, you are plainly just guessing. Sensitizing dyes and the kind of sensors you might personally work with aren't the same thing. Your sensors must have an already known, calibrated, and programmable set of parameters.

Sounds like you're plainly just guessing.

DREW WILEY said:
every illuminance wavelength will be handled in the same proportion by the sensitizing dyes

This is, with all due respect, meaningless word salad. "Wavelengths handled in the same proportion." What mean does even that LOL having.

DREW WILEY said:
The other must be equivalent to the manufacturer's idea of daylight

Ah, the mother of all f*ups - the assumption. A confused one, at that. In this case, Kodak is actually specific about what the basis for their normalization is:

Erg - Wikipedia

en.wikipedia.org

So Kodak has plotted density against actual exposure energy. Since exposure energy correlates with wavelength and a tungsten light source very closely approximates a black body radiator, you can in fact fairly easily convert the Ilford plot to one comparable to the Kodak plot. Mark J is right. But then again, this is his day job, so he knows what he's talking about.

DREW WILEY said:
dyes typically don't behave as neutral linear profiles

That's where the wavy bits in the relatively flat plateau of the plots originate in. None of that, however, stands in the way of converting the data in one plot to the other. There are (small, practically insignificant for our purpose) methodological problems, but they are in entirely different areas.

I wouldn't be too quick in dismissing what @Mark J said. What he proposes is fairly simple - and quite robust.

Mark J · May 15, 2024

All we are doing here is taking out (reversing) the known effect of a tungsten source at 2850 K.
It's fairly simple maths if you do this sort of thing at work.

Jonathan, I have just created a set of black body data on Excel at work.
Your part of the bargain is to blow up that Ilford density/wavelength graph and get me a set of points every 20nm - that's the tedious bit !

snusmumriken · May 15, 2024

Mark J said:
All we are doing here is taking out (reversing) the known effect of a tungsten source at 2850 K.
It's fairly simple maths if you do this sort of thing at work.

Jonathan, I have just created a set of black body data on Excel at work.
Your part of the bargain is to blow up that Ilford density/wavelength graph and get me a set of points every 20nm - that's the tedious bit !

OK, I'll get onto that straight away. I could have handled the Excel stuff too, I just didn't understand the reasoning for taking the inverse of the 2850K black body data. I've thought about it a bit since, and now I get it!

Also, surely the answer to @DREW WILEY's insistence, that emulsion characteristics need to be taken into account, is that they are taken into account: that's exactly what the Ilford sensitivity curve illustrates.

loccdor · May 15, 2024

Regarding the OP's spectral sensitivity inquiry generally, I prefer the way films with more red sensitivity, like Fomapan 400, render skies in a more contrasty way, as if you used a filter. But you don't incur the loss of light or potential for additional flare that a filter would introduce.

Red sensitive films tend to make skin go paler, green sensitive will make it seem more tan with more contrast to freckles and the like. A note if you do portraits, they used to give actresses lipsticks with odd colors when they shot them on films with less balanced spectral sensitivities.

Check out the portraits below: Fomapan 400 @ 1600 (red sensitive) vs. Ferrania P30 @ 80 (red insensitive). Which renders a face better? I personally like both effects.

Yes, color filters have different factors depending on the spectral sensitivity of a film. It is often on the film datasheets.

Which films make shadows more luminous? Well, outdoor shadows tend to have a blue tinge from the blue sky, so I would think the more blue sensitive film would have an edge there. But I doubt that has as big an impact as the film's latitude.

snusmumriken · May 15, 2024

@Mark J: Here are the numbers for HP5+, for a start. FP4+ to follow later today. Regrettably, Photrio doesn't allow me to attach an Excel file, but hopefully you can copy and paste into Excel. Ignore column 2, obviously.

snusmumriken · May 15, 2024

Sorry, the first line should have read wavelength 355 (not 308), sensitivity 0.

snusmumriken · May 15, 2024

And here are the data for FP4+:

Mark J · May 15, 2024

Thanks Jonathan - you're a star !

On the FP4+ data, is the point where it drops to zero actually 673nm rather than 623 ?

Mark J · May 15, 2024

Note : Since the Kodak data is presented as a 'spectral sensitivity curve', my assumption is that this was done with a monochromator, which provides a (calibrated) flat response vs. wavelength over the whole spectrum. If there was any use of 'Daylight' ( eg. D65 ) in the test, Kodak would have stated that, and they didn't.

Spectral sensitivity and its possible consequences

snusmumriken

Rudeofus

Andrew O'Neill

Moderator

Mark J

albireo

Mark J

snusmumriken

Andrew O'Neill

Moderator

Mark J

Spectral Calculator-Hi-resolution gas spectra

Attachments

Mark J

DREW WILEY

Mark J

DREW WILEY

reddesert

RalphLambrecht

DREW WILEY

koraks

Moderator

Erg - Wikipedia

Mark J

snusmumriken

loccdor

Attachments

snusmumriken

snusmumriken

snusmumriken

Mark J

Mark J

Spectral sensitivity and its possible consequences

Moderator

Moderator

Attachments

Moderator

Attachments

Privacy & Transparency

Privacy & Transparency