Upside down – How color film and paper are fundamentally different

This is a story about chromogenic color film and paper that’s not a secret at all, but it remains untold much of the time. Perhaps because it’s rather technical and it goes beyond what an analog photographer or darkroom printer needs to know. But once you start thinking about the structure of C41 (and ECN2 and E6) film and RA4 paper in particular, it becomes an unavoidable topic. It’s the question how massive color crossover is in these materials. And in RA4 paper in particular, despite the absence of substantial color filtering in the emulsion layers.

Silver-halide based color photography is a bit of a miracle if you think about it. It’s a range of very neat tricks that in combination allow a photochemical process to fairly accurately reproduce real-world colors. One of those tricks has to do with how silver halides are sensitized to different colors of light, and the implications this has for photographic film and paper.

Silver halide is the light sensitive agent, and it is the generic term for silver salt molecules that combine a silver ion and a halide ion, with the halides being chloride, bromide and iodide. So silver halides in actual photographic practice are silver chloride (AgCl), silver bromide (AgBr) and silver iodide (AgI). Generically they are called AgX, with the signifying any halide.

Let’s start with a very basic observation. Silver halides are inherently sensitive to light – but only small wavelengths. Their light sensitivity is limited to UV and a little blue light, with no appreciable sensitivity to green or red light.

For color photography, this means that silver halides as such aren’t very useful – they can only record blue light and as such would only be capable of differentiating colors in terms of “blue” and “not blue”. That’s a bit of a limited palette to work with.

However, there are tricks to make silver halides more sensitive to a broader range of visible light (i.e. color). Firstly, sulfide sensitization boosts overall visible light sensitivity. Secondly, and more importantly in terms of color, it’s possible to adhere dye molecules to silver halide particles. These dye molecules have their own light sensitivity, but they alone don’t actually transform under the influence of visible light in a photographically meaningful way. However, they can interact with the silver halide grains they’re attached to and thus (to put it simply) impart their sensitivity to certain colors of visible light on silver halide grains. These silver halide grains can then be developed into a metallic silver image, and this mechanism is a big part of the basis of color photography.

The problem, however, is that attaching color-sensitizing dyes to silver halide particles changes almost nothing about the inherent UV- and blue-sensitivity of the silver halide grains themselves. In other words, you can sensitize a silver halide to, let’s say, red light, but it will also remain sensitive to blue and especially UV at the same time.

A little detour, but without going into it very deeply at this point: in a color film or paper, the actual color of the material (the negative or the print) is ultimately formed by another set of dyes that are created within the emulsion during development. The red-sensitive emulsion contains the precursor to a cyan dye (cyan being the complimentary color of red) and this dye is actually formed upon development of the image recorded by the red-sensitive silver halide.

This way, a color film or paper really forms an image consisting of the opposite colors of what it sees. Since you’d typically first make a negative (with the camera) and then a print, the two consecutive inversions of the color result ultimately in the same colors of the original scene. I.e. red becomes cyan in the negative, and this cyan image then turns into red again in the final print.

Back to our story about the actual color sensitivity. The overlapping color-sensitivity of dye-sensitized silver halides highlighted above implies that if you were to coat a very simple ‘color’ film with dye-sensitized silver halides for yellow, green and red sensitivity, you would effectively get an emulsion where each part has its desired sensitivity peak, but there will also be considerable overlap, especially due to all emulsion parts being sensitive to blue light.

As a result, a photograph of a red object would render fairly true to life. One of t a green object may render somewhat foul due to a probably crosstalk between the red- and green-sensitive channels. A blue object would become pretty much a desaturated, muddy mess since it would activate all emulsion layers and you’d get the photographic equivalent of the well-known kindergarten experiment of “hey let’s mix all these paints together…!”

Enter another trick: think in terms of layers. A photographic film, and particularly a color film, isn’t an amorphous emulsion on a plastic support. It’s really a stack of thin, discrete layers. That means there’s an order to these layers, and it’s of course no coincidence. By ordering these layers in a smart way, and introducing some essential layers in-between to help along, it’s possible to make a color film that doesn’t exhibit horrible crosstalk between the color channels.

Let’s start by cutting out the unwanted UV sensitivity, at least to a large extent. This happens in the top part of the film, since we want this selectiveness for “everything except UV” to apply to all light sensitive layers below. So a layer with UV-blocking properties is put on top.

Then follows the layer of blue-sensitive emulsion. Light of all colors can fall onto this layer, but since it’s only sensitive to blue (and UV, but we blocked that out), it’ll only respond to blue light. Since the layer is only very thin, most light will simply pass through this layer and only the blue light that happens to hit the silver halide molecules in this layer will actually contribute to image forming in this layer.

Next comes the smart thing: a layer with a filter that allows all light to pass, except blue light. So essentially, all red and green light will pass, but blue light is blocked. This is useful, because the emulsions underneath are naturally blue-sensitive – but as long as no blue light actually reaches them, it won’t affect them either!

So next comes the green-sensitive layer, which will actually see only green and red light. Since it’s only sensitive to green (and blue and UV, but these were blocked out already), it’ll only respond to the green light hitting it, and the red light won’t do anything here. Again, we’re still talking about very thin layers, so there’s ample of red light that passes through.

Next comes another filter layer, and this time, it blocks all green light. Since blue (and UV) was already blocked in a higher filter layer, what’s left now is only red light. At the bottom, there is the red-sensitive layer. As it’s sensitive to blue (and probably to a considerable part of the green/cyan spectrum as well), it would respond to these colors of light just as well. But at this point, there’s no blue or green light left, so it’ll only form an image in response to any red light reaching it.

Put together, it looks like this:

So the selectivity of the different layers in a color film emulsion is a combination of the innate sensitivity of each emulsion layer and the filter layers strategically placed between the image-forming layers. You’ll find the same general concept in all color films (C41, ECN2 and E6) and in fact, it’s been that way since some bright engineers at AGFA figured this all out in the early 1930s.

NB: the presence of a red filter layer is as far as I can tell not necessarily always the case, and it likely depends on how well the red-sensitive emulsion can be made to respond specifically to red light only, and not to green light.

Coincidentally, what happens if you expose the film not from the top, but from the bottom (i.e. through the base)? The incoming light isn’t filtered in this case (apart from tinting of the base and blockage due to the anti-halation layer) and it will all hit the red-sensitive layer. Since this layer is sensitive to red, but also blue light and a little to green light as well, it will respond to, well, pretty much everything that falls through the film. Depending on how much light makes it through the bottom interlayer with the (optional) red filter, the yellow-sensitive layer may also receive a good dose of light that it’s sensitive to – i.e. blue and green light. And if a whole lot of light is available, it will ultimately even make it through to the blue-sensitive layer. This is how ‘redscale’ film works – it’s just regular color film, exposed through the back.

So the above applies to color film. It’s a story that’s told in several places, and I was aware of it for quite some time. Indeed, early color paper such as 1940s Kodacolor also followed the same logic and layer order. But what I personally never realized until fairly recently, is that more modern color RA4 paper is really a different story altogether. The eye-opener for me was a paper by Weaver and Long from 2009 that explains it briefly.

Apparently, the early color papers were plagued by many issues, one of them being mottling. It’s in fact a problem still present in modern color papers, and one of the main reasons I prefer higher-end papers like FUJIFILM DPII instead of the lower-end Crystal Archive. The core of the problem is that the coated paper base of a color paper isn’t perfectly flat. It’s actually a bit wavy, like a very flat landscape of sand dunes. This waviness results in small density differences in the emulsion layers coated on top of that support.

Somewhere in the 1950s, Kodak engineers realized that if you put the red-sensitive, cyan-forming dye layer at the bottom of the layer stack and directly on top of this slightly wavy paper base, the density differences tend to stand out painfully clear. If, however, you flip things over and put the yellow-forming (blue-sensitive) layer at the bottom, the waviness is much less apparent, because the human eye apparently does a relatively poor job discerning density differences in a yellow image, whereas it’s particularly good at spotting them in a cyan image.

An interesting diversion at this point would be a piece about how we see color, and the fact that yellow really is a special case. It would stretch too far to really explore this in depth, but human color vision is, well, really weird if you think about it. And when you do, it becomes clear that yellow is kind of a special case, since it’s the brightest color we see – and oddly enough, it’s a color we actually don’t directly observe, but rather our brain deduces it from seeing red and green at the same time. But that’s a story for (maybe) another day.

Back to our paper issue: flipping around the color layer stack helps to prevent the cyan mottling issue due to an irregularly formed base. But it also means that the neat set of tricks with the layer order and filter layers no longer works. If you put the blue-sensitive layer at the bottom, this means that you have no choice but to allow blue light to pass through both of the other layers as well. And since they are inherently sensitive to blue, this would result in anything of the color blue being rendered as a kindergarten-paint-mix-mess. How come this doesn’t happen, and blues on RA4 paper (and any other color, for that matter) actually look perfectly decent?

Enter another trick: a difference in speed. Color paper has an advantage that film doesn’t have: it can be used under controlled conditions. Film needs to accurately record whatever nature throws at it in the sense of an infinite number of hues and an infinite range of brightness. Regardless if you’re recording a very low-contrast scene with muted colors or a high-contrast, candy cane colored scene, it’ll need to record it as faithfully as possible.

Color paper we can expose in a way that’s convenient – with discontinuous light sources, and at illumination levels that we can more or less freely choose. That’s how Kodak engineers worked around the layer order problem, and in fact more or less did away with the need for a filter layer in the layer stack to begin with. They decided that if blue light needed to penetrate the entire layer stack, the upper (red- and green-sensitive) layers simply shouldn’t care. And that’s possible if you make those layers a lot less sensitive altogether.

So the trick here is that in color paper, the top layer is red-sensitive, but it will just as well respond to blue light. However, it’s also relatively insensitive to light of whatever color compared to the other layers. Small amounts of green and blue light can fall on and through it and it won’t even notice.

The green-sensitive layer is also sensitive to blue light, but not to red light (due to dye sensitization for green light, specifically). And it’s also fairly slow, but a already a lot faster than the red-sensitive layer on top. This means that any green light that is not bright enough to activate the red-sensitive layer (which has no actual shielding against this color of light) will form an image just fine on the somewhat faster green-sensitive layer. But it won’t respond to tiny amounts of blue light that fall up and through it.

The most sensitive layer is the blue-sensitive one, and it’s really fast and trigger-happy when it comes to blue light. But since it’s only sensitive to blue light, the intense red and green light needed to pummel an image from the upper two layers won’t do anything down there anyway.

The difference in speed between the top red-sensitive and bottom blue-sensitive layers is apparently about 5 stops. At least, according to the legendary Ron Mowrey. Looking at the spectral sensitivity curves of RA4 papers, I think the quantitative story is a little more complex than that and depends a lot on which wavelengths are exactly used to expose the paper. But the net result is still that there’s a substantial speed difference that prevents crosstalk between the different layers.

An essential part of this mechanism is the high inherent contrast of RA4 paper. The gamma (which is basically the steepness of the H/D curve) of RA4 paper is around 2.25, which means it’ll produce 2.25 log density for 1 log exposure. Given that the density range of a paper like DPII is actually around 2.1 log, this means that an exposure range of roughly 0.9 will cover the entire density range of the paper. Simply put: it only takes roughly 3 stops to go from paper white to full density on each of the color channels. Kodak paper can be even steeper, depending on the channel, with the steepest red channel having a gamma of as high as 3.0 and the green channel being more similar to Fuji’s at around 2.2:

Anyone who has done any color printing in a darkroom has probably seen hints of the layer order and different sensitivities. For instance, you may have accidentally scratched the surface of a wet print in a spot where the image was really dark or even black. If so, you probably noticed that the edges to those scratches appeared to be yellow or orange. Like so:

The coloration of the scratches makes perfect sense if you realize that the cyan layer is the first to be removed, as it’s on top of the layer stack. This leaves the yellow and magenta layers underneath, which together form a red hue. As more of the magenta and ultimately yellow layers is removed, the color changes from orange (magenta partially removed) to yellow (magenta entirely removed) to pale yellow (yellow partially removed) and ultimately white (no more emulsion left). So such scratches will never show up as blue, green or cyan hues; they will always be in a red – orange – yellow gamut.

On the other hand, chemical attack on the emulsion tends to affect the top (cyan) layer first and only at a later stage penetrate deeper into the stack. I did an experiment some time ago to demonstrate the presence of unused dye couplers in fully processed RA4 paper. I forced the colorless dye couplers to form colored dyes by applying oxidized developer onto the emulsion – a process that remains mostly limited to the cyan layer, because that is on top and therefore the most exposed, and also because it’s not protected by any interlayers that contain scavenging compounds that capture excess oxidized developer from migrating through the emulsion stack. As a result, the color gamut of such damage tends to be the opposite of scratching: it’s cyan, shifting to green and ultimately violet hues:

Now you also understand why color negative film comes with an orange mask. It’s sometimes stated that this mask is there to correct for problems with unwanted light absorption in the dyes of color film. To my best understanding, this is not correct, since it applies to the complementary ‘masking’ dyes that are formed along with the main cyan, magenta and yellow dyes in color negative film. These complementary dyes, however, are invisible to us, since their influence is overshadowed by the main dyes as well as the orange mask.

The actual orange mask serves to filter the light with which RA4 paper (and its predecessors back to the mid-1950s) is historically exposed. The orange-brown color is no coincidence. After all, an orange filter means that red light is transmitted almost unhindered, green is passed partly and blue is mostly blocked. This matches the sensitivity differences of the RA4 emulsion layers exactly: it needs a lot of red light, less green light and only a little blue light in order to form pure and faithful colors.

The nature of the orange mask of color negative film can be spotted in the datasheet of such films, by the way. Look at two particular plots for Kodak Ektar color negative film:

The left plot (characteristic curves) shows the density that’s produced in relation to exposure. All the way to the left of that plot, we find the densities for virtually unexposed film. Note that this density is very low for the red channel, much higher for green, and even higher for blue. In other words: black C41 film passes lots of red, far less green and even less blue light. What does “a lot of red, much less green and only a little blue” look like? Exactly – orange-brown:

Coincidentally, the fact that historically, tungsten light bulbs were used (and often, still are, by darkroom printers) for exposing RA4 color paper also aligns very nicely with the difference in sensitivity of the color layers. After all, the emission spectrum of tungsten bulbs is balanced very heavily towards the red end of the spectrum:

Finally, there’s the phenomenon that when printing a middle-of-the-road color negative, the starting point for filtration settings will be around 40 units of both yellow and magenta (on a regular dichroic filter head), which boils down to blocking a significant amount of blue and green light from the already red-leaning ‘white’ light of a tungsten light source.

Well, so much for chromogenic theory, today. I admit that the practical relevance of all this for the average analog photographer is rather limited. For the experimental among us, all of the above has implications when DIY-ing or modifying light sources for color printing and perhaps scanning, or doing all kinds of experimental torture on color film and paper. For the rest of us, it’s perhaps still an amusing story.