How Photography Was Optimized For White Skin
April 28, 2015 1:35 PM   Subscribe

"the lighter you were, the more likely it was that the camera got your likeness right.” "In film photography, color balance has a lot to do with the chemical composition of the film. For many decades, color film in the United States was calibrated to highlight Caucasian skin tones. This was the most fundamental problem. With an unusual degree of skill and attention, a photographer could compensate for the biases in most stages of production. But there was nothing they could do about the film’s color balance. When the famous New Wave filmmaker Jean Luc Godard was commissioned to make a film about Mozambique, he reportedly refused to use Kodachrome film -- the most popular color film at the time. He complained the film, developed for a predominantly white market, was 'racist.'"
posted by minhrootloop (58 comments total) 51 users marked this as a favorite
 
"Teaching The Camera To See My Skin", a similar article previously on MetaFilter.
posted by jedicus at 1:49 PM on April 28, 2015 [7 favorites]


The camera can only lie.
posted by the man of twists and turns at 1:49 PM on April 28, 2015 [5 favorites]




For anyone wondering, non HDR digital imaging has a similar problem. In 8 and 16 bit images all of the color values are fixed, the spacing of the values is non-linear with the values at the light end of the range being much closer together thanthe values at the low end, essentially meaning there are more values to describe lighter things than there for darker things. This is based on how the eye works, so while it may be racist in practice, I doubt it was intended that way.
posted by doctor_negative at 2:19 PM on April 28, 2015 [2 favorites]


For goodness sake don't read the comments.
posted by asavage at 2:31 PM on April 28, 2015 [19 favorites]


I think you have that backwards-- most commonly used color spaces (e.g., sRGB) are gamma encoded so that more bits are devoted to darker colors than to lighter ones. See the graph on this page: 50% of the range of sRBG is devoted to the darkest 20%.
posted by Pyry at 2:36 PM on April 28, 2015 [4 favorites]


My best man was not Caucasian and doesn't show up in any of our wedding shots. Like just a couple of eyes hanging there in the non-posed shots.

The photographer did tried to explain himself and said "Well, he is kind of dark"
posted by bottlebrushtree at 2:42 PM on April 28, 2015


This article really seems intent on picking an argument. I won't (can't) deny that there had been racial assumptions made in the formulations, color and white balances of film stock from the mid-20th century. Cursory searches on "Shirley Cards" demonstrated some pretty insensitive behaviour by Kodak during that period. But there were also those in the Kodak Research Labs working on the problems both with the film stock as well as the distributed cards. From the NPR article referenced in the OPs link:

In the 1970s, photographer Jim Lyon joined Kodak's first photo tech division and research laboratories. He says the company recognized there was a problem with the all-white Shirley cards.

"I started incorporating black models pretty heavily in our testing, and it caught on very quickly," he says. "It wasn't a big deal, it just seemed like this is the right thing to do. I wasn't attempting to be politically correct. I was just trying to give us a chance of making a better film, one that reproduced everybody's skin tone in an appropriate way."


All that said, during my time in China I struggled with getting skin tones correct when shooting film. There's a huge amount of ethnic diversity in China and I felt it extremely difficult to get consistent, representative results. I'm pretty sure I moved to Fuji Film Velvia on the recommendation of a very good local film processor.
posted by michswiss at 3:02 PM on April 28, 2015 [5 favorites]


I learned a fair bit about photography working for my university newspaper, including turning photos into halftones for publication.

My first experience there was halftoning a bunch of individual student photos (we had a weekly feature where we asked a bunch of students what they thought of the plans for the new student centre/what their plans for spring break were/their choice of best live music venue on campus. etc. and ran a picture along with their name and answer).

Almost the very first thing I was told while doing this was that photos of students of colour presented some technical challenges to the process. The exact wording was "black people can be tricky to halftone."
posted by ricochet biscuit at 3:02 PM on April 28, 2015


Everyone's favorite big box employees, Desi and Wanda.
posted by Tell Me No Lies at 3:02 PM on April 28, 2015 [3 favorites]


The response curve that gives more latitude toward the higher end of exposure is primarily an attempt to not have completely blown-out skies. Even with modern digital cameras, it's common see completely blown-out skies -- basically, they're brighter than the available dynamic range when exposing for a subject in the foreground -- so you just see pure white notihingness, rather than blue with clouds.
posted by the jam at 3:07 PM on April 28, 2015 [4 favorites]


Oh man, the Vaseline thing was real? I just read about that in Percival Everett's Erasure, when a black character is about to go in front of the camera and is Vaselined by the make-up artist; I thought it was a joke!
posted by mittens at 3:44 PM on April 28, 2015


Pyry, you are right, I have it exactly backwards.
posted by doctor_negative at 3:45 PM on April 28, 2015 [2 favorites]


This is why I used German-made Agfa slide film when I photographed pale European musicians and actors for a living. It was good at warming up their skin tones a little, adding a touch of golden Autumn afternoon.

American Kodak slide film (both Ekta and Koda chrome) was very Hollywood, and couldn't handle anyone pasty. It needed a bit of tan or naturally darker skin to work with.

Japanese Fujichrome had wonderful color but it had a slight greenish flavor to its skin tones, flattering for people with Mrs w0mbat's beautiful eastern skin, but not good for other skin tones.

With skin at the darker end of the range it's more a matter of exposure, both in color and black & white. Film isn't perfect so you need to open up half a stop to keep them on the flat part of the s-shaped curve, and you can't rely so much on a single soft light + reflector -- it's better to have the light wrapped around a little more. If film had a bit more dynamic range you wouldn't have to adjust like that.

This is the kind of stuff that photographers think about when trying to get the perfect solo portrait. With diverse people standing next to each other it's more of a compromise to get everyone looking pretty good.
posted by w0mbat at 3:53 PM on April 28, 2015 [9 favorites]


"For anyone wondering, non HDR digital imaging has a similar problem. In 8 and 16 bit images all of the color values are fixed, the spacing of the values is non-linear with the values at the light end of the range being much closer together thanthe values at the low end, essentially meaning there are more values to describe lighter things than there for darker things. This is based on how the eye works, so while it may be racist in practice, I doubt it was intended that way."

Interestingly, cameras (and film) bought in Asia generally have a slightly different color balance to catch the local skin tone better.
posted by klangklangston at 3:53 PM on April 28, 2015 [3 favorites]


ricochet biscuit The other problem is that newsprint soaks up ink like a sponge and has crazy gain. Darker halftones gain a lot more than mid to upper ranges (i.e., white folks).
posted by nathan_teske at 4:02 PM on April 28, 2015 [1 favorite]


For the sake of understanding, (especially having read the awful comments to the original article) I'd like to make an important distinction here as camera wonks will continue adding technical notes to the discussion:

The article is not saying that film or its accompanying techniques were designed to make it difficult to photograph darker skin tones. As we'll hear people continue to talk about balancing for light and the dynamic range involved in photographing subjects in sunlight etc... it's important to realize that the technical side of photographing a wide range of skin tones has a wide variety or origins, yes. BUT, the article is attempting to address a history where those who were interested in photographing or filming people with dark skin tones found very little help from film manufacturers for a depressingly long time. Of particular note in the article, which speaks volumes beyond any discussion of the properties of light, is this passage:
Kathy Connor, an executive at Kodak, told Roth the company didn’t develop a better film for rendering different gradations of brown until economic pressure came from a very different source: Kodak’s professional accounts. Two of their biggest clients were chocolate confectioners, who were dissatisfied with the film’s ability to render the difference between chocolates of different darknesses. “Also,” Connor says, “furniture manufacturers were complaining that stains and wood grains in their advertisement photos were not true to life.”

This kicked off the scientific investigation into rendering different shades of dark brown. As Kodak expanded into global markets, they leverage this technology to develop different emulsions to market in different parts of the globe based on local aesthetic preferences.
It would be one thing to say "therefore the people at kodak are racist." But no one is saying that. But it's important to remember that there simply wasn't that much interest from people to create the techniques and materials for photographing dark skin until it coincided with the demands of chocolate makers.

That lack of interest is disturbing and sad.
posted by shmegegge at 4:06 PM on April 28, 2015 [28 favorites]


My photography professor told me (many years ago) that Kodachrome exported to different parts of the world had subtly different color balance, but that was based on meteorology rather than skin tones - Northern Europe Kodachrome was a little warmer to offset gray weather, while that sent to Mediterranean and North African regions tended a little blue.
posted by tommyD at 4:31 PM on April 28, 2015 [2 favorites]




I think it's worthwhile reading the original paper (Looking at Shirley, the Ultimate Norm) referenced it the article. While I still think most of the observations in the paper have been over played in the article, there's enough information there to do a more extensive search on the topic to form your own view.

tommyD: To your comment, here's a paragraph from the paper:

Ongoing international skin colour preference tests conducted by Kodak have also generated much data useful in informing Kodak film chemists and “designers” of the skin colour biases preferred in different parts of the world. This information has shaped Kodak’s geography of emulsions, which conform to these preferences, rather than to considerations of exact reproduction. The industry term used for this business choice is “optimum reproduction” (Winston, 1996). For this accommodation, film inventory is batched by region and distributed in accordance with these researched preferences. Although there are no explicit signs on Kodak’s film boxes of where each target market is located geographically, the film is coded numerically to indicate countries or regions. So, for example, in 1996, several Kodak sources indicated the regional codes to be as follows:#1–U.S. & Canada; #2–Latin America, Central America, South America, and Mexico; #3–Asia Pacific, China, and Japan (often treated as a separate entity);#4–Europe, Middle East, Africa, India.

So say, you were a Journalist from North America that travelled to Asia, Africa or Europe and didn't buy your film stock locally, you'd likely be carrying an inappropriate tonal bias in your bag in so much as Kodak varied its emulsions.
posted by michswiss at 5:05 PM on April 28, 2015 [1 favorite]


praiseb: I'm certainly thinking it.

The thing about racism that a lot of people forget is that in the vast majority of cases it's unintended and unconscious -- but it's still racism.
posted by flatluigi at 5:11 PM on April 28, 2015 [10 favorites]


The USA (and Australia, Canada, and most other places) used to be explicitly, formally racist. It's inevitable that this racism would be codified in all sorts of assumptions that persist even in the minds of otherwise decent people.1 Saying that the people at Kodak were racist is a statement about their background and society, not their aspirations: it would be totally possible for them to wish to be not-racist but still carry a whole lot of baggage from the way they were raised.

1 But not either of us, of course.
posted by Joe in Australia at 5:15 PM on April 28, 2015 [1 favorite]


Ongoing international skin colour preference tests conducted by Kodak have also generated much data useful in informing Kodak film chemists and “designers” of the skin colour biases preferred in different parts of the world. This information has shaped Kodak’s geography of emulsions, which conform to these preferences, rather than to considerations of exact reproduction.

To me, the interesting part is how openly Kodak was targeting consumers' preferences, rather than "exact reproduction." It makes sense -- if people aren't happy with the photographs, they won't buy your film.
posted by Dip Flash at 5:25 PM on April 28, 2015 [1 favorite]


Eponymorrific!
posted by not_that_epiphanius at 5:31 PM on April 28, 2015


~Almost the very first thing I was told while doing this was that photos of students of colour presented some technical challenges to the process. The exact wording was "black people can be tricky to halftone."

~The other problem is that newsprint soaks up ink like a sponge and has crazy gain. Darker halftones gain a lot more than mid to upper ranges (i.e., white folks).


I do print ads for an area free newspaper, and one of the local colleges regularly places an ad for a series of talks they sponsor, featuring speakers from all over the world, especially Africa. Inevitably, the photos they provide of the speakers are such that I have to spend an inordinate amount of time in Photoshop, adjusting tones and exposures so that we can get an acceptable reproduction of the speakers' faces on newsprint. You're constantly trying really hard not to tread into that "you obviously lightened the black man" territory, but also avoiding the dreaded "picture of black man printed like a big blob of ink" disaster.

The worst days are when you have to work with a picture of two people, who are, inevitably, on the absolute opposite ends of the grayscale. You just stare at the picture and accept your doom.
posted by Thorzdad at 5:38 PM on April 28, 2015 [6 favorites]


This matches my experience, even with a digital camera. I took photos at an all-black wedding as a favor once... I was horrified at how poorly everything came out at the time. It took a lot of tweaking to have people look normal.
posted by MikeWarot at 6:09 PM on April 28, 2015 [1 favorite]


...in so much as Kodak varied its emulsions.

I don't think they purposely varied their emulsion, it seems more likely that they couldn't make exactly the same thing every time. They then just measured how a particular batch varied and shipped it to the appropriate place.
posted by Confess, Fletch at 6:41 PM on April 28, 2015 [1 favorite]


OK, true story: some years back, I attended the World Science Fiction Convention in Anaheim, California. There was a costume event and I took photos of every entry. Being naive, I dropped my film off at my friendly, neighborhood photo shop for development.

Here's the thing: one of the entries was a striking group of blue-skinned elves. When I got my prints back, imagine my surprise that they were white! The photo shop had "color corrected" the hell out my pictures to turn their blue skins white! I have to believe this is some sort of automated process, because no one in their right mind would have done this.

So: imagine what a photo shop is doing to "color correct" YOUR photos!
posted by SPrintF at 7:18 PM on April 28, 2015


The photo shop corrected your prints though - there's not much they could do to correct your film. It's still possible for you to take your prints somewhere better and get the prints you want out of those negatives (or stand behind the photo guys and make them give you what you want)

What this article/post is talking about is biases that are built into the film - these mean that when you make prints from the negatives, some tones look more natural than others. That's fixable to an extent when making prints, but it's Complicated.
posted by RustyBrooks at 7:28 PM on April 28, 2015 [1 favorite]


I rarely see such a large set of mistaken assumptions leading to a pre-selected conclusion, as in that article. There are actually people who study the representation of colors in images using the scientific method. They actually teach this science of "color theory" in places called "art schools." Obviously the author has never heard of this, she seems to have done her research solely on wikipedia. And the misinformation in this thread is nearly as bad.

Photography is a process that maps a range of brightnesses and colors onto another medium. It is highly subjective because it is a process unlike our human vision, which dynamically changes its sensitivity without our being aware of it. The dynamic range of the world is greater than any sensor, biological, chemical, or digital, can capture. The measurement of the transfer functions of these media is called "sensitometry" and there are tools we use like light meters and film densitometers and colorimeters to make accurate assessments of a range of brightness and how it is best represented in a physical media. Here is a short primer on sensitometry of both color and B&W film.

No, I am terribly sorry, but color films are not optimized for representation of caucasian skin. They are optimized for accurate colors over the broadest range of color and brightness, for the most dynamic range. Film has a nonlinear response curve. If you look at a response curve from the tutorial I linked to, there is a representation of a typical response curve. Note that it has curved bits at the extreme ends of the curve, we call that the "heel" and the "toe." But also notice that the majority of the curve is almost completely linear. It is the job of the photographer to expose the film accurately, so the entire dynamic range of the scene is mapped onto the linear part of the curve. If you underexpose, you squeeze all the shadow tones together in the heel. If you overexpose, the highlights are all blown out in the toe.

The reason people have a hard time accurately capturing ANY tone, whether it is skin tones or anything else, is that they suck at photography. They don't use light meters correctly, they don't understand the dynamic range of the subject, or of the media used to capture it. They don't understand how light affects the subject. They don't understand how to see what the light is doing, so they can't represent it accurately.

It is absolutely possible to use color controlled photo systems, in either chemical-based film or digital sensors, and accurately represent a wide diversity of skin tones in a single image. If Goddard thinks Kodachrome is racist because it can't accurately represent the skin tones of Africans, then he is a goddam idiot.
posted by charlie don't surf at 7:33 PM on April 28, 2015 [7 favorites]


Erm. I get this exact same problem shooting black horses, panthers, and many other black furred critters. Border collies are a big pain to shoot in good light - expose for the white, black fur becomes a black splotch. Expose for the black sheen, white goes nuclear.

So, are film makers speciests as well?!

I'm don't propose to understand the technicalities of film, but I find it hard to believe race has anything to do with it. Now, if it was about mass market, sure, that I can get behind.
posted by TrinsicWS at 7:57 PM on April 28, 2015 [1 favorite]


i.e. this is an issue with the dynamic range of the delivery medium, i.e. film/prints.
posted by TrinsicWS at 7:59 PM on April 28, 2015


I am not sure how true it is exactly, but my computer vision instructor was adamant that all humans have skin hues in a very narrow range -- only the lightness is different. It's just skin with different amounts of melanin, not some completely different substance with a different color. If that's true, the real problem is lighting and exposure, not anything wrong with the film itself.

(But I'd be more than willing to believe that the lighting that is typically used for school photos, office IDs, weddings, etc. is optimized for white skin.)
posted by miyabo at 8:53 PM on April 28, 2015


I am not sure how true it is exactly, but my computer vision instructor was adamant that all humans have skin hues in a very narrow range -- only the lightness is different. It's just skin with different amounts of melanin, not some completely different substance with a different color. If that's true, the real problem is lighting and exposure, not anything wrong with the film itself.

That's not entirely true. The color of skin is affected by subsurface scattering. Lighter skin is more translucent than dark skin, it allows more light penetration and the color comes more from reflection off deeper skin structures, which includes more pink and yellows (from blood vessels, fat, and connective tissue). Dark skin coloration comes more from reflection off the melanin in the epidermis, closer to the surface.
posted by charlie don't surf at 9:21 PM on April 28, 2015 [2 favorites]


Here's the thing: one of the entries was a striking group of blue-skinned elves. When I got my prints back, imagine my surprise that they were white! The photo shop had "color corrected" the hell out my pictures to turn their blue skins white! I have to believe this is some sort of automated process, because no one in their right mind would have done this.

There's a good story from the early days of Star Trek about how color tests of a green-skied Orion slave kept coming back from the lab with her looking normal. "Make her greener!" was the resulting order and they kept making the green more and more intense and dark until somebody finally realized the poor lab tech was moving Heaven and Earth to "fix" the color.
posted by LastOfHisKind at 9:23 PM on April 28, 2015 [3 favorites]


The reason people have a hard time accurately capturing ANY tone, whether it is skin tones or anything else, is that they suck at photography.

And yet somehow Kodak created a product such that we idiots who suck at photography could mostly get a decent photo of white skin tones but frequently couldn't get a decent photo of black and dark brown skin tones.

Good photography is difficult. Fine. But this is about the priorities of the companies making film, which difficult things they used their technology to make easier and which difficult things remained difficult because they weren't a priority.
posted by straight at 10:32 PM on April 28, 2015 [6 favorites]


I remember seeing this image about the difficulties of interracial selfies, which illustrates this problem as it applies to popular consumer cameras (as opposed to professional ones).
posted by NoraReed at 11:19 PM on April 28, 2015 [1 favorite]


I remember seeing this image about the difficulties of interracial selfies, which illustrates this problem as it applies to popular consumer cameras (as opposed to professional ones).

They managed to find lighting conditions where no camera settings will look good for either of them. Seriously, if you cut each of those pictures in half, none of the 8 are exposed well.
posted by aubilenon at 11:55 PM on April 28, 2015 [3 favorites]


"To me, the interesting part is how openly Kodak was targeting consumers' preferences, rather than "exact reproduction.""

Exact reproduction doesn't exist. Even reproduction faithful to what he eye sees doesn't exist.

"No, I am terribly sorry, but color films are not optimized for representation of caucasian skin. "
The new PORTRA 160 features a significantly finer grain structure for improved scanning and enlargement capability in today’s workflow. Choose PORTRA 160 to deliver exceptionally smooth and natural skin tone reproduction, the hallmark of the KODAK PROFESSIONAL PORTRA Film Family. It's the ideal choice for portrait, fashion and commercial photography — whether in the studio or on location.
The idea that most color film, especially consumer film, is not designed to get the best reproduction of skin tones is bullshit and you should know better. The "neutral skin tones" used as their model for the majority of Kodak's history were Caucasian.

If you're going to go off on people not knowing about film response curves, you should know enough to remember that the frequency response curves for the film (hence color balance) is a separate question from the overall lux/density exposure question. At the very least, you should remember that Fuji usually has a cool cast and Kodak usually has a warm one. The fact sheets for color films all include dye sensitivity information based on color temperature.

And that picture you linked to? Weird one to use here — the whole picture's color balance is off. The dude on the right looks like a fucking Oompa Loompa.
posted by klangklangston at 2:51 AM on April 29, 2015 [6 favorites]


I thought it was a weird picture to use because there aren't any black people in that picture.
posted by ROU_Xenophobe at 6:24 AM on April 29, 2015 [3 favorites]


There's a good story from the early days of Star Trek about how color tests of a green-skied Orion slave kept coming back from the lab with her looking normal. "Make her greener!" was the resulting order and they kept making the green more and more intense and dark until somebody finally realized the poor lab tech was moving Heaven and Earth to "fix" the color.

Yeah, I found it funny that SPrintF's story was at a science fiction convention -- as soon as I read it I thought of Susan Oliver being colour-corrected back to her standard flesh tones.
posted by ricochet biscuit at 7:51 AM on April 29, 2015


. If Goddard(sic) thinks Kodachrome is racist because it can't accurately represent the skin tones of Africans, then he is a goddam idiot.

Yes, I trust that you know more about film than someone who's only been directing movies for the last sixty years.
posted by octothorpe at 8:21 AM on April 29, 2015 [1 favorite]


I thought it was a weird picture to use because there aren't any black people in that picture.

The guy in the middle certainly is. They look a bit odd because they're all covered in wind-blown clay dirt.

Look, do I really need to beat this to death? You guys are so goddam skeptical that you won't even believe your eyes. I guess I do have to hammer it home.

I chose photographs taken during the early years of Kodachrome, when presumably it was still most challenging to make chemically consistent, accurate photo emulsions. Most of these (like the one I already posted) were taken on Kodachrome 4x5 sheet film. Let's look at a few more.

This kid seems to have fairly well-represented skin tones, as well as accurate highlights that are not blown out.

This portrait seems to represent very dark skin with some accuracy, although the woman's face is in shadow. This is the most challenging lighting condition to reproduce, when the majority of the subject is lit by direct sunlight at mid-day, but some of the subject is in shadow. As I said before, everything depends on accurate exposure.

The idea that most color film, especially consumer film, is not designed to get the best reproduction of skin tones is bullshit and you should know better. The "neutral skin tones" used as their model for the majority of Kodak's history were Caucasian.

There is no such thing as "consumer film." Film is all made with the same process, coating a plastic substrate with thin layers of photochemical emulsion. The color sensitivity is optimized to be accurate across the entire color gamut. Even the earliest "consumer" color films were used as a technical recording medium for scientific photography.

There is a huge mythology over the representation of skin tones in film, for example the "China Girls" test images, the models you are presumably referring to. The reason these caucasian women were chosen is because their tones are the most difficult to render accurately, far more difficult than dark skin. This is a mere technical detail of color photography, a small color shift in very pale tones is easily visible, where a small color shift in dark tones is more difficult to detect. It was fairly common for test models to be extremely fair skinned, even some pale redheads, just to make it harder to accurately capture. But you might notice that all these test images have color swatches in them, to check the accuracy of tones other than flesh color.

No, even your Portra film blurb is just marketing crap. The film is marketed at portrait photographers, it says it is designed for accurate skin tones, it doesn't specifically say caucasian skin tones. We can look at the tech data sheet for the film, on page 5 it shows the sensitometry data. It's a C-41 film, it produces color negatives, not transparencies. If you look at the Characteristic Curves, it shows a pretty standard curve, with a distinct heel and almost no dropoff at the toe. The spectral sensitivity curve is most particularly relevant. It shows the sensitivity range of each of the three film layers. There is some variation in sensitivity of each layer, the whole trick of color film is optimizing them so that as sensitivity to one frequency range drops off, the next layer sensitivity begins. But you may also notice in the other curves, that the response to red light is lowest. This is a problem known since before color photography, early B&W films were usually orthochromatic, sensitive primarily to red light, or panchromatic, sensitive to all frequencies. Film emulsions are sensitive to different colors of light in different ways, and the different colors of light carry different amounts of energy. This is a very complex subject.

But let's compare the curves to the data sheet for Ektar 100, which is marketed without any reference to skin tones:

Featuring ISO 100 speed, high saturation and ultra-vivid color, EKTAR 100 offers the finest, smoothest grain of any color negative film available today.

Ideal for scanning, and offers extraordinary enlargement capability from a 35mm negative. A perfect choice for commercial photographers and advanced amateurs.


Even the demo is a landscape photo with no human skin tones.

But again, let's look at the curves. Let's directly compare them by overlaying the curves from the two different films. This is a little tricky since Kodak stupidly changed the scale of the Y axis, so I had to match the scales by rescaling them in Photoshop. Surely this will be a bit inaccurate, but good enough for a rough comparison. And you can see the films are almost exactly the same, except for the Cyan layer. Now remember this is color negative film, so the Cyan layer is capturing the red tones. Red is the basic color of pinkish caucasian skin, but hey wait a minute, the Ektar film has a broader response curve than the Portra film intended for portraiture. Technically, the Ektar film would be a better choice for representing caucasian skin tones accurately, the Portra film is less sensitive to pink skin tones. And yet, Kodak is marketing the Portra film for portraits, even though it is less accurate. Maybe there is some art to this, as well as science.

Yes, I trust that you know more about film than someone who's only been directing movies for the last sixty years.

Why yes, actually, I do. Goddard is a director, not a cinematographer. I was a professional film tech in a Hollywood graphics studio, and I went to art school and studied color theory, film sensitometry, photochemistry, and technical aspects of color reproduction. My current project involves making my own color printing paper, I make my own color film emulsions and coat the chemicals onto the paper by hand. I have very specific and detailed technical expertise in this particular subject.
posted by charlie don't surf at 8:50 AM on April 29, 2015 [7 favorites]


For what it's worth, I actually find paler people much easier to overexpose when shooting digital - I honestly can't speak to film at all, but in digital I find darker skin tones easier to work with, or at least it gives me way more latitude, because it's far easier to lose detail on overexposure than underexposure with today's high-end cameras and a lot of detail can be recovered in RAW.

Each film had its own tonal response. I honestly find it hard to believe that of the many, many, many varieties of film, there wasn't a single one that performed well with darker skin tones. I doubt many posters here could list every single variety of film that was produced 1960-2000.

However, I would find it easy to believe that the most commonly used, popular film stocks of the time might not perform as well. I would also find it easy to believe that a lot of photographers would botch their results.

One thing that's nice to keep in mind about digital photography is that it no longer requires photographers to be chemists or rely on chemists to develop their prints properly. I think we've all seen shots from pros that aren't quite there, technically. It's more common than you might think. So while unconscious bias could easily be a factor from the film manufacturers, couldn't inexperience be a factor from the photographers? It's not like we look back on school photos and think they're the best that could be done under the circumstances. Even today's digital cameras can only expose a certain range of values "perfectly" and once you get beyond that range, you're exposing for two different shades. Pale people and darker people are two fundamentally different tones, exposure-wise. You can't get both of them "perfect" on a single in-camera exposure, without varying lighting levels or camera settings.

*For purposes of argument, let's just say that "perfect" means the camera's metering says it's exposed properly - which usually means what the camera thinks is the right exposure for an 18% gray card.
posted by Strudel at 9:31 AM on April 29, 2015 [1 favorite]


couldn't inexperience be a factor from the photographers?

That is probably the primary problem. Some of the Kodachrome photos I linked to were done by documentary photographers from the Farm Services Administration, they were specifically tasked with documenting poor farmers so they certainly had more experience than most, at photographing minorities with diverse skin types.

I doubt many posters here could list every single variety of film that was produced 1960-2000.

They used to make annual books that had the tech sheets of every film on the market, color & B&W. I have one around here somewhere, the last time I bought one was in the late 80s. But now you just look up the data sheets online, or it is included with pro film stocks. This data was something every film tech was expected to be able to analyze.

Pale people and darker people are two fundamentally different tones, exposure-wise. You can't get both of them "perfect" on a single in-camera exposure, without varying lighting levels or camera settings.

*For purposes of argument, let's just say that "perfect" means the camera's metering says it's exposed properly - which usually means what the camera thinks is the right exposure for an 18% gray card.


I am saying just the opposite, you can get both light and dark skin exposed well in the same shot. Your 18% gray card is a quick trick, I used it a lot, but it only puts one point on the right spot in the film's response curve. If you use standard light meters to measure skin tone, you can easily put both dark and light skin at the exact same brightness. Each skin tone will turn out "middle grey," each photo will look fine, taken by itself, since the brightnesses are all correct relative to each other. But you need to make some assessments by eye and decide what the proper brightness really is, and compensate for it.

This problem is (mostly) solved by using the Zone System, which usually involves using spot meters to measure multiple points in a scene, so you can pin several different brightnesses to the film response curve. For example, average caucasian skin is Zone VI, lighter skin is Zone VII. Yes, the Zone System has specific calibration points for caucasian skin, because it was invented by a caucasian photographer, and most photographers were caucasians. We used it because we could spot meter our own skin for an index value, and we are pretty familiar with our own skin, it doesn't change much unless we get a dark tan.

But if you meter accurately in the Zone System, and get two different spot meter values, you can pin the scene brightness to the entire curve accurately. More technical photogs (like me) would use the Zone System to customize the film's development to give the best film curves, which was a lot easier when you were shooting sheet film and could develop each sheet differently. I personally used to use a spot meter and when I was shooting outdoors, I'd read a gray card in direct sunlight for Zone V, and then use a shaded ambient incident meter reading to get Zone IV. That way, I was exposing for the shadows, I had two data points. But in order for this to work, you had to do extensive calibration shots with your camera on different films, with different developing methods, oh it was a huge pain in the ass. But if you could do this, you got what we called a "Perfect Print." Every single brightness in the photo was perfectly matched to every brightness in the scene.

But nowadays this is all much simpler. Most serious photographers use something like an X-Rite ColorChecker. It's a set of swatches that covers the entire gamut. They are a known standard that you can use to calibrate your camera, and there are simple digital tools so you can calibrate your individual photos to match the swatches, you can recover them a bit even if they are poorly exposed. But you can only go so far. The dynamic range of digital photo sensors is not inherently better than film, they are pretty evenly matched unless you are using expensive HDR sensors. Most HDR systems combine exposures from bracketed exposures. Some advanced photogs will manually composite different shots, to combine the best exposed parts. This is more of a landscape technique since it requires you to lock down your camera on a tripod.

BTW one catch-up remark..

Japanese Fujichrome had wonderful color but it had a slight greenish flavor to its skin tones, flattering for people with Mrs w0mbat's beautiful eastern skin, but not good for other skin tones.

Fuji films are kind of an exception. Unlike everyone else that produces tricolor film, some Fuji films have four layers, two of them cover the green/blue spectrum (e.g. Figure 15 here). I believe this was done not so much for enhancing asian skin tones, but merely as a monopolistic move, since you had to use Fuji developing systems specifically for their films. But other tricolor films from manufacturers like Agfa could all be developed with the same chemicals.

My personal preference was to shoot Agfa color films. Agfa films were much better at producing accurate colors. I believe this was mostly because their film emulsion coatings were made with more precision, and the Germans were always way ahead of the rest of the world in dye and pigment technology.
posted by charlie don't surf at 10:14 AM on April 29, 2015 [1 favorite]


The guy in the middle certainly is. They look a bit odd because they're all covered in wind-blown clay dirt.

I hear your larger point about kodachrome and brown skin -- I was originally going to link to a different example that I found -- but that guy in the middle is just a grubby white dude. Most obviously because you can see that he stops being "black" where his neck meets his chest, but also simply because the probability of an integrated (as opposed to all black) tank crew in 1942 was just about zero.
posted by ROU_Xenophobe at 10:54 AM on April 29, 2015


Charlie, you're going from "talking about highly technical photographic technique" to "making shit up."

Godard was a cinematographer. He understood cameras on a technical level, photography and cinematography on a creative level, and the art of photographing with deliberate goals at a level that neither you nor I will ever come close to.

Beyond that, Kodak themselves eventually acknowledged that they didn't put effort into creating film that represented various brown tones well until chocolate makers demanded it.

You can blame everything on the cinematographer as much as you want, but you're simple and completely plainly wrong, and in your attempts to hang on to this point you've stopped representing history honestly.
posted by shmegegge at 10:58 AM on April 29, 2015 [1 favorite]


Oops, an editing note:

orthochromatic, sensitive primarily to red light

Ortho film is insensitive to red light. I apologize for the edit error.

On edit:

Beyond that, Kodak themselves eventually acknowledged that they didn't put effort into creating film that represented various brown tones well until chocolate makers demanded it.

citation needed
posted by charlie don't surf at 11:36 AM on April 29, 2015


... it's in the article.
posted by shmegegge at 12:32 PM on April 29, 2015 [2 favorites]


From the paper the article is derived from:


Kodak’s drive to increase the dynamic range of its film products was motivated by two other (seemingly irrelevant) issues. These had to do with the photography of brown objects. Here is Kathy Connor’s description of the experience of Earl Kage, former head of the Color Photo Studio at Kodak Park in the 60s and70s, and former Manager of Kodak Research studios:
Well, he said that it was interesting, that in the mid-sixties and seventies there was a coincidental problem that the company was facing. Two of their biggest professional accounts were, he didn’t name the company, but somebody said that they made chocolate candies....Apparently, in reproducing chocolate candies, Kodak was receiving complaints that they weren’t getting the right brown tones on the chocolates. Also, furniture manufacturers were complaining that stains and wood grains in their advertisement photos were not true to life, and that they weren’t appropriate, so the chemists did some work on that. Earl also said to acertain extent, that research to improve those professional markets and addressing their questions helped them to do a little bit better with ethnic skin colours. I was amazed. (Kathy Connor, Executive, Kodak, Rochester, NY, personal communication, August 16, 1995)

In his own words, Earl Kage remarked to me: In the 4 x 5, 5 x 7, or even 8 x 10 colour transparency area that manufacturers of furniture were using to display their wares and to advertise their furniture for catalogues, they were having a good deal of difficulty indemonstrating the subtle differences of certain woods. Now, whether it was maple vs. oak vs. a couple of dark woods, this couldn’t be distinguished in the photographs. This was also about the same time that we got some interesting observations from chocolate manufacturers who, in displaying Whitman’s chocolate or whatever the names were in any case, the subtle variations between the dark and bittersweet and milk chocolates weren’t as discernible and so some modifications were tried and consequently my little department became quite fat with chocolate, because what was in the front of the camera was consumed at the end of the shoot. (Earl Kage, former Manager, Kodak Research Studios,Rochester, NY, personal communication, August 21, 1995)


I posted this earlier, but here it is again to bring them together. It's another quote from the referenced NPR article from a technician that possibly worked in the same group in the 1970's. I dare say Jim Lyon and Earl Cage probably knew each other.

In the 1970s, photographer Jim Lyon joined Kodak's first photo tech division and research laboratories. He says the company recognized there was a problem with the all-white Shirley cards.
"I started incorporating black models pretty heavily in our testing, and it caught on very quickly," he says. "It wasn't a big deal, it just seemed like this is the right thing to do. I wasn't attempting to be politically correct. I was just trying to give us a chance of making a better film, one that reproduced everybody's skin tone in an appropriate way."


The problem in my mind is that these are both anecdotal accounts of personal experiences working on particular assignments. Which is consistent with my experience and understanding of how a lot of advancements in photography in the early days came about.

I still feel the opening statement in the first quote regarding "Kodak's drive" is disingenuous. It's the author's viewpoint, fair enough, but it doesn't seem to be backed up with anything other than a couple of anecdotal memories of former employees.
posted by michswiss at 1:18 PM on April 29, 2015


LOC index page for the tank crew photo.
posted by XMLicious at 1:42 PM on April 29, 2015


"There is no such thing as "consumer film." Film is all made with the same process, coating a plastic substrate with thin layers of photochemical emulsion. The color sensitivity is optimized to be accurate across the entire color gamut. Even the earliest "consumer" color films were used as a technical recording medium for scientific photography."

Man, stop before you embarrass yourself further. There is such a thing as consumer film, with consumer film generally made with broader exposure latitude and longer shelf life at the expense of grain and color reproduction. Portra is pro grade; Kodak Gold is consumer grade. Just admit you're wrong and move along.

"There is a huge mythology over the representation of skin tones in film, for example the "China Girls" test images, the models you are presumably referring to. The reason these caucasian women were chosen is because their tones are the most difficult to render accurately, far more difficult than dark skin. This is a mere technical detail of color photography, a small color shift in very pale tones is easily visible, where a small color shift in dark tones is more difficult to detect. It was fairly common for test models to be extremely fair skinned, even some pale redheads, just to make it harder to accurately capture. But you might notice that all these test images have color swatches in them, to check the accuracy of tones other than flesh color."

If the reference is caucasian skin, that does mean that caucasian skin tones are the reference. Jesus, man. And the Ektar is optimized for shooting high contrast, no grain in stable lighting. Of course they're going to use a landscape to demonstrate that.

"The spectral sensitivity curve is most particularly relevant."

Yeah, I know, chief. The dye density is also relevant for printing or projecting.

"Technically, the Ektar film would be a better choice for representing caucasian skin tones accurately, the Portra film is less sensitive to pink skin tones. And yet, Kodak is marketing the Portra film for portraits, even though it is less accurate. Maybe there is some art to this, as well as science."

Portra is better for skin tones under a broader set of lighting conditions, in part because Ektar can oversaturate the colors (including skintones) and has a lower exposure latitude. Higher contrast can also be unflattering to skin tones. Why, it's almost as if Portra was optimized for shooting (white) skin tones in professional portraiture and the question of pure reproductive accuracy is the wrong one!

"Yes, the Zone System has specific calibration points for caucasian skin, because it was invented by a caucasian photographer, and most photographers were caucasians."

That's the whole goddamned point of the article.

"Most obviously because you can see that he stops being "black" where his neck meets his chest, but also simply because the probability of an integrated (as opposed to all black) tank crew in 1942 was just about zero."

The first African American tank crews were constituted in June 1941, and trained at Ft. Knox, Ky. And while they were an M5 light tank battalion, the M5 and M4 were both being produced then and both trained at Knox, so it's not implausible on its face that African American tank crews would be photographed. I also know that most African American units had white officers, in part because my grandfather started the war as a white officer training black transportation corps.
posted by klangklangston at 2:22 PM on April 29, 2015 [2 favorites]


The problem with chocolate really is a matter of subsurface scattering again. Waxy materials like chocolate are slightly translucent so they let light penetrate and reflect back out through the subsurface layers.

Also some browns are terribly difficult to reproduce in tricolor, or worse, CMYK printing like in offset printed advertisements. In oil painting, my favorite browns come from Cobalt Violet (very expensive) and Cadmium Yellow, both are very intense pigments and there really is no way to reproduce it photographically, since there is no primary purple in the CMY or RGB systems. Purple is just a pain to reproduce, for example there are purple gemstones that just cannot be photographed on film, they turn out blue. But let's stop there because I don't want to get into spectroscopy.

The wood stain is a perpetual problem even today. I recall one of my most difficult prepress jobs ever, a photographer came in with a shot of polished wood floors that was lit from behind through a bright window, so the sun was reflecting right into the camera. The floor was almost all highlights, where the sun was most intense it was pure white. Most of the colors were very pale, so when they were color separated, they came out in CMYK like 10:7:8:0. And even worse, there were gradations in color from very small values, like from 8Y to 6Y, so you could see banding on the yellow plate. When you have such small values, the slightest problem (like dot gain on the printing press) causes very noticeable shifts in color. I had to darken the whole image and fiddle with the contrast to get the scene to look realistic, so the brightnesses were all correct relative to each other. It ended up quite unlike the photograph, although the photographer liked the result. When you have such light tones, it really is very difficult to reproduce in any medium without fudging it considerably. And note that these light tones are very similar to caucasian skin colors. Once again I reiterate: test images of caucasian women are used because it is very easy to detect small color deviations in pale skin colors.

The dark tones (like the chocolate) are at the other end of the problem. The usual approach to fixing this as a prepress job is with a "touch plate." The easiest way to explain this is with oranges. Bright orange is beyond the color gamut of CMYK. So if you're printing pictures of oranges, you can add a fifth color, CMYK + orange, and the orange plate is only used to print the intense orange on the fruit, nowhere else. This is called "arbitrary color separation" and is most common in fine art printmaking, where it is common to have 20 or more colors. But this is way beyond photographic film.

Anyway, I read through that paper "Looking at Shirley" and that is the worst bunkum I could imagine. There isn't one single iota of quantitative data, there isn't even any qualitative data. It's a rant about "cognitive equity" and "flesh tone imperialism." I'll just tell you the point where you can tell this entire paper is ridiculous crap. On page 188, figure 4, there is a photo showing two people, an attempt to demonstrate the author's point, "An example of the challenge of photographing highly contrasted skin colours in the same frame and the difficulty of later recognizing who is in the photo." Sorry, that is a black & white photo, it has no color (other than an overall sepia tone) and it is badly printed on high contrast paper. There are no midtones whatsoever, just black, and white. The photo is a vast exaggeration, reductio ad absurdum. A competent photographer can use a high constrast effect even on dark skinned persons and they are still easily recognizable.
posted by charlie don't surf at 2:33 PM on April 29, 2015


Oh man here we go again.

Man, stop before you embarrass yourself further. There is such a thing as consumer film, with consumer film generally made with broader exposure latitude and longer shelf life at the expense of grain and color reproduction.

You know what I mean. They are all made with the same process, on the same machines, using the same photochemistry and the same dyes, and scientists have been using "consumer" grade films like Kodachrome for scientific photography ever since it was released. I know researchers that used Kodachrome up until it was discontinued, long after it became impractical compared to digital, because they had fifty years of Kodachrome slides they made, they didn't want to recalibrate their whole process.

The biggest difference is that pro films have a shorter expiration date, so you're not buying film that has been sitting on a shelf in the hot sun for ten years, and that you can buy film in bulk from a single batch so you can calibrate the film carefully, to compensate for any miniscule variations in film sensitivity from batch to batch.

If the reference is caucasian skin, that does mean that caucasian skin tones are the reference.

You didn't even read the paper I referenced. Caucasian skin tones are a single point of reference. There are ten basic tones in the Zone System. Us white people all look alike, we generally have very similar skin tones, but there is more variation in darker skinned people so you have to pick what Zone to put them in. You know this crap, why are you disputing it? Maybe you don't know it, hardly anyone uses the Zone System.

The first African American tank crews were constituted in June 1941, and trained at Ft. Knox, Ky. And while they were an M5 light tank battalion, the M5 and M4 were both being produced then and both trained at Knox, so it's not implausible on its face that African American tank crews would be photographed.

Well now at least we agree on some basic facts. These Federal photography projects were tasked with documenting the diversity of Americans, and often went out of their way to include under-represented people like for example, this woman.

BTW I hate those watermarks copyright Library of Congress. No, those photos are Public Domain because they are Federal work products.
posted by charlie don't surf at 3:03 PM on April 29, 2015


Went to a talk on implicit bias today, and one of the speakers mentioned this: HP Face-Tracking Webcams Don't Recognize Black People
posted by larrybob at 3:55 PM on April 29, 2015 [1 favorite]


Zone system aside, I was referring to the camera's metering system, not an actual gray card or actual exposure. With the same setting, on a contemporary DSLR, you cannot use its built in metering and get the same exact exposure reading for two wildly different skin tones. Which, like you said, is why people use more expert metering schemes. But the kind of photography most people are likely to run into on a daily basis, be it smartphones, or graduation photographers, is unlikely to use more elaborate metering than the built in DSLR's meter.

I'm talking about the actual readout on the hardware here, not what the most technically proficient user can produce. I don't actually use a gray card when I shoot, and I have a ColorChecker Passport. To my knowledge, (and feel free to correct me on this, it's what I've read) The hardware is designed to meter for 18% gray. 18% gray is two different exposures when the camera is pointed at two very different skin tones. As you pointed out, this is not optimal.

But it's what the default hardware that most people are using tells them, regardless of whether or not they listen to it.
posted by Strudel at 12:09 AM on May 1, 2015 [1 favorite]


This isn't like the olden days of SLRs that had a TTL sensor that looked at the average of the entire scene. I have never owned a serious DSLR but metering in most of them essentially use spot metering, looking at multiple points inside the frame and doing some fancy evaluation to optimize the dynamic range. Software from big compaines like Canon tend to have one standard camera OS and they use it even on simpler cameras, without all the high end features but usually with evaluative metering. Someone asked me to take a pic of them with their new pocket-sized Canon, I could see it did face recognition so it presumably metered for that, and I could see little markers where it was metering other spots. I don't know much about that, my only digital camera is from 1996, it's so stupid I actually do use a grey card when metering. I did a huge catalog shoot for a T shirt company with my antique digital camera, which is easy when you can control the lighting. And yeah, I basically did run into the same problems this FPP alleges, I had black t shirts with dark details that were hard to capture without good lighting and good exposure, but once you dial in the right settings, it captures the shadow tones as well as it captures the bright white t shirts. I just set up the lights, did some test shots of test targets, and locked it into manual exposure mode and kept the same exposure for the whole shoot.
posted by charlie don't surf at 9:59 AM on May 1, 2015


Most DSLRs allow you to do custom point metering, but often make you go deep into the programable functions for it. I could do it on my Pentax K10, but it'd take a good half hour fiddling with sub-menus or I could just use one of the handheld meters I've got and set the DSLR to manual.
posted by klangklangston at 6:55 PM on May 1, 2015


« Older “Has the Department of Education learned nothing?”   |   What it Says on the Tin, Collect Cats Newer »


This thread has been archived and is closed to new comments