I posted this, and definitely wasn't thinking of it as clickbait, or even as controversial. As Dan says, I think this history is both intellectually interesting and relevant to my work as an engineer, today.
Over time, as the technology of photography was invented and iteratively improved, film formulations came to work better for taking photographs of light-skinned people than of dark-skinned people. Inventors and engineers didn't set out to disadvantage dark-skinned people specifically. But it happened anyway.
In the relatively short history of commercial "AI", we've already seen examples of machine learning systems that followed a similar development path. Non-representative data sets[0], replicating existing bias[1], over-fitting data sets to the point where the outcomes violate non-discrimination laws[2], etc.
One way to help avoid making these mistakes is to think about how they were made in the past, in other engineering contexts.
In the days of film photography this was an important issue - negative/slide film chemistry had a huge impact on the output of your work.
I used to shoot a lot of Fuji Velvia due to its excellent contrast and reasonable resolving power. It was terrible for portraits, though, of any race - it picked up too much red in both light and dark skin tones, and the result rarely looked good. However I definitely think that it was true back then that more portrait orientated film chemistry was poorly optimised for those of African descent or with similar skin tones.
I'm less certain it's such an issue today. Some cameras have face recognition built in as a focus/af assist, and I suspect that these are better with some faces than others (most consumer cameras are developed in asia so I'd except them to do best with asian faces, but as I don't use this sort of feature I don't know). Other than that, I'm struggling to come up with a way that bias - in the way I'd define it - could creep in. In-body JPEG colour profile perhaps, but most serious photographers don't depend on that, and it can be rectified in post anyway.
On the other hand, there's the unfortunate technical problem that dynamic range is limited, and having a subject with very large contrast between skin and clothing brightness is technically challenging to shoot, especially in a situation like the one illustrated where the subject can move around so you have limited fine grained control over lighting.
I'm totally on board with presenting the former film chemistry issue as bias - but that's pretty much just historical at this point. The number of people shooting events on slide film is pretty close to zero these days. There's also very little if any investment into film chemistry as a consequence of this.
The second issue irks me a bit. If you have very dark skin and wear very light clothing, or very light skin and wear very dark clothing then it's going to be more of a challenge to capture you effectively. That's physics not racism, and conflating the two issues feels a bit disingenuous to me.
True. So maybe skin darkening would be more practical than CV Dazzle, because it wouldn't be so unusual. Horrible legacy and PR, though.
Anyway, I suspect that was an evolutionary advantage. For hunting in forest and savannah, to hide in shadows. The human rootstock was black, I gather. And lighter skin evolved where soft UV was limiting for vitamin D synthesis.
But that's solvable with the right technology. Maybe something like getting overall color value, and then tweaking color contrast. If we can do it in astronomy, we can do it with people.
I would guess that the evolutionary advantage of black skin was not primarily camouflage. Tool-using humans of any colour can do that with mud and grass.
My bet is on UV tolerance. Black people have lower rates of skin cancer in the US compared to whites. https://www.jaad.org/article/S0190-9622(05)02730-1/fulltext (Mortality is higher in blacks, but that's because it's diagnosed at later more severe stages.) The effect would have been far bigger when foraging and hunting all day every day under the equatorial sun.
Or perhaps light skin was an evolutionary development as humans moved to higher latitudes in order to obtain high enough levels of vitamin D. See [1]: The evolutionary significance of vitamin D, skin pigment, and ultraviolet light.
The genetic history of human skin colour is really amazing. All our ancestor had fair skin because of body hair (Chimps have fairish skin under all their body hair), we then became dark skinned as we lost most of our body hair, and then those of us who had ancestors that lived far from the equator became fair again.
As someone who has the ancestry of people who lived far from the equator and now lives in a country that is much closer to the equator I wish we hadn't lost our body hair.
Yes if you are covered with hair you not only don’t need dark skin for UV protection, but it will likely interfere with your ability to synthesise vitamin D. Also why waste energy synthesising something you don’t need (melanin).
Dark-skinned faces are inherently more difficult to photograph. Dark skin is in the lowest range of values in most images, but the specular highlights of that skin can reach the highest range of values. Underexpose and you lose all shading; overexpose and you lose the geometric definition given by specular highlights.
Pale skin can be made easier to photograph in challenging lighting conditions by using a matte powder, which diffuses specular highlights and reduces the value range. This doesn't work well on dark skin, because subtle shadows are less visible and you lose the impression of shape; you need the specular highlights, but they greatly increase the value range.
The problem with early colour films wasn't really hue (all human skin falls into a remarkably narrow range of hues), but a lack of sensitivity and dynamic range compared to monochrome films. They coped with over-exposure reasonably well, but even slight under-exposure would obliterate the detail on dark skin. This limitation of chemistry was a constant problem in all forms of colour photography, but the social burden fell disproportionately on dark-skinned portraiture subjects. Later emulsions had far better rendition of lowlights and smooth compression of highlights, which vastly improved their performance with dark skin.
I just want to note that the technical causes you list here are, arguably, totally beside the point.
All technical systems have strengths and weaknesses. When systems are conceived and developed their goals are set through a culturally-created understanding of what the system "should" do. The lack of attention to darker faces (from chemical process selection to selection of people depicted in kodak's system calibration cards) explains how we got this particular set of technical limitations instead of one that, say, tended to over-expose easily.
> Dark-skinned faces are inherently more difficult to photograph
I would say that dark-skinned faces are inherently poorly captured by the photographic technology we have developed, which has focused on detail rather than dynamic range.
>selection of people depicted in kodak's system calibration cards
The choice of the model does reflect some amount of bias, but it doesn't cause bias in the printing process.
Precise calibration was done with a test chart and a densitometer. Shirley was just a quick reference used throughout the day, to check at a glance whether prints were unnaturally blue or purple or washed out due to changes in the developer chemistry. Fuji usually supplied test films and prints with Asian models, even in European markets; the motion picture industry originally used porcelain dolls. A multi-ethnic test image wouldn't have provided better accuracy.
>I would say that dark-skinned faces are inherently poorly captured by the photographic technology we have developed, which has focused on detail rather than dynamic range.
Except that the opposite was true for the entire duration of the film era. The photographic film industry was acutely aware of the fact that most consumer photographs were taken on compact cameras with mediocre lenses and usually printed at 6"x4". They were also aware that the main complaints from consumers and darkroom technicians were about exposure latitude and colour accuracy. Their development efforts and marketing materials reflected this awareness.
Most consumers had never even heard the word "resolution" until the megapixel wars. Ektar, the only Kodak colour negative film specifically marketed as being especially fine-grained and high-resolution, failed commercially and was discontinued after five years.
This is not the only argument that color films were developed in a way that excluded people of color[1]. I'm not an expert.
You're correct that the film ecosystem of photography was not developed with resolution in mind. I was thinking of the way in which digital sensors capture the most bit depth in the upper tones [2]. This would likely not have been pursued if test targets had subjects with darker skin colors than their surroundings.
Fascinating. For about a decade my father owned a one-hour photo lab back in the eighties. I worked for him.
When he purchased the equipment for his lab, he didn’t have a ton of money so the equipment was only semi automated. There were two machines, one to develop the color negative film and another to make prints.
I operated the printing machine. This involved sitting at a console and feeding each strip of negatives (typically 36 frames) one image at a time over a lamp which lit the negative from below. I then punched some buttons to make each print. I was assisted by a computer. But the computer was dumb. It looked at the negative and tried to average out the exposure so that the resulting print on average would be grey. 18% grey to be exact.
So my job was to look at this inverse negative that was 1” x 1.5”, figure out what the scene actually was, then override the computer’s exposure and color balance to get a correct print. 12 minutes later when the print had developed I’d know if I guessed correctly.
On a good day, I’d print about 100 rolls of film, 3600 prints, and maybe I’d have to redo about 36 of those.
Now this was in Miami and so I was printing picture of every subject and skin type under the sun.
And the funny thing is, I don’t recall ever noticing, particularly, that there were any issues with prints of persons of color.
What I do remember distinctly is that my dad and I never agreed on blue/yellow color balance. I thought he made everyone look like Smurfs. I pleaded with him to look at the whites of people’s eyes. Never won that battle. It was his shop.
Also, beach scenes. Snow scenes. Cameras and film often (counterintuitively) underexposed those.
But this was long ago and maybe I didn’t notice or don’t remember.
> Also, beach scenes. Snow scenes. Cameras and film often (counterintuitively) underexposed those.
The snow being under-exposed is the camera is trying to balance out the image to 18% grey [usually]. This means in a scene with mainly bright white snow it will expose to make the scene grey, unless you override it. The camera can't [with current technology as far as I know] know that you are shooting snow. If you're shooting in snow then you probably want to adjust exposure by +2 as a starting point.
Photographers used to carry around a physical grey card and would use that to take a light meter reading from. That would give the base standard exposure for the ambient light so the scene should be correctly exposed. [ it's more complex due to dynamic range etc but that covers the basics ]
All: this post was flagged. But it is obviously intellectually interesting and therefore on topic here, regardless of one's views on race and other issues. It's full of fascinating detail, and even has a technical aspect for the less-curious among us who only find technical things interesting. I've turned off the flags and rolled back the timestamp on the post.
We can debait the title a bit, but a substantive article with a slightly baity title is not one that should be flagged. If the title is your issue, you're welcome to let us know at hn@ycombinator.com, or suggest a better title in the thread, which we'll happily use if we see the suggestion.
One last thing: if you're going to comment on a thread like this, can you please top up on the site guidelines first? Especially this one: Comments should get more thoughtful and substantive, not less, as a topic gets more divisive. Remember that we're going for good conversation—respectful and two-sided—rather than ideological or political battle.
Thanks for resurrecting this post! I also appreciate the clear guidance that "a substantive article with a slightly baity title is not one that should be flagged". Maybe this could be added to the https://news.ycombinator.com/newsguidelines.html?
I'm not sure, though, that it makes sense to focus on the idea that it was a "mistake" to flag this article. I think flagging is currently one of the weakest areas of the HN moderation system, in that it allows a minority to suppress discussion of otherwise good articles that they disagree with. Clearer guidelines can help, but I think you should also think hard about how the system can be changed to be more resilient to "mistakes".
Maybe flagging of articles with some minimum number of upvotes should act only as a "request for moderator review" rather than having any immediate visible effect?
Maybe there could be a way (either mandatory or optional) to explain why an article was flagged? I presume it would be easier to review a flag that says "Bad title, should be X" than just an opaque and universal flag.
Maybe there could be some better feedback for users who flag articles, to tell them if they are flagging articles "correctly"? As it is, I presume a lot of people persist in doing what they think is best, even if you think they are using the system incorrectly.
It's great that thanks to good moderation HN is still going strong after all these years. Obviously you are doing many things right! But I strongly feel that current flagging system needs to be rethought for this success to continue. Explanatory comments like the one you made here are helpful, but I think some deeper changes will also be necessary.
Ok I took out "That was a mistake." Fair point. Also, I looked at the history of the users who flagged it, and none of them had a history of flagging obviously good articles.
I don't know if the flagging system needs rethinking though. The way it works has been stable for years. I appreciate that such claims are a bit irritating, because they're based on a view of the site that not everybody gets to see. But the [flagged] marker is publicly visible, and if you or anyone sees an article in [flagged] state that ought not to be flagged—say, one whose signal/noise ratio is as high as the OP or better—then please tell us (hn@ycombinator.com is best, because that way we're sure to see it). You especially, because we know how much and how sincerely and for how long you have cared about open discussion on this site.
Posts can be intellectually stimulating yet not belong on this site. What's important about a submission to users is a lot more nuanced, and I absolutely see why it was flagged. A slightly baity title on the wrong topic generally causes a derailed comment section since the title primes how you interpret the article(and many uses don't read the article).
For example, consider a post about the political analysis of the changing voting habits of Congress. The post is great, it is filled with technical details and an explanation on the application of game theory on voting habits in Congress, yet it is titled "Racism in the American Political System along Party Lines." The title enough is alone to disqualify such a post here IMO as it would in itself cause a dumpster-fire of a comment section, or at least have a much higher likelihood of it.
Fair points as well. And sure, the title is by far the strongest influence on a new thread. But there are other ways, as I mentioned, for users to request a title change, and if they're flagging an article based on title alone without looking at the text, that's not good.
I appreciate how the moderation call could seem non-obvious in this case, even though it was obvious to me. Here's the most important thing: people shouldn't just flag every divisive article. It depends on the quality of the specific article. We don't want HN to be taken over by political battle. But we also don't want to sanitize it of social issues, because it would be less intellectually interesting that way, and that is the entire point of the site. Here's a good question to ask when deciding whether to flag: what is the information-to-indignation ratio of the article? If it's high, don't flag it. I wrote about this the other day, for anyone who wants more: https://news.ycombinator.com/item?id=19720659.
What prompted me to comment publicly in the current case is that the article was high in information and low in indignation—surprisingly low, considering where the author is coming from. Which makes the article even more interesting, btw—any time an article defies expectation for its genre, that's good for HN.
Thanks for this comment. I tend to flag "every divisive article" myself, and maybe I'll resist that a bit more now.
Essentially what I use HN for is the comment discussion. So for me, the main thing to look at is whether an article will generate a good discussion. Saying "I don't actually care if the article is good" is not exactly true, but it's also not that far off: a well-written article that generates a bad discussion is still a bad thing IMO.
Whereas someone who uses HN as just a news aggregator may have the opposite thought -- the comments don't matter, all that matters is the article quality.
I agree—but your comment omits something important, which is that article quality also conditions comment quality. It's a reciprocal loop. To get the most out of it means watching for quality articles that otherwise might fall through the cracks, and it also means coaxing threads to rise to a higher quality level instead of collapsing into default sludge.
This submission is a good example to make the point; it sits just on the interesting edge. That's why I've spent so much time commenting in this thread: not because I know anything about the social implications of photography tech, but because it reveals something about how HN can get better.
It is unclear to me why there is "inherited bias" in film/photography technology. Inherited bias implies that the biased film practices of the past have spilled over into the present, causing further racial bias in photography.
While it's true that photography in the past was racially biased, it is unclear why modern photography is still racially biased. It may just be genuinely harder to photograph dark skin, after all, the article even says that attempts have been made to fix racial bias in photography.
All of this being said, even if photography is not racially biased and the issue is more technological rather than social, the problem should still be fixed.
I could also be wrong about all of this, and in reality dark skin is just as easy to photograph as light skin, but the article did not give a modern example of the racially biased "Shirley card"--except for facial recognition technology, but I didn't really consider that photography in the traditional sense.
Edit:
While modern photography may not be racially biased, Kwindla does make an excellent point that machine learning is replicating the past mistakes of photography.
> While it's true that photography in the past was racially biased, it is unclear why modern photography is still racially biased. It may just be genuinely harder to photograph dark skin
Perhaps if those with dark skin were a majority, far more R&D spending would go into dealing with dark skin tones than say, better zoom lenses or image processing speed. As a practical example, I suspect HDR would have arrived much faster than it did.
Camera companies have been competing on ISO/gain performance and dynamic range for as long as they've existed. It has little or nothing to do with skin color. There are plenty of motivations for delivering high ISO and good dynamic range beyond capturing skin colors.
The idea that it was furniture companies and chocolate vendors that drove this development is absurd, that is probably just Kodak sales pitches backfiring on them 50 years later. Anyone that wanted to take a photo of a white person, let alone any subject, in a slightly darker environment wanted the same thing.
Or for an even better way to look at it: if you can photograph fast moving subjects twice as well, you can also photograph subjects twice as dark. Again, this is something everyone that uses a camera wants, independent of race.
As for HDR: it is only relatively recently that digital sensors have started outperforming film's dynamic range, which was pretty good for a very long time. Digital camera vendors absolutely had massive financial incentive to fix this deficit. When they started to even come close, that's when the market exploded for movie production, and probably a lot of other commercial applications as well.
The issue is probably the dynamic range of the sensor or film. Our eyes are incredibly sensitive to brightness and not very sensitive to color, so as a result, we are kinda good at seeing things with different illumination very well and the brain corrects in the color if it's missing it (in fact, most of your vision is black-and-white being colored in by your brain).
Cameras on the other hand have one sensor that is only color and not black and white. And with low brightness, noise becomes a huge problem with the electronics. The modern smartphone sensors is incredibly small and even moderate lighting conditions already require some post processing or special capturing methods like HDR to cook up an acceptable image.
The issue is most likely fixable by making HDR a standard feature on all cameras and extending the dynamic ranges even further so that darker skin tones are appropriately captured in all lighting conditions.
There's no "merely" about it. ML algorithms are decided on historical data, and it's not surprising that historical data has historical bias.
There's little "amusing" about a piece of software that re-enforces prejudices, only now in a way that is supported by a technology culture which prides itself on blind faith in machines. For victims of those prejudices, it presents a real and terrifying future.
The 1995 Multicultural Shirley Card image in the article is beyond my skill, but I was a little confused by it, in a way that might be relevant to the article.
I'd expect it to let you calibrate for the black woman's hair detail without blowing out the white dress, but this image on my screen (print? scan?) doesn't seem to have done that.
It looks like the black woman's hair is almost lost in this image, yet it looks like the photographer put a hair light on the blonde woman, and not on the black woman.
(Lighting: In the reflections in their eyes, it looks like two studio lights. A narrow hair light for the blonde, from camera left, might be a third. Also looks like a backdrop light behind, which might also be doing that small hint of rear lighting on camera right of the black woman's hair, which emphasizes that the rest of the hair isn't lit as well as it could be.)
(Outfit exposure: The gray outfit might be Gray Card gray, and you can also compare it to the white outfit. The black dress looks hopeless for exposure, at least in this print/scan.)
(Focus: The blonde head looks in a bit softer focus, which might be accidental or glamorous.)
Thank you for sharing - not being familiar with calibration cards, I only paid attention to the features of faces (which seem well-rendered on the print), though the white dresses seemed to unnaturally lack shadow. I did not notice how the black hair texture on the women to the left and right was completely lost until you pointed it out.
With studio lighting (often powerful strobes, though this time it might just be continuous hot lights), it's pretty easy for something white to get blown out completely. If you also have something dark in the scene, you might need more light on that, at the same time that you're trying not to get too much light off the white. One tool in the studio is to use multiple light sources and modifiers. I don't know why that didn't seem to happen for the black woman's hair in this particular print. (The exposure of the white dress might be exactly as they intended, however -- almost an extreme, but the folds still detectable.)
You can do an image search on your favorite search engine and look for someone like Donna Brazile, who has been in politics for a long time and understands appearing on TV.
Then you can look at the clothes she is wearing and realize that in every photo where she is wearing a lighter dress or jacket, the automatic exposure settings of the camera adjusted for the average lighting of the scene and underexposed her face.
That is why, because Brazile has done countless TV appearances for decades ... in most of the photos you can see that she wears darker jackets or clothing.
You can barely make out Michael Jordan's facial features. Michael Jordan, in a team photograph of the 1990/91 Chicago Bulls.
The lesson here is to design technology so it works for everyone, not just people like you. Sometimes it's a matter of literal skin color, sometimes it's a question of physical size and shape or abilities.
That's a bad example, half the pic is white (white guys + white shirts), the other half is black (black guys + black suits and dark background). Even with today's tech it's easy to mess a picture like that. It also looks like it was badly scanned / digitalised., cameras in the 90s were way more capable than that.
There is no magic, photography is all about light, darker skin colors reflect less light, thus you need to ramp up the exposure or the sensitivity of the film, which destroys details/highlights. The only racial related thing here would be the photographer chosing to correctly expose one skin color over another.
Idk what type of film was used for that particular pic but some films are made to better render white people's skin tones (kodak portra for example), while other are made for the asian markets, some are better for snowy/sandy landscapes, other for blueish/greenish scenes, &c.
I did specifically go looking for a more difficult photo; adjusting exposure can help to capture a single skin tone, so a diverse group of skin tones is going to be harder. MJ’s skin tone in particular is pretty dark, though I remember it looking a shade or two lighter at times; I don’t know if that’s because the photographers and TV cameraman adjusted exposure so we could more clearly see him or just normal suntans (off-seasons back home in NC in the summer vs. winter in Chicago). Scottie Pippen, for example, looks relatively fine in the photo.
Although it’s a harder photograph to get right in terms of technical difficulty, it’s also more than a fair chance to the photographer since everyone is more or less sat down in a controlled environment. Maybe the Bulls didn’t have the best equipment or photographers in the world, but it’s not a total amateur operation, either.
This is interesting, I've always thought that makers of image capture equipment were striving to create technology to capture the world, as it is, without any bias towards any particular subject.
Maybe, in a way they were, and they simply went for the lowest hanging fruit first? or maybe they simply put most effort where they believed that there would be most gain for least effort.
The reason why they started at one point is not really important, the important thing is that they continue to strive to improve. Not because some group of people are excluded, but because, from a technical standpoint, they're not done before they can capture and display an accurate, neutral, non-biased, factual representation of reality. Same goes for audio, I'm sure getting voice right had priority over music, and that some styles of music had priority over others, but, we're not done before all SOUND, not just music, can be captured and represented accurately.
> I've always thought that makers of image capture equipment were striving to create technology to capture the world, as it is, without any bias towards any particular subject.
You'd be surprised, especially with pre-digital photo. Each film emulsion has its own characteristics. I won't go to much into details because it's explained all over the internet already, but here is a quick comparison of a few emulsions:
As fun of darker skin adult models, I always found it annoying how most adult videos mishandle dark skin. I always thought there must be something inherent in the technology that makes it not optimized for darker skins, especially in the hands of less skilled cameramen.
I hope no one will be offended by my comment, I mean it factually.
The people who invented the tech (US/Europe/Japan) optimised it for consumers around them.
If darker skinned people had invented it, or had been a richer consumer group things would have been different - to think otherwise you'd have to think greedy capitalists would give up piles of cash to be racist.
Why hate on inventors who create something cool just because it doesn't quite work as well for all groups of people?
Surely this also left a gap in the market - someone could have optimized film for darker skin tones and made a lot of money?
The article covers all of this closely, and I see nothing in it that can fairly be described as "hate on inventors". Would you mind reviewing the site guidelines at https://news.ycombinator.com/newsguidelines.html? I think you'll notice that they require HN comments to be more substantive than this one here.
For one, there's this: "Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith." The author of an article is someone.
Then there's this: "Eschew flamebait." Taking an HN thread further into flamewar, which is the direction your comment points in and alas even moves a little into, definitely breaks that guideline. Keep in mind that once one person goes there, a lot of others are going to go there—for and against, bashing each other along the way—so the biggest responsibility is not to be first to go there. If we all avoid that, no flamewar.
There's also: "Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something." It seems to me that the dismissals here were indeed shallow, reminiscent of instant objections that pop into one's mind when encountering what, for whatever reason, we dislike. That mechanism is hardly unique to you—it is active in 100% of us. But the HN guidelines have been carefully written to ask all of us to slow down and inhibit that mechanism—to be more reflective and less reflexive—because this is the only way to get good conversation on the internet.
There are two other guidelines that the comment breaks, but I'll leave them as an exercise for the curious.
The article does not cover at all why black people didn't invent their own film technology or were a sufficiently organised or attractive consumer group to spur someone else to (it was chocolate and furniture makers)
"Hating" is perhaps too strong a word, but the author makes multiple accusations of racism against companies, technicians etc - from my reading thinking things are far too intentional and getting angry/offended - where in reality it's a mix of money and physics - not racism against a people.
"Hating" is much too strong a word to describe the article—so much so that invoking it in a thread like this is a big upping of the flamewar ante. There's no need to do that, but it's hard to resist, when a topic stirs up powerful emotions. This creates a need for relief, and venting that energy in the form of extreme words is one way to get relief. Unfortunately, it doesn't relieve anything at a community level. It just tosses the hot potato around in a way that only makes the potato hotter and more painful to the one who catches it next. What provides relief at a community level is when people find ability in themselves to acknowledge truth in what the other is saying.
I would say there is more love in the article—consider the passages about the author's grandfather, whose humiliation she in a way dedicated her career to repairing—and later about her father. Note how she includes a moving (to me at least) moment of reconciliation at the end ("Her eyes were glassy as she said goodbye. Mine were, too, grateful for her vulnerability."). How easy it would have been to shame the woman who made the faux pas instead. Such moments of acknowledgment are hard to come by, and are worth emulating. This is not someone who's just out to hate.
> thinking things are far too intentional and getting angry/offended
A more charitable interpretation of the article is that a series of omissions can compound into a bias, even without deliberate attempts to exclude. That's interesting, and I'm a bit puzzled by the aversion in some commenters to look at it. Yes, angry accusations have been made and still get made, but that leads us to hear them also when they're not really there. We need the ability to notice when they're not really there, so as to respond in kind. That would be a de-escalating movement.
> the author makes multiple accusations of racism against companies, technicians etc
I'd urge you to read the article again and see whether those are really there, or if you haven't somehow filled those details in, perhaps because it felt that way while reading it. I just reread the whole thing myself, to see whether I had missed some accusation of racism against a company or a technician. I didn't find any. In fact, the second reading convinced me that the author must have taken great pains to restrain herself from doing that—since nothing would be easier for someone in her position to do.
The article does contain a lot of pain—the pain of being unseen, excluded. And she does do something difficult for the reader: she creates tension by never expressing the pain directly. It's there implicitly, which heightens the effect. That's a pretty effective device for making a point, and I wonder if that's really what people are reacting to: the discomfort we feel when something intense is present but not expressed. But there's also generosity in this. If someone holds back from expressing as much as they could, even when feelings are intense, it creates space for others to do the same. Those opportunities are worth noticing and acting on, because otherwise we all just repeat the cycle.
> The people who invented the tech (US/Europe/Japan) optimised it for consumers around them.
I must confess that it grinds my gears too when people rail against makers for being 'selfish' or discriminatory when the things they make end up best suited for people like them.
You could even see such a accusations as implicitly insinuating that the non-catered-to groups are somehow less capable of innovating their own more suitable alternatives.
The dominant culture will get blamed for either not doing enough or for doing too much, so you have to do what you can, and what you feel is right, accept useful feedback, and realize you can't please all the people all the time.
This seems to have touched a nerve, which is not unexpected.
Examples:
- colonial powers are criticized both for exporting agricultural technologies westernizing other countries and for not exporting enough
- Hollywood is criticized both for cultural imperialism and for not including enough diversity as well as for cultural appropriation
- foreign aid is insufficient/creates dependency/empowers corrupt governments/is colonialism/...
- tourism is exploitation but no tourism is isolationist, crippling to the economies, ...
etc.
Any action (including inaction) is likely to be criticized, so you've just got to do your best, and try to improve as you learn.
The Kipling reference shows that the issues with technology raised in the thread are merely new aspects of an old dilemma.
What is my responsibility to share something I value?
That could be culture, technology, religion, defense, ways of raising children, marriage age, purifying water, sanitation, etc.
Where does "sharing" and "helping" become "imperialism" and "patronizing"?
Where does not sharing become negligence?
What level of customization for varying physical/cultural needs/preferences is required?
We make our best choices but may get criticized by others who choose differently, and will likely be judged harshly by future commentators, just like Kipling is now.
Take a raking kind of light, and shine it on a beige to brown colored wall, so that the light will be stronger on one side and gradually fade in intensity (that is, less reflected light) across the field of view.
Now place a different shade of brown upholstered chair, in front of this wall.
Take a photo using the Hasselblad X1D, currently considered to have the most accurate color of mainstream cameras.
Now try to print it accurately, even with the best quality paper and a high end inkjet printer.
Having read the article (and being just enough of an amateur photographer to know how hard it is to take good pictures, and how much harder it was when working in film instead of digital) I don't actually think the headline is over-stating the case at all.
Photography is not just a system of calibrating light, but a technology of subjective decisions. Light skin became the chemical baseline for film technology, fulfilling the needs of its target dominant market. For example, developing color-film technology initially required what was called a Shirley card. When you sent off your film to get developed, lab technicians would use the image of a white woman with brown hair named Shirley as the measuring stick against which they calibrated the colors. Quality control meant ensuring that Shirley’s face looked good. It has translated into the color-balancing of digital technology. In the mid-1990s, Kodak created a multiracial Shirley Card with three women, one black, one white, and one Asian, and later included a Latina model, in an attempt intended to help camera operators calibrate skin tones. These were not adopted by everyone since they coincided with the rise of digital photography. The result was film emulsion technology that still carried over the social bias of earlier photographic conventions.
I can see the case she’s making for film photography, but her remarks on digital photography came off as having a bit of a gripe against physics itself. Any darker and lower-contrast object is going to be more difficult to photograph in detail than a lighter higher-contrast object. To imply that the reason technology hasn’t sufficiently (in her opinion) overcome this challenge is due to racial bias is quite a leap. Especially since low light and low contrast photography has such a broad application outside of photographing people, and is met with all the same challenges.
No, not any more than eyes are. At least, with a neutral projection surface.
But once you add technology for recording photographs on film, or creating and displaying digital images, there's potential for racism. I know next to nothing about photography. But it seems pretty clear that building stuff that works well for everyone is a hard problem.
I take issue with this article simply because photography isn't a technology at all. It is an art that uses technology. There are millions of pictures of people of all races that look perfectly fine. I take issue with stirring up people for no reason.
If a maker of paints in the 1800s owned slaves does that mean that painting (then, now, in the future) is racist? How ridiculous can we get?
Sure it's ridiculous. But also it's not what the author says. Therefore your comment breaks this site guideline: "Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith."
Would you mind reviewing https://news.ycombinator.com/newsguidelines.html and doing a better job of posting substantively here? You've posted one-liner flamebait twice already, which is not cool, especially on a divisive topic.
If "millions of pictures of people of all races ... look perfectly fine", that just means that those photographers had requisite skills. And cared enough to get it right.
But the issue isn't so much about those photographers. It's about the defaults designed into film, cameras, photosensors, and so on. As TFA discusses. So random users -- and even many technicians -- don't get good results when they rely on the defaults. As in "your jacket is lighter than your face". And that's even an issue for many "white" people, which is why TV makeup is such a thing.
And about your paint maker. If they didn't care very much about painting dark-skinned people, maybe they wouldn't put much effort into researching and developing paints that worked well for that. It's exactly the same issue as with Kodak in TFA.
So is that racism? One could argue that it's just not caring very much about some race(s) of people. But one could also argue that anything that's race-based is racism.
I posted this, and definitely wasn't thinking of it as clickbait, or even as controversial. As Dan says, I think this history is both intellectually interesting and relevant to my work as an engineer, today.
Over time, as the technology of photography was invented and iteratively improved, film formulations came to work better for taking photographs of light-skinned people than of dark-skinned people. Inventors and engineers didn't set out to disadvantage dark-skinned people specifically. But it happened anyway.
In the relatively short history of commercial "AI", we've already seen examples of machine learning systems that followed a similar development path. Non-representative data sets[0], replicating existing bias[1], over-fitting data sets to the point where the outcomes violate non-discrimination laws[2], etc.
One way to help avoid making these mistakes is to think about how they were made in the past, in other engineering contexts.
[0]-https://www.theatlantic.com/technology/archive/2016/04/the-u... [1]-https://www.theverge.com/2018/10/10/17958784/ai-recruiting-t... [2]-https://www.vox.com/2019/3/21/18275746/facebook-settles-ad-d...
In the days of film photography this was an important issue - negative/slide film chemistry had a huge impact on the output of your work. I used to shoot a lot of Fuji Velvia due to its excellent contrast and reasonable resolving power. It was terrible for portraits, though, of any race - it picked up too much red in both light and dark skin tones, and the result rarely looked good. However I definitely think that it was true back then that more portrait orientated film chemistry was poorly optimised for those of African descent or with similar skin tones.
I'm less certain it's such an issue today. Some cameras have face recognition built in as a focus/af assist, and I suspect that these are better with some faces than others (most consumer cameras are developed in asia so I'd except them to do best with asian faces, but as I don't use this sort of feature I don't know). Other than that, I'm struggling to come up with a way that bias - in the way I'd define it - could creep in. In-body JPEG colour profile perhaps, but most serious photographers don't depend on that, and it can be rectified in post anyway.
On the other hand, there's the unfortunate technical problem that dynamic range is limited, and having a subject with very large contrast between skin and clothing brightness is technically challenging to shoot, especially in a situation like the one illustrated where the subject can move around so you have limited fine grained control over lighting.
I'm totally on board with presenting the former film chemistry issue as bias - but that's pretty much just historical at this point. The number of people shooting events on slide film is pretty close to zero these days. There's also very little if any investment into film chemistry as a consequence of this.
The second issue irks me a bit. If you have very dark skin and wear very light clothing, or very light skin and wear very dark clothing then it's going to be more of a challenge to capture you effectively. That's physics not racism, and conflating the two issues feels a bit disingenuous to me.
It is not just photography, even in real life under low light conditions it can be hard to read the facial expressions of dark skinned people.
True. So maybe skin darkening would be more practical than CV Dazzle, because it wouldn't be so unusual. Horrible legacy and PR, though.
Anyway, I suspect that was an evolutionary advantage. For hunting in forest and savannah, to hide in shadows. The human rootstock was black, I gather. And lighter skin evolved where soft UV was limiting for vitamin D synthesis.
But that's solvable with the right technology. Maybe something like getting overall color value, and then tweaking color contrast. If we can do it in astronomy, we can do it with people.
I would guess that the evolutionary advantage of black skin was not primarily camouflage. Tool-using humans of any colour can do that with mud and grass.
My bet is on UV tolerance. Black people have lower rates of skin cancer in the US compared to whites. https://www.jaad.org/article/S0190-9622(05)02730-1/fulltext (Mortality is higher in blacks, but that's because it's diagnosed at later more severe stages.) The effect would have been far bigger when foraging and hunting all day every day under the equatorial sun.
Or perhaps light skin was an evolutionary development as humans moved to higher latitudes in order to obtain high enough levels of vitamin D. See [1]: The evolutionary significance of vitamin D, skin pigment, and ultraviolet light.
1. https://www.ncbi.nlm.nih.gov/m/pubmed/1211435/
Good points.
And maybe there's linked stuff that got selected for.
The genetic history of human skin colour is really amazing. All our ancestor had fair skin because of body hair (Chimps have fairish skin under all their body hair), we then became dark skinned as we lost most of our body hair, and then those of us who had ancestors that lived far from the equator became fair again.
As someone who has the ancestry of people who lived far from the equator and now lives in a country that is much closer to the equator I wish we hadn't lost our body hair.
Huh. I guess that I've never seen a shaved chimp.
But now that I think of it, hairless mice have light-colored skin. Some hairless dogs too.
Yes if you are covered with hair you not only don’t need dark skin for UV protection, but it will likely interfere with your ability to synthesise vitamin D. Also why waste energy synthesising something you don’t need (melanin).
Dark-skinned faces are inherently more difficult to photograph. Dark skin is in the lowest range of values in most images, but the specular highlights of that skin can reach the highest range of values. Underexpose and you lose all shading; overexpose and you lose the geometric definition given by specular highlights.
Pale skin can be made easier to photograph in challenging lighting conditions by using a matte powder, which diffuses specular highlights and reduces the value range. This doesn't work well on dark skin, because subtle shadows are less visible and you lose the impression of shape; you need the specular highlights, but they greatly increase the value range.
The problem with early colour films wasn't really hue (all human skin falls into a remarkably narrow range of hues), but a lack of sensitivity and dynamic range compared to monochrome films. They coped with over-exposure reasonably well, but even slight under-exposure would obliterate the detail on dark skin. This limitation of chemistry was a constant problem in all forms of colour photography, but the social burden fell disproportionately on dark-skinned portraiture subjects. Later emulsions had far better rendition of lowlights and smooth compression of highlights, which vastly improved their performance with dark skin.
I just want to note that the technical causes you list here are, arguably, totally beside the point.
All technical systems have strengths and weaknesses. When systems are conceived and developed their goals are set through a culturally-created understanding of what the system "should" do. The lack of attention to darker faces (from chemical process selection to selection of people depicted in kodak's system calibration cards) explains how we got this particular set of technical limitations instead of one that, say, tended to over-expose easily.
> Dark-skinned faces are inherently more difficult to photograph
I would say that dark-skinned faces are inherently poorly captured by the photographic technology we have developed, which has focused on detail rather than dynamic range.
>selection of people depicted in kodak's system calibration cards
The choice of the model does reflect some amount of bias, but it doesn't cause bias in the printing process.
Precise calibration was done with a test chart and a densitometer. Shirley was just a quick reference used throughout the day, to check at a glance whether prints were unnaturally blue or purple or washed out due to changes in the developer chemistry. Fuji usually supplied test films and prints with Asian models, even in European markets; the motion picture industry originally used porcelain dolls. A multi-ethnic test image wouldn't have provided better accuracy.
>I would say that dark-skinned faces are inherently poorly captured by the photographic technology we have developed, which has focused on detail rather than dynamic range.
Except that the opposite was true for the entire duration of the film era. The photographic film industry was acutely aware of the fact that most consumer photographs were taken on compact cameras with mediocre lenses and usually printed at 6"x4". They were also aware that the main complaints from consumers and darkroom technicians were about exposure latitude and colour accuracy. Their development efforts and marketing materials reflected this awareness.
Most consumers had never even heard the word "resolution" until the megapixel wars. Ektar, the only Kodak colour negative film specifically marketed as being especially fine-grained and high-resolution, failed commercially and was discontinued after five years.
This is not the only argument that color films were developed in a way that excluded people of color[1]. I'm not an expert.
You're correct that the film ecosystem of photography was not developed with resolution in mind. I was thinking of the way in which digital sensors capture the most bit depth in the upper tones [2]. This would likely not have been pursued if test targets had subjects with darker skin colors than their surroundings.
[1] https://www.vox.com/2015/9/18/9348821/photography-race-bias [2] https://digital-photography-school.com/exposing-to-the-right...
Fascinating. For about a decade my father owned a one-hour photo lab back in the eighties. I worked for him.
When he purchased the equipment for his lab, he didn’t have a ton of money so the equipment was only semi automated. There were two machines, one to develop the color negative film and another to make prints.
I operated the printing machine. This involved sitting at a console and feeding each strip of negatives (typically 36 frames) one image at a time over a lamp which lit the negative from below. I then punched some buttons to make each print. I was assisted by a computer. But the computer was dumb. It looked at the negative and tried to average out the exposure so that the resulting print on average would be grey. 18% grey to be exact.
So my job was to look at this inverse negative that was 1” x 1.5”, figure out what the scene actually was, then override the computer’s exposure and color balance to get a correct print. 12 minutes later when the print had developed I’d know if I guessed correctly.
On a good day, I’d print about 100 rolls of film, 3600 prints, and maybe I’d have to redo about 36 of those.
Now this was in Miami and so I was printing picture of every subject and skin type under the sun.
And the funny thing is, I don’t recall ever noticing, particularly, that there were any issues with prints of persons of color.
What I do remember distinctly is that my dad and I never agreed on blue/yellow color balance. I thought he made everyone look like Smurfs. I pleaded with him to look at the whites of people’s eyes. Never won that battle. It was his shop.
Also, beach scenes. Snow scenes. Cameras and film often (counterintuitively) underexposed those.
But this was long ago and maybe I didn’t notice or don’t remember.
> Also, beach scenes. Snow scenes. Cameras and film often (counterintuitively) underexposed those.
The snow being under-exposed is the camera is trying to balance out the image to 18% grey [usually]. This means in a scene with mainly bright white snow it will expose to make the scene grey, unless you override it. The camera can't [with current technology as far as I know] know that you are shooting snow. If you're shooting in snow then you probably want to adjust exposure by +2 as a starting point.
Photographers used to carry around a physical grey card and would use that to take a light meter reading from. That would give the base standard exposure for the ambient light so the scene should be correctly exposed. [ it's more complex due to dynamic range etc but that covers the basics ]
All: this post was flagged. But it is obviously intellectually interesting and therefore on topic here, regardless of one's views on race and other issues. It's full of fascinating detail, and even has a technical aspect for the less-curious among us who only find technical things interesting. I've turned off the flags and rolled back the timestamp on the post.
We can debait the title a bit, but a substantive article with a slightly baity title is not one that should be flagged. If the title is your issue, you're welcome to let us know at hn@ycombinator.com, or suggest a better title in the thread, which we'll happily use if we see the suggestion.
One last thing: if you're going to comment on a thread like this, can you please top up on the site guidelines first? Especially this one: Comments should get more thoughtful and substantive, not less, as a topic gets more divisive. Remember that we're going for good conversation—respectful and two-sided—rather than ideological or political battle.
https://news.ycombinator.com/newsguidelines.html
Thanks for resurrecting this post! I also appreciate the clear guidance that "a substantive article with a slightly baity title is not one that should be flagged". Maybe this could be added to the https://news.ycombinator.com/newsguidelines.html?
I'm not sure, though, that it makes sense to focus on the idea that it was a "mistake" to flag this article. I think flagging is currently one of the weakest areas of the HN moderation system, in that it allows a minority to suppress discussion of otherwise good articles that they disagree with. Clearer guidelines can help, but I think you should also think hard about how the system can be changed to be more resilient to "mistakes".
Maybe flagging of articles with some minimum number of upvotes should act only as a "request for moderator review" rather than having any immediate visible effect?
Maybe there could be a way (either mandatory or optional) to explain why an article was flagged? I presume it would be easier to review a flag that says "Bad title, should be X" than just an opaque and universal flag.
Maybe there could be some better feedback for users who flag articles, to tell them if they are flagging articles "correctly"? As it is, I presume a lot of people persist in doing what they think is best, even if you think they are using the system incorrectly.
It's great that thanks to good moderation HN is still going strong after all these years. Obviously you are doing many things right! But I strongly feel that current flagging system needs to be rethought for this success to continue. Explanatory comments like the one you made here are helpful, but I think some deeper changes will also be necessary.
Ok I took out "That was a mistake." Fair point. Also, I looked at the history of the users who flagged it, and none of them had a history of flagging obviously good articles.
I don't know if the flagging system needs rethinking though. The way it works has been stable for years. I appreciate that such claims are a bit irritating, because they're based on a view of the site that not everybody gets to see. But the [flagged] marker is publicly visible, and if you or anyone sees an article in [flagged] state that ought not to be flagged—say, one whose signal/noise ratio is as high as the OP or better—then please tell us (hn@ycombinator.com is best, because that way we're sure to see it). You especially, because we know how much and how sincerely and for how long you have cared about open discussion on this site.
Posts can be intellectually stimulating yet not belong on this site. What's important about a submission to users is a lot more nuanced, and I absolutely see why it was flagged. A slightly baity title on the wrong topic generally causes a derailed comment section since the title primes how you interpret the article(and many uses don't read the article).
For example, consider a post about the political analysis of the changing voting habits of Congress. The post is great, it is filled with technical details and an explanation on the application of game theory on voting habits in Congress, yet it is titled "Racism in the American Political System along Party Lines." The title enough is alone to disqualify such a post here IMO as it would in itself cause a dumpster-fire of a comment section, or at least have a much higher likelihood of it.
Fair points as well. And sure, the title is by far the strongest influence on a new thread. But there are other ways, as I mentioned, for users to request a title change, and if they're flagging an article based on title alone without looking at the text, that's not good.
I appreciate how the moderation call could seem non-obvious in this case, even though it was obvious to me. Here's the most important thing: people shouldn't just flag every divisive article. It depends on the quality of the specific article. We don't want HN to be taken over by political battle. But we also don't want to sanitize it of social issues, because it would be less intellectually interesting that way, and that is the entire point of the site. Here's a good question to ask when deciding whether to flag: what is the information-to-indignation ratio of the article? If it's high, don't flag it. I wrote about this the other day, for anyone who wants more: https://news.ycombinator.com/item?id=19720659.
What prompted me to comment publicly in the current case is that the article was high in information and low in indignation—surprisingly low, considering where the author is coming from. Which makes the article even more interesting, btw—any time an article defies expectation for its genre, that's good for HN.
Thanks for this comment. I tend to flag "every divisive article" myself, and maybe I'll resist that a bit more now.
Essentially what I use HN for is the comment discussion. So for me, the main thing to look at is whether an article will generate a good discussion. Saying "I don't actually care if the article is good" is not exactly true, but it's also not that far off: a well-written article that generates a bad discussion is still a bad thing IMO.
Whereas someone who uses HN as just a news aggregator may have the opposite thought -- the comments don't matter, all that matters is the article quality.
I agree—but your comment omits something important, which is that article quality also conditions comment quality. It's a reciprocal loop. To get the most out of it means watching for quality articles that otherwise might fall through the cracks, and it also means coaxing threads to rise to a higher quality level instead of collapsing into default sludge.
This submission is a good example to make the point; it sits just on the interesting edge. That's why I've spent so much time commenting in this thread: not because I know anything about the social implications of photography tech, but because it reveals something about how HN can get better.
It is unclear to me why there is "inherited bias" in film/photography technology. Inherited bias implies that the biased film practices of the past have spilled over into the present, causing further racial bias in photography.
While it's true that photography in the past was racially biased, it is unclear why modern photography is still racially biased. It may just be genuinely harder to photograph dark skin, after all, the article even says that attempts have been made to fix racial bias in photography.
All of this being said, even if photography is not racially biased and the issue is more technological rather than social, the problem should still be fixed.
I could also be wrong about all of this, and in reality dark skin is just as easy to photograph as light skin, but the article did not give a modern example of the racially biased "Shirley card"--except for facial recognition technology, but I didn't really consider that photography in the traditional sense.
Edit:
While modern photography may not be racially biased, Kwindla does make an excellent point that machine learning is replicating the past mistakes of photography.
> While it's true that photography in the past was racially biased, it is unclear why modern photography is still racially biased. It may just be genuinely harder to photograph dark skin
Perhaps if those with dark skin were a majority, far more R&D spending would go into dealing with dark skin tones than say, better zoom lenses or image processing speed. As a practical example, I suspect HDR would have arrived much faster than it did.
Camera companies have been competing on ISO/gain performance and dynamic range for as long as they've existed. It has little or nothing to do with skin color. There are plenty of motivations for delivering high ISO and good dynamic range beyond capturing skin colors.
The idea that it was furniture companies and chocolate vendors that drove this development is absurd, that is probably just Kodak sales pitches backfiring on them 50 years later. Anyone that wanted to take a photo of a white person, let alone any subject, in a slightly darker environment wanted the same thing.
Or for an even better way to look at it: if you can photograph fast moving subjects twice as well, you can also photograph subjects twice as dark. Again, this is something everyone that uses a camera wants, independent of race.
As for HDR: it is only relatively recently that digital sensors have started outperforming film's dynamic range, which was pretty good for a very long time. Digital camera vendors absolutely had massive financial incentive to fix this deficit. When they started to even come close, that's when the market exploded for movie production, and probably a lot of other commercial applications as well.
The issue is probably the dynamic range of the sensor or film. Our eyes are incredibly sensitive to brightness and not very sensitive to color, so as a result, we are kinda good at seeing things with different illumination very well and the brain corrects in the color if it's missing it (in fact, most of your vision is black-and-white being colored in by your brain).
Cameras on the other hand have one sensor that is only color and not black and white. And with low brightness, noise becomes a huge problem with the electronics. The modern smartphone sensors is incredibly small and even moderate lighting conditions already require some post processing or special capturing methods like HDR to cook up an acceptable image.
The issue is most likely fixable by making HDR a standard feature on all cameras and extending the dynamic ranges even further so that darker skin tones are appropriately captured in all lighting conditions.
Machine learning merely learns patterns that are fed into it. I find it rather amusing that people do not like the result.
There's no "merely" about it. ML algorithms are decided on historical data, and it's not surprising that historical data has historical bias.
There's little "amusing" about a piece of software that re-enforces prejudices, only now in a way that is supported by a technology culture which prides itself on blind faith in machines. For victims of those prejudices, it presents a real and terrifying future.
The 1995 Multicultural Shirley Card image in the article is beyond my skill, but I was a little confused by it, in a way that might be relevant to the article.
I'd expect it to let you calibrate for the black woman's hair detail without blowing out the white dress, but this image on my screen (print? scan?) doesn't seem to have done that.
It looks like the black woman's hair is almost lost in this image, yet it looks like the photographer put a hair light on the blonde woman, and not on the black woman.
(Lighting: In the reflections in their eyes, it looks like two studio lights. A narrow hair light for the blonde, from camera left, might be a third. Also looks like a backdrop light behind, which might also be doing that small hint of rear lighting on camera right of the black woman's hair, which emphasizes that the rest of the hair isn't lit as well as it could be.)
(Outfit exposure: The gray outfit might be Gray Card gray, and you can also compare it to the white outfit. The black dress looks hopeless for exposure, at least in this print/scan.)
(Focus: The blonde head looks in a bit softer focus, which might be accidental or glamorous.)
Thank you for sharing - not being familiar with calibration cards, I only paid attention to the features of faces (which seem well-rendered on the print), though the white dresses seemed to unnaturally lack shadow. I did not notice how the black hair texture on the women to the left and right was completely lost until you pointed it out.
With studio lighting (often powerful strobes, though this time it might just be continuous hot lights), it's pretty easy for something white to get blown out completely. If you also have something dark in the scene, you might need more light on that, at the same time that you're trying not to get too much light off the white. One tool in the studio is to use multiple light sources and modifiers. I don't know why that didn't seem to happen for the black woman's hair in this particular print. (The exposure of the white dress might be exactly as they intended, however -- almost an extreme, but the folds still detectable.)
Can someone explain the following,
The author says: instead of seeking a solution, the technician had decided that my body was somehow unsuitable for the stage"
But the technician said: "We have a problem. Your jacket is lighter than your face, That’s going to be a problem for lighting"
In what sense did the technician decide the author should be removed from the stage, and seek no other solution?
You can do an image search on your favorite search engine and look for someone like Donna Brazile, who has been in politics for a long time and understands appearing on TV.
Then you can look at the clothes she is wearing and realize that in every photo where she is wearing a lighter dress or jacket, the automatic exposure settings of the camera adjusted for the average lighting of the scene and underexposed her face.
That is why, because Brazile has done countless TV appearances for decades ... in most of the photos you can see that she wears darker jackets or clothing.
For an illustration of how bad photography used to be, consider this photograph of the 1990-1991 Chicago Bulls: https://www.theguardian.com/sport/2017/apr/20/craig-hodges-m...
You can barely make out Michael Jordan's facial features. Michael Jordan, in a team photograph of the 1990/91 Chicago Bulls.
The lesson here is to design technology so it works for everyone, not just people like you. Sometimes it's a matter of literal skin color, sometimes it's a question of physical size and shape or abilities.
That's a bad example, half the pic is white (white guys + white shirts), the other half is black (black guys + black suits and dark background). Even with today's tech it's easy to mess a picture like that. It also looks like it was badly scanned / digitalised., cameras in the 90s were way more capable than that.
There is no magic, photography is all about light, darker skin colors reflect less light, thus you need to ramp up the exposure or the sensitivity of the film, which destroys details/highlights. The only racial related thing here would be the photographer chosing to correctly expose one skin color over another.
That one is similar to your pic but even with the bad resolution you can tell it's much better : https://usercontent1.hubstatic.com/13875852_f520.jpg
Idk what type of film was used for that particular pic but some films are made to better render white people's skin tones (kodak portra for example), while other are made for the asian markets, some are better for snowy/sandy landscapes, other for blueish/greenish scenes, &c.
On the other hand, black and white film works very well on black people: https://i2-prod.mirror.co.uk/incoming/article8111113.ece/ALT...
I did specifically go looking for a more difficult photo; adjusting exposure can help to capture a single skin tone, so a diverse group of skin tones is going to be harder. MJ’s skin tone in particular is pretty dark, though I remember it looking a shade or two lighter at times; I don’t know if that’s because the photographers and TV cameraman adjusted exposure so we could more clearly see him or just normal suntans (off-seasons back home in NC in the summer vs. winter in Chicago). Scottie Pippen, for example, looks relatively fine in the photo.
Although it’s a harder photograph to get right in terms of technical difficulty, it’s also more than a fair chance to the photographer since everyone is more or less sat down in a controlled environment. Maybe the Bulls didn’t have the best equipment or photographers in the world, but it’s not a total amateur operation, either.
This is interesting, I've always thought that makers of image capture equipment were striving to create technology to capture the world, as it is, without any bias towards any particular subject. Maybe, in a way they were, and they simply went for the lowest hanging fruit first? or maybe they simply put most effort where they believed that there would be most gain for least effort. The reason why they started at one point is not really important, the important thing is that they continue to strive to improve. Not because some group of people are excluded, but because, from a technical standpoint, they're not done before they can capture and display an accurate, neutral, non-biased, factual representation of reality. Same goes for audio, I'm sure getting voice right had priority over music, and that some styles of music had priority over others, but, we're not done before all SOUND, not just music, can be captured and represented accurately.
> I've always thought that makers of image capture equipment were striving to create technology to capture the world, as it is, without any bias towards any particular subject.
You'd be surprised, especially with pre-digital photo. Each film emulsion has its own characteristics. I won't go to much into details because it's explained all over the internet already, but here is a quick comparison of a few emulsions:
https://www.thephoblographer.com/2017/11/27/quick-comparison...
As fun of darker skin adult models, I always found it annoying how most adult videos mishandle dark skin. I always thought there must be something inherent in the technology that makes it not optimized for darker skins, especially in the hands of less skilled cameramen.
I hope no one will be offended by my comment, I mean it factually.
Is there any way to read that not being logged in?
Switching to another browser might help; it's what I do when I've reached my free article limits.
The people who invented the tech (US/Europe/Japan) optimised it for consumers around them.
If darker skinned people had invented it, or had been a richer consumer group things would have been different - to think otherwise you'd have to think greedy capitalists would give up piles of cash to be racist.
Why hate on inventors who create something cool just because it doesn't quite work as well for all groups of people?
Surely this also left a gap in the market - someone could have optimized film for darker skin tones and made a lot of money?
The article covers all of this closely, and I see nothing in it that can fairly be described as "hate on inventors". Would you mind reviewing the site guidelines at https://news.ycombinator.com/newsguidelines.html? I think you'll notice that they require HN comments to be more substantive than this one here.
For one, there's this: "Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith." The author of an article is someone.
Then there's this: "Eschew flamebait." Taking an HN thread further into flamewar, which is the direction your comment points in and alas even moves a little into, definitely breaks that guideline. Keep in mind that once one person goes there, a lot of others are going to go there—for and against, bashing each other along the way—so the biggest responsibility is not to be first to go there. If we all avoid that, no flamewar.
There's also: "Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something." It seems to me that the dismissals here were indeed shallow, reminiscent of instant objections that pop into one's mind when encountering what, for whatever reason, we dislike. That mechanism is hardly unique to you—it is active in 100% of us. But the HN guidelines have been carefully written to ask all of us to slow down and inhibit that mechanism—to be more reflective and less reflexive—because this is the only way to get good conversation on the internet.
There are two other guidelines that the comment breaks, but I'll leave them as an exercise for the curious.
The article does not cover at all why black people didn't invent their own film technology or were a sufficiently organised or attractive consumer group to spur someone else to (it was chocolate and furniture makers)
"Hating" is perhaps too strong a word, but the author makes multiple accusations of racism against companies, technicians etc - from my reading thinking things are far too intentional and getting angry/offended - where in reality it's a mix of money and physics - not racism against a people.
"Hating" is much too strong a word to describe the article—so much so that invoking it in a thread like this is a big upping of the flamewar ante. There's no need to do that, but it's hard to resist, when a topic stirs up powerful emotions. This creates a need for relief, and venting that energy in the form of extreme words is one way to get relief. Unfortunately, it doesn't relieve anything at a community level. It just tosses the hot potato around in a way that only makes the potato hotter and more painful to the one who catches it next. What provides relief at a community level is when people find ability in themselves to acknowledge truth in what the other is saying.
I would say there is more love in the article—consider the passages about the author's grandfather, whose humiliation she in a way dedicated her career to repairing—and later about her father. Note how she includes a moving (to me at least) moment of reconciliation at the end ("Her eyes were glassy as she said goodbye. Mine were, too, grateful for her vulnerability."). How easy it would have been to shame the woman who made the faux pas instead. Such moments of acknowledgment are hard to come by, and are worth emulating. This is not someone who's just out to hate.
> thinking things are far too intentional and getting angry/offended
A more charitable interpretation of the article is that a series of omissions can compound into a bias, even without deliberate attempts to exclude. That's interesting, and I'm a bit puzzled by the aversion in some commenters to look at it. Yes, angry accusations have been made and still get made, but that leads us to hear them also when they're not really there. We need the ability to notice when they're not really there, so as to respond in kind. That would be a de-escalating movement.
> the author makes multiple accusations of racism against companies, technicians etc
I'd urge you to read the article again and see whether those are really there, or if you haven't somehow filled those details in, perhaps because it felt that way while reading it. I just reread the whole thing myself, to see whether I had missed some accusation of racism against a company or a technician. I didn't find any. In fact, the second reading convinced me that the author must have taken great pains to restrain herself from doing that—since nothing would be easier for someone in her position to do.
The article does contain a lot of pain—the pain of being unseen, excluded. And she does do something difficult for the reader: she creates tension by never expressing the pain directly. It's there implicitly, which heightens the effect. That's a pretty effective device for making a point, and I wonder if that's really what people are reacting to: the discomfort we feel when something intense is present but not expressed. But there's also generosity in this. If someone holds back from expressing as much as they could, even when feelings are intense, it creates space for others to do the same. Those opportunities are worth noticing and acting on, because otherwise we all just repeat the cycle.
> The people who invented the tech (US/Europe/Japan) optimised it for consumers around them.
I must confess that it grinds my gears too when people rail against makers for being 'selfish' or discriminatory when the things they make end up best suited for people like them.
You could even see such a accusations as implicitly insinuating that the non-catered-to groups are somehow less capable of innovating their own more suitable alternatives.
"Take up the White Man’s burden—
And reap his old reward:
The blame of those ye better
The hate of those ye guard—"
Rudyard Kipling
The dominant culture will get blamed for either not doing enough or for doing too much, so you have to do what you can, and what you feel is right, accept useful feedback, and realize you can't please all the people all the time.
This seems to have touched a nerve, which is not unexpected.
Examples:
- colonial powers are criticized both for exporting agricultural technologies westernizing other countries and for not exporting enough
- Hollywood is criticized both for cultural imperialism and for not including enough diversity as well as for cultural appropriation
- foreign aid is insufficient/creates dependency/empowers corrupt governments/is colonialism/...
- tourism is exploitation but no tourism is isolationist, crippling to the economies, ...
etc.
Any action (including inaction) is likely to be criticized, so you've just got to do your best, and try to improve as you learn.
> touched a nerve, which is not unexpected
With this you've confessed to trolling. I mean, the White Man’s Burden? Come on.
Posting a viewpoint, sincerely held, which you know will be unpopular, is not trolling. They seem to believe what they're saying.
> They seem to believe what they're saying
Most trolls do.
The Kipling reference shows that the issues with technology raised in the thread are merely new aspects of an old dilemma.
What is my responsibility to share something I value?
That could be culture, technology, religion, defense, ways of raising children, marriage age, purifying water, sanitation, etc.
Where does "sharing" and "helping" become "imperialism" and "patronizing"?
Where does not sharing become negligence?
What level of customization for varying physical/cultural needs/preferences is required?
We make our best choices but may get criticized by others who choose differently, and will likely be judged harshly by future commentators, just like Kipling is now.
don't worry, dang is not adding anything to the conversation. He just enjoys virtue signaling. Hard!
Great more outrage culture.
> Kodak Gold Max became appealing. This new film was billed as being “able to photograph the details of a dark horse in lowlight
If this is true please show me the ad? Magazines from the era are easy to get, many are online already.
1995 was when I think Kodak Gold Max came out, do you really think humans way back then wouldn't get what is being said?
It may have been used in personal correspondence, by someone, in an unknown context is the best source I can see.
There is a interesting discussion here, but this is not it.
Take a raking kind of light, and shine it on a beige to brown colored wall, so that the light will be stronger on one side and gradually fade in intensity (that is, less reflected light) across the field of view.
Now place a different shade of brown upholstered chair, in front of this wall.
Take a photo using the Hasselblad X1D, currently considered to have the most accurate color of mainstream cameras.
Now try to print it accurately, even with the best quality paper and a high end inkjet printer.
Even master printmakers will struggle with it.
Why?
> suggest a better title in the thread
Will do! How about:
> Dark Things Are Dark. Film At Eleven!
Steve Sailer suggested:
> Let's Talk About My Face
Or the outrage built into clickbait
Having read the article (and being just enough of an amateur photographer to know how hard it is to take good pictures, and how much harder it was when working in film instead of digital) I don't actually think the headline is over-stating the case at all.
Is the camera obscura racist?
From the article:
Photography is not just a system of calibrating light, but a technology of subjective decisions. Light skin became the chemical baseline for film technology, fulfilling the needs of its target dominant market. For example, developing color-film technology initially required what was called a Shirley card. When you sent off your film to get developed, lab technicians would use the image of a white woman with brown hair named Shirley as the measuring stick against which they calibrated the colors. Quality control meant ensuring that Shirley’s face looked good. It has translated into the color-balancing of digital technology. In the mid-1990s, Kodak created a multiracial Shirley Card with three women, one black, one white, and one Asian, and later included a Latina model, in an attempt intended to help camera operators calibrate skin tones. These were not adopted by everyone since they coincided with the rise of digital photography. The result was film emulsion technology that still carried over the social bias of earlier photographic conventions.
I can see the case she’s making for film photography, but her remarks on digital photography came off as having a bit of a gripe against physics itself. Any darker and lower-contrast object is going to be more difficult to photograph in detail than a lighter higher-contrast object. To imply that the reason technology hasn’t sufficiently (in her opinion) overcome this challenge is due to racial bias is quite a leap. Especially since low light and low contrast photography has such a broad application outside of photographing people, and is met with all the same challenges.
No, not any more than eyes are. At least, with a neutral projection surface.
But once you add technology for recording photographs on film, or creating and displaying digital images, there's potential for racism. I know next to nothing about photography. But it seems pretty clear that building stuff that works well for everyone is a hard problem.
I take issue with this article simply because photography isn't a technology at all. It is an art that uses technology. There are millions of pictures of people of all races that look perfectly fine. I take issue with stirring up people for no reason.
If a maker of paints in the 1800s owned slaves does that mean that painting (then, now, in the future) is racist? How ridiculous can we get?
Sure it's ridiculous. But also it's not what the author says. Therefore your comment breaks this site guideline: "Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith."
Would you mind reviewing https://news.ycombinator.com/newsguidelines.html and doing a better job of posting substantively here? You've posted one-liner flamebait twice already, which is not cool, especially on a divisive topic.
I will just avoid these articles. Sorry i hurt the outrage.
If "millions of pictures of people of all races ... look perfectly fine", that just means that those photographers had requisite skills. And cared enough to get it right.
But the issue isn't so much about those photographers. It's about the defaults designed into film, cameras, photosensors, and so on. As TFA discusses. So random users -- and even many technicians -- don't get good results when they rely on the defaults. As in "your jacket is lighter than your face". And that's even an issue for many "white" people, which is why TV makeup is such a thing.
And about your paint maker. If they didn't care very much about painting dark-skinned people, maybe they wouldn't put much effort into researching and developing paints that worked well for that. It's exactly the same issue as with Kodak in TFA.
So is that racism? One could argue that it's just not caring very much about some race(s) of people. But one could also argue that anything that's race-based is racism.