Luminous Landscape Forum
The Art of Photography => The Coffee Corner => Topic started by: HonorableSensor on April 29, 2019, 09:28:19 am
-
I hesitate to post this since I don't believe her argument is even remotely close to the truth and it may be seen as flamebait. Given her network of contacts, the author could have very easily found a color scientist to speak with, but didn't.
However, given the popularity of this article (which I have seen on various non-photography sites) it might be good to know about it, since, friends and family might bring up the claims made by the author.
https://www.nytimes.com/2019/04/25/lens/sarah-lewis-racial-bias-photography.html
-
It's an old argument. I remember it from 20 or 30 years ago. It resurfaces from time to time. Recycled work as epiphany. It would have been better if she had focused on the contributions of underappreciated black photographers.
-
It is a long article, and behind a paywall for some. The Left, most academia included, already think everything is racist. Can you summarize her argument?
-
It is a long article, and behind a paywall for some. The Left, most academia included, already think everything is racist. Can you summarize her argument?
The writer starts with an image of a Shirley card, having a white woman and the color-checker squares on it, then asks: "Can a photographic lens condition racial behavior? "
====
We have a problem. Your jacket is lighter than your face,” the technician said from the back of the one-thousand-person amphitheater-style auditorium. “That’s going to be a problem for lighting.” She was handling the video recording and lighting for the event.
It was an odd comment that reverberated through the auditorium, a statement of the obvious that sounded like an accusation of wrongdoing. Another technician standing next to me stopped adjusting my microphone and jolted in place. The phrase hung in the air, and I laughed to resolve the tension in the room then offered back just the facts: “Well, everything is lighter than my face. I’m black.”
“Touché,” said the technician organizing the event. She walked toward the lighting booth. My smile dropped upon realizing that perhaps the technician was actually serious. I assessed my clothes — a light beige jacket and black pants worn many times before in similar settings.
====
A person making a serious accusation doesn't take the time to understand how metering works...
You should be able to see the article here: http://archive.is/igUgi
-
The first anecdote is the lighting technician (they are taking video of the event) stating matter-of-factly that the jacket is lighter than the woman's face, and that is a problem. Which the author, being black, takes as an offense, instead of understanding how sensors and metering works...
Light colored clothing is a lighting problem for white people too. The eye is naturally drawn to the lightest object in the frame, so if you want someone focusing on your face (whatever color it is) and not your clothing, don't wear light clothing. That's not to say racism doesn't exist; it is just that this is not a good example of it. Of course it is up to you to decide whether this is a racist remark or a fashion tip. It tells at least as much about you as it does the person making the remark.
-
Racism is a word that covers a great deal of territory these days.
She's essentially just going over the same ground that has been covered a bunch of times about the Shirley card (a picture of a white woman) that Kodak distributed to help technicians do color corrections for Kodak films. While she makes mention of the fact that Fuji film did better with non-white people. she leaves unstated the obvious fact that it was formulated by non-white people.
People tend to assume that the world looks like themselves, and when they're devising technology (be if color film, stage lighting, or AI algorithms for detecting faces) they tend to bias the technology so that it works better for people who look like themselves. For a variety of reasons. Given the size of the groups that work on these things, we may assume that it contains a few people who outright despise everyone who looks different, and we may also assume that access to test subjects is also tilted, so there's a large envelope for containing reasons here.
Given that technology is disproportionately developed by white people, and disproportionately not developed by people of African descent, this means that a hell of a lot of technology works badly for people with very dark skin. Sometimes appallingly badly, sometimes only slightly badly, and occasionally there is no detectable bias.
Contemporary usage lumps all this under the label "racism" which you may or may not agree with, but anyways do distinguish the usage of language from the underlying phenomena being talked about.
It is all too common to assume that the word "racism" somehow implies lynching and screeching bigots, but that is simply not contemporary usage.
-
Thanks for the summary. It just confirms that most of the so-called experts are actually idiots.
-
Shirley! Forgot about that. Good times.
Article sounds silly. I shoot people of all skin colours. Fact of life where I live. Takes a bit of extra skill. Local black people can be quite light compared to the West Africans who are really very dark. We get a lot of West African models. Shoot a lot of what we call coloureds as well, mixed race people. Kind of honey coloured, easy to make look good. My sister in law is black, quite dark very attractive woman. I find white people the most difficult, especially when the client decides to shoot them with a Nigerian in a white shirt. Was close to impossible back in the days of EPR. Shirley is going to be white where most people are white and that’s how Kodak came up with it I’m sure.
Does that make me racist?
-
White wedding dress + black tux is a problem as old as the wedding photography. Films like Portra, with gentler contrast, tried to solve that. Nothing to do with race.
-
While it's very tempting to say "well, it's just human nature" or "that's how it is, wot wot?" the fact is that some things don't work as well for some people as they do for other people.
When that happens, the people for whom it doesn't work as well are perfectly reasonable to stand up and ask "hey, can we make this work better for me, and people who look like me?" and they deserve to not be dismissed, especially if the answer is obviously "yes, of course it can be."
-
A more scholarly article (from 2009):
https://www.cjc-online.ca/index.php/journal/article/view/2196/3069
-
I just finished a studio lighting class at my local community college. I had as models, black, white and mixed race individuals, and I thought my Fuji XT2 did an equally good job with all skin tones. I don't remember anyone complaining about skin tone reproduction with their Canon, Nikon and Lumix cameras. Maybe companies are doing better color science today. Nevertheless, I am glad I use an X-Rite Color Checker and not a Shirley card. No sense perpetuating a stereotype.
-
According to the article, Kodak was aware that their film did not respond appropriately to darker skinned folks, yet it was the criticism from the chocolate and furniture industries that motivated Kodak to work on its emulsions.
::)
-
We have a problem. Your jacket is lighter than your face,” the technician said from the back of the one-thousand-person amphitheater-style auditorium. “That’s going to be a problem for lighting.”
I remember photographing the wedding of two friends of mine
He was a German paleface and she a lady from Uganda.
He was in black and she in white.
Yes that was a problem outdoors, outside the church. The whole zone system.
-
Day exterior: group shot, office workers in Zaire (now, Congo) The Zairois are really dark, as mentioned before. The guys I was photographing were all wearing white shirts and it was near noon. At the equator.
I was shooting ECO 7252, a reversal Ektachrome stock in 16mm. ASA 25. Fine-grained and contrasty.
No reflector available. Worst photographic case I've yet encountered.
-
The article simply illustrates that there is hardly a thing that somebody, somewhere, cannot take and turn on its head to be something else: a skewered version of reality.
If the blessed professor knew, the problem is scientific, not racist, but as folks say, to a hammer...
I wouldn't lose any sleep about it.
Rob
-
The article simply illustrates that there is hardly a thing that somebody, somewhere, cannot take and turn on its head to be something else: a skewered version of reality.
If the blessed professor knew, the problem is scientific, not racist, but as folks say, to a hammer...
I wouldn't lose any sleep about it.
Rob
Amen, brother.
It is a “white hare in the snow, black cat in a tunnel” problem, not race. It is 18% gray-card problem, not race. It is academic idiots problem. It is a knee-jerk reaction problem that looks at at the end result and immidiatelly cries “racism,” without understanding the underlying reasons. Ultimately, it is idiots problem, even if Harvard professors. Especially Harvard professors.
-
It's all very well to wave it off as a metering problem, but to do so is to indicate that you understand neither modern metering, nor the NYT times.
Metering in today's modern cameras encompasses rather more than taking a reading off a grey card, and the article mentions more things than metering.
The point is not that metering black people is hard. The point is that metering, lighting, AI facial recognition, color science, a host of other things, and all the combinations of these things (like, for instance, metering in modern cameras) tends to work better for white people than for not-white people. This is a bummer for not white people, who would like things to be different.
Look, people on Lula are apparently allowed to bitch about how not all cameras produce DNG files, and how cameras ought to but do not have an ETTR exposure mode, and they are taken seriously. These people would like cameras to work better for them. What, exactly, is wrong then with a dark-skinned person asking that cameras work better for them?
Why do the concerns of dark-skinned people get waved off, while idiotic ideas like "ETTR exposure mode" are taken seriously?
C'mon, people. You're better than this.
-
It's all very well to wave it off as a metering problem, but to do so is to indicate that you understand neither modern metering, nor the NYT times.
Metering in today's modern cameras encompasses rather more than taking a reading off a grey card, and the article mentions more things than metering.
The point is not that metering black people is hard. The point is that metering, lighting, AI facial recognition, color science, a host of other things, and all the combinations of these things (like, for instance, metering in modern cameras) tends to work better for white people than for not-white people. This is a bummer for not white people, who would like things to be different.
Look, people on Lula are apparently allowed to bitch about how not all cameras produce DNG files, and how cameras ought to but do not have an ETTR exposure mode, and they are taken seriously. These people would like cameras to work better for them. What, exactly, is wrong then with a dark-skinned person asking that cameras work better for them?
Why do the concerns of dark-skinned people get waved off, while idiotic ideas like "ETTR exposure mode" are taken seriously?
C'mon, people. You're better than this.
You're getting off on the wrong foot. It's not about metering, but about dynamic range. Of course you can adjust your exposure on both film and digital to encompass the zone/tone important to you; just don't expect that being the righteous bro' makes a blind bit of difference to the science: the science and problem in this instance is the difference between dark skin and light clothing. One or the other is gonna come out second-best.
I stick to my opinion: hammer looking for nail. Reversed racialism, perchance?
Take it a step or two backwards into your photographic education: lost detail in highlights is usually more disturbing than lost shadow. Blown highlights grab attention and make things look terrible, even on white skin.
-
Rob, god knows I have nothing but love for you! But this hasn't got anything to do with any specific technical detail. It's about the shape of the whole creature here, which is made up of myriad individual technical details, influenced by myriad technical choices.
-
You're getting off on the wrong foot. It's not about metering, but about dynamic range.
Hear! Hear!
That's what I meant too
-
I love it when a guy who has no clue about photography, and in particular the technical side of it, pontificates about the technical side of photography and claims that the science of photography is inherently racist. Where is the proof that modern cameras do not do well with dark skins? That is, anymore than black tux/white wedding dress, for instance.
-
This is a discussion that is indeed history, and not relevant anymore.
If you look at the images put out at the moment, every type of skin and race is pictured, and well exposed.
- skins getting too red is still a problem- a technical problem ...
If you see the old target in the article - it looks awful... but that might be an analogue to digital technical problem.
-
This is a discussion that is indeed history, and not relevant anymore.
If you look at the images put out at the moment, every type of skin and race is pictured, and well exposed.
- skins getting too red is still a problem- a technical problem ...
If you see the old target in the article - it looks awful... but that might be a analogue to digital technical problem.
This is a fair point. Things have obviously comes a long ways since Shirley. Cameras, for all I know, work equally well or poorly across all skin colors for the most part. Up and coming technologies, however, don't. Google's facial recognition system famously tagged africans as gorillas not too long ago, and, frankly, it takes only a few seconds on google to come up with a bunch of software that was discovered to fail miserably on this racial group or that.
Now, I don't expect that to roll in to Canon's next generation facial-recognition focus/metering algorithms, and therefore collapse when confronted with African people. Why do I expect the best from Canon, rather than the worst? Because the media continues to cover these issues, continues to make sure that the not-African people working on the code continue to have somewhat forward in their minds "right, right, test on black people, don't forget" rather than gradually forgetting about that continent in the rush to deliver features.
Canon knows that there's likely to be a ****storm if their next generation cameras can't successfully eye-focus, or meter, when confronted with dark-skinned people. If they didn't know this, they might very well drop the ball.
-
As a humble newbie I must concede that I don't understand the issue.
Dark skin falls on zone IV, white skin on zone VI. The gray card on zone V.
Has anything changed? Is there a real problem?
And yes, of course I had to google to know who this Ms.Lewis is.
-
Yes, this specific issue is history.
But similar issues arise in other ways in tech. Facial recognition softwares have been found to not handle non-white faces very well. I listened to a podcast recently (sorry, forgotten which one) about this very problem and in at least one case was traced to the fact that the AI was "taught" faces from a skewed data base of primarily white faces. Easy enough to fix, perhaps or perhaps not, depending on how well the program was written. This points to inadequate testing but as an ex-IT guy, I can tell you that very little software gets properly tested these days, especially if it delays delivery (and payment). It's easy to write this off as an outlier problem until the wrong person is stopped at a border with modern-day reduced due process. Aside from the obvious problem of an innocent person being victimized by bad AI design, it also means the AI may be missing the real bad guys. As we rely more and more on this stuff, these issues will become more and more important.
The point I am trying to make is that it's better not to dismiss a topic too quickly until you know all the ramifications. The specific problem of photo exposure can and has been dealt with, but the larger problem will turn up in other areas. The article that was originally referred to may have done a bad job of describing this, but that doesn't mean it isn't real.
-
As a humble newbie I must concede that I don't understand the issue.
Dark skin falls on zone IV, white skin on zone VI. The gray card on zone V.
Has anything changed? Is there a real problem?
Color rendition has been a problem in the past with film. I don''t know if it is now a problem with digital sensors.
-
As a humble newbie I must concede that I don't understand the issue.
Dark skin falls on zone IV, white skin on zone VI. The gray card on zone V.
Has anything changed? Is there a real problem?
And yes, of course I had to google to know who this Ms.Lewis is.
Skin falls where you want it to fall, eh? Just for fun I pulled out my copy of Ansel Adams "The Negative" and my copy of Walker Evans "American Photographs" and started comparing the little zone system chart in the former with skin tones in the latter, and, well, the answers were all over the place. Globally, skin tones had by various people probably cover 4-5 zones all by themselves, and when you chuck in lighting you pretty much run a gamut from Zones III through IX without looking unnatural or "improperly exposed."
But, Lewis's issue isn't really with this or that technicality.
-
My feeling is that the issue is mostly being confused. You all know how to meter. How skin tones look becomes an aesthetic decision. I don't see any issue there.
I had in mind as an example Zanele Muholi (https://www.thisiscolossal.com/2019/04/monograph-zanele-muholi/)
On the other hand, out in the real world, most technology (cameras, face recognition etc) is used on automatic. People press a button. That's it.
That's where some questions need to be raised about how well the tech is working.
-
Skin falls where you want it to fall, eh? Just for fun I pulled out my copy of Ansel Adams "The Negative" and my copy of Walker Evans "American Photographs" and started comparing the little zone system chart in the former with skin tones in the latter, and, well, the answers were all over the place. Globally, skin tones had by various people probably cover 4-5 zones all by themselves, and when you chuck in lighting you pretty much run a gamut from Zones III through IX without looking unnatural or "improperly exposed."
But, Lewis's issue isn't really with this or that technicality.
You compared tones from books? And on top of that from two different books?
Well, I wouldn't do that. Just MHO
-
I could probably expand on this in endless boring detail, but perhaps it will suffice to say I did not select the books at random, and I that have reason to believe that the brief investigation I conducted provides useful information.
-
... Google's facial recognition system famously tagged africans as gorillas not too long ago...
Don't blame the technology, geeks, or AI, but mother nature.
-
I could probably expand on this in endless boring detail, but perhaps it will suffice to say I did not select the books at random, and I that have reason to believe that the brief investigation I conducted provides useful information.
Whatever.
You selected Walker Evans' book and one of the editions of The Negative (both editions I have, 1948 and 1981, are more or less poorly printed - but do their job of explaining the ZS if you WORK with it) and I can assure you from my humble seat that comparing the tones of the printed images with Walker Evans' whatever edition it is you have, is not the right thing to do.
That said, I propose to leave it at that. ;)
-
Relevant technical details for my comparison: roughly similar print capability and rendering intent.
Check.
Check.
-
I think I have seen skin tones in Zone IX. Avedon and Gibson come to mind. It is a particular aesthetic decision. What that has to do with photography being racist is anyone's guess.
-
Yes, this specific issue is history.
But similar issues arise in other ways in tech. Facial recognition softwares have been found to not handle non-white faces very well. I listened to a podcast recently (sorry, forgotten which one) about this very problem and in at least one case was traced to the fact that the AI was "taught" faces from a skewed data base of primarily white faces. Easy enough to fix, perhaps or perhaps not, depending on how well the program was written. This points to inadequate testing but as an ex-IT guy, I can tell you that very little software gets properly tested these days, especially if it delays delivery (and payment). It's easy to write this off as an outlier problem until the wrong person is stopped at a border with modern-day reduced due process. Aside from the obvious problem of an innocent person being victimized by bad AI design, it also means the AI may be missing the real bad guys. As we rely more and more on this stuff, these issues will become more and more important.
The point I am trying to make is that it's better not to dismiss a topic too quickly until you know all the ramifications. The specific problem of photo exposure can and has been dealt with, but the larger problem will turn up in other areas. The article that was originally referred to may have done a bad job of describing this, but that doesn't mean it isn't real.
I understand what you say, the original problem was when there were no coloured models; Since then things have changed in a multi-cultural way in that field but not (yet) in others...
There are enough battles for equal rights to be fought. Men are still paid more than women etc etc..
-
NYT has zero credibility with me. I won't waste my time.
Kent in SD
-
NYT has zero credibility with me. I won't waste my time.
Kent in SD
So what news paper do you 'trust' / find credible
-
Don't blame the technology, geeks, or AI, but mother nature.
Since you've not added an emoticon, I have to assume you are being serious. Are you?
Cheers,
Bart
-
Since you've not added an emoticon, I have to assume you are being serious. Are you?
Yes.
-
Yes.
So you equate africans to gorillas?
-
So you equate africans to gorillas?
Perhaps he means that it's natural for people of one ethnicity to kind of forget about the existence, or importance, of other ethnic groups.
-
So you equate africans to gorillas?
You are smarter than that, creating a strawman.
-
Perhaps he means that it's natural for people of one ethnicity to kind of forget about the existence, or importance, of other ethnic groups.
And that was not even a strawman, just plain stupid.
-
Honestly, Bart, I would not assume that Slobodan has a point here. He often does not, he's just being a vaguely anti-PC contrarian. That's his schtick.
-
I am surprised that this thread added so many posts so quickly!
One thing occurred to me, which is, that the digital sensor, and the film, both respond in ways that are very very different from the way our eyes actually process a scene in front of us.
Given this, it seems even more silly for the author of the original piece to complain about the reaction of chemicals or silicon in comparison to the human eye...
-
You are smarter than that, creating a strawman.
Then care to explain what you did mean when you responded:
Don't blame the technology, geeks, or AI, but mother nature.
-
1. So you equate africans to gorillas?... Bart
2. Don't blame the technology, geeks, or AI, but mother nature... Slobodan
Bart, that's silly. The point being made is one of tonality. Even I could understand that, me wot ain't even no professor!
I'm a skinny, damaged old white guy; I'd rather fight an African gentleman than a gorilla any day, even though given the chance, I'd fight nobody or anything. Even old age is beating the hell out of me.
;-)
-
While it is tempting to say that google's AI was simply foxed by the skin color, that explanation is incorrect.
-
HonorableSensor had a point in his opening post, for even here, honorable members are also incapable - at times - of accepting that sometimes a zebra is just a zebra and neither an electronic camera aid nor a horse in drag.
I thought gorillas were mostly rolls of gaffer tape? That said, I do remember - just - that during my teens we would occasionally refer to ciggies as gorillas too, but I have no idea of the provenance of that one; I just joined the mob, which I was later to learn was the worst path to follow even if the most travelled. That was perhaps the second most significant lesson of my life.
Rob
-
We are talking at cross purposes, Rob.
I, at any rate, accept that a zebra is a zebra. The point is, with this many zebras around, every one of them an honest zebra, it becomes reasonable to suspect that we might be in Africa.
-
..He often does not, he's just being a vaguely anti-PC contrarian...
Vaguely? Vaguely!? That’s an insult, my friend*
*Vaguely
-
We are talking at cross purposes, Rob.
I, at any rate, accept that a zebra is a zebra. The point is, with this many zebras around, every one of them an honest zebra, it becomes reasonable to suspect that we might be in Africa.
Just as well we are not referring to professors, then, but honest, bona fide animals with either black and white stripes or, as some cynics would have it, white and black stripes. With all that confusion in nature already, the uncertainties, you'd think folks would have some sympathy with the poor old camera manufacturers and cut them some slack here and there.
Regarding the point made elsewhere about wedding pictures and grooms (nothing to do with horse or zebra) in black, and brides in white - yet another gesture devoid of current-day rationality but full of wishful thinking and post-modern hopes - we can only safely assume that those who still continue to clamour for same are of split personality bordering on the rabid: don't they grasp that all those pictures that will cost them so much and that they will never look at again other than as evidence one way or the other, won't even be properly exposed to make the cake look right? As bad, that the files will have to be tampered with just to accommodate the silly clothes, will possibly lead to their dismissal in court.
Thinking of the cake reminds me that it's the real victim of the event, about to be ravished, torn asunder and tossed to the masses out there. I was about to mention the horseshoe, but realised it would just bring us full circle to the bloody zebras once again.
-
For anyone interested, removing bias in AI is a growing field of study. If you google the term, you'll get tons of review and more detailed articles. Just one example https://www.technologyreview.com/s/612876/this-is-how-ai-bias-really-happensand-why-its-so-hard-to-fix/ (https://www.technologyreview.com/s/612876/this-is-how-ai-bias-really-happensand-why-its-so-hard-to-fix/). There may even be TED talks on the subject. I tried to locate the podcast that I mentioned earlier on this thread but I failed to find it, sorry.
-
... I was about to mention the horseshoe, but realised it would just bring us full circle to the bloody zebras once again.
Speaking of horses, some white faces resemble horses (e.g. French actor Fernandel), some pigs (my high school psychology teacher).
-
Since this is a photographic forum, I'll contribute a self-portrait.
So much humanity in that what-are-you-looking-at stare:
(https://live.staticflickr.com/4054/4639524173_d350932e84_z.jpg) (https://flic.kr/p/84YLU6)
Ape (https://flic.kr/p/84YLU6) by Slobodan Blagojevic (https://www.flickr.com/photos/slobodan_blagojevic/), on Flickr
-
So what news paper do you 'trust' / find credible
Pretty much none of them. WSJ seems most fact based but isn't a general interest publication.
Kent in SD
-
We take the saturday WSJ, which is in general excellent. Avoid the op-ed pages, which can approach comical, and are only occasionally sensible.
It comes monthly with a surprisingly good arts/fashion/culture magazine which keeps me peripherally apprised both of what the major labels are doing photographically in their advertising (they *all* take ads, which means that there's almost always 1 or 2 things that are truly wonderful), as well as lightly acquainted with the world of high end art and design.
The Christian Science Monitor was once very well respected, but I have not looked in on them in a while.
The NYT is a demented joke at this point, but pieces of the sort cited stand apart in the NYT or any medium. They are statements of opinion, and of personal experience. There is no "journalistic" aspect to be judged here. Just as the WSJ opinion pieces stand apart from their paper, so does this one. In general the NYT's photography coverage seems to exist in its own little world. Not a world I always love, but anyways separate.