I mean, you can read the linked-to article that explains their position pretty clearly. They basically say that they believe that it separates one from the original experience.

I have no idea if they are really prominent (Art History is not my field), but I will say, I am sure if you looked you'd find historians who also thought this sort of thing was fine. Historians are of many opinions. It makes for a better article to focus on those who say, "this popular thing is actually bad," but that doesn't make it at all representative. They are entitled to their views; so are the people who like these things.

I think of it sort of like textual primary sources. When I am assigning something to students, I prefer to give them a scan of an original source, rather than a re-type version of it. Struggling through the ugliness of the original source, or its awkward typography (including such oddities as the long S), gives one a much better feeling of communing with the past in all its messiness, as opposed to a retyped version typeset in clean, crisp word processing software. I feel that it does engage one's "historical instincts" a bit better — you have to struggle a bit more, and that struggling is part of the work.

But I'm also aware that this is somewhat bullshit. It's essentially Barthe's "reality effect" applied to textual analysis — as if it makes it "more historical" to read it in handwriting rather than Times New Roman. It doesn't, really. Struggling to read a text can give one the experience of a historian but it does not actually bring one closer to historical truth or even historical empathy. That is a separate process. The only advantage to the struggling is that it makes you very cognizant of your struggling to grasp the past.

That is basically what these art historians are saying, as I read it. Seeing an upscaled, colorized version of the past is seeing a constructed, mythical, modern, "easy to process" version that keeps you from having to do the hard work of really extending your empathy backwards in time. OK, I guess, maybe. And they are unhappy that some people don't realize they're modified. One says students are submitting colorized pictures in essays without realizing they are colorized. I mean, that's annoying, sure, but there's a lot of fake stuff out there, and the only answer is going to be to teach students to be more critical of these things. And what a wonderful "teachable moment" it is to point this out, eh? And let's be honest — if "students can get things wrong" is a criteria for stopping something, shit, we're gonna be stopping a lot of things. :-)

Ultimately the audience for these creations isn't other historians. It's not people who are actually making that hard struggle. It's the everyday people who find the past hard to relate to, and are agog when they can suddenly relate to it. Personally, I think that's a great impulse to encourage, even if it's artificial. Who knows where it may lead? I can't see the harm of it, to be honest. It seems like a very abstract objection.

I loved They May Not Grow Old. I have studied World War I, I have read historical works on it, I have struggled to empathize with the past, etc. And yet there is something transformative and magical about Jackson's work, something that extends it beyond reality and makes it into something new. It's not a perfect reproduction of the past. It's an artificial past, one partially created by the present. But not wholly created by the present: when done well, there is still a wonderful core of historical reality in there, and the increase in historical empathy can be palpable. You stare into those boys' faces and you think, "shit, those are just kids... kids at war. What the hell?" And yeah, you could shame people into not feeling that instinctively, or not seeing the same thing you see when you look at an image (after how many years of post-graduate education?), but really, what is the point? One of the historians in the article did not like the film and thought it was gimmicky — that's a perfectly valid opinion! But to jump from "I found it gimmicky" to "it is actually causing people to misunderstand World War I" strikes me as quite a leap.

Again, they are 100% entitled to their opinions and to tell the world what they think, and Wired is 100% allowed to report on them. What would academia be without cranky academics? Not academia. But I also think we should recognize that alleged "harm" here feels pretty low.

I could go on a rant about the hyper-criticism that is common in history and the humanities (and its ultimately uselessness — if you only tear down, you never build anything up), but that seems tangential to the above, though it comes to mind. I will just end with the note that many historians (and other humanists) are prone to lecture people on how other (non-historians) will understand something, and why it will lead them astray, and so on. But in my experience none of them have ever actually done any empirical research to find out if that is true. I have taken to pointing out, whenever I can, "that is an empirical question" when such things come up — do people understand it in mode X, rather than mode Y? We could actually find out with some good social science research. We maybe should if we believe it is important. But this is a mode of thinking very alien to historians; I am only infected with it because I work with an excellent social scientist on how people understand various forms of risk communication (relating to nuclear matters) and have to come to appreciate that the way academics perceive things can be dramatically different from how everyone else perceives things ("I am not the target audience" is how I like to put this).

While I cannot provide a well researched historical answer, I have expertise in computer vision and image processing for humanities applications and can explain the argument presented in this article.

McKernan is making an argument about the historical photograph as a text. In much the same way that we don’t “correct” old newspaper articles with more modern language, we shouldn’t “correct” old photographs by inferring color and resolution information. Instead, we should try our best to interpret them as-is, experience them in the context of period technology and approach their content critically: every photograph was taking by someone for a reason and it is ahistorical to ignore that.

Image upscaling and colorization can either be done by hand or automatically using an interpolation strategy. If done by hand, the editor decides what pixel values to use to fill in the gaps. Automatic interpolation strategies now usually make use of machine learning and infer the pixel values based on training data: existing color images and their downsampled/grayscale counterparts. Regardless of which method is used, colorization and upscaling are fundamentally ill-posed problems, meaning they have multiple solutions which are equally correct, from a mathematical perspective. Deciding which color image to use injects new information and biases which weren’t in the original.

On the other hand, I would argue that that isn’t such a bad thing. Colorized images are much like translations of archaic texts. They’re not suitable for scholarly purposes, but they allow us to relate to them in new ways and they’re entertaining. As long as we treat these images as interpretive illustrations and not any kind of neutral historical fact, there’s really not that much harm done.

Example reference on modern colorization techniques: https://arxiv.org/pdf/1603.08511.pdf

Example reference for modern image upscaling: https://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Schulter_Fast_and_Accurate_2015_CVPR_paper.pdf

I have lots more where that came from if anyone is interested, sorry they’re not really accessible reading. I would appreciate if anyone knows more accessible references.

I am not a historian, so this comment may be removed, but I am a Certified Professional Photographer (CPP) and Photographic Craftsman (Cr.Photog.) and I will do my best to thoroughly back up my answer with examples.

This is a complex issue masquerading as a very simple one. On the face of it, it would seem obvious that colorizing images would help modern viewers connect with and understand the subject matter better. (The underlying assumption there is essentially that the black and white presentation creates a distraction and/or a sense of distance. I would question whether that assumption is even true, but that’s a conversation for another day.)

The reality, however, is much more nuanced. Firstly, we have to remember that while the photographers who made the images in question worked with black and white film, they lived in a color world. In other words, a photographer shooting in black and white has to constantly navigate that difference between the living scene before her eyes and the way that scene was translating on her film, without the visual assistance provided by today’s digital preview screens. She had to be intimately acquainted with all the varied ways in which the scene would read differently in black and white, in order to effectively capture the image. She was constantly making active choices and decisions in her framing, composition, and exposure with the shift to black and white in mind. Colorizing the image disregards all of those decisions, thereby potentially altering the entire mood, message, or even meaning of the image in a way the photographer did not intend.

I realize this all sounds a bit vague, so let me use the thumbnail image from your linked article as an example. The image in question is Migrant Mother, photographed by Dorothea Lange. The original image is incredibly powerful, and a lot of that power comes from the impact of how forcefully the viewer’s attention is drawn to the mother’s face. The children’s faces are turned away, perhaps to protect privacy, but also making the mother’s face the only fully visible one in the image. Evolutionarily, our eyes naturally go toward a face in an image, but in this case that effect is strengthened by the fact that everyone’s clothing and the surrounding environment is dark enough to make her face the lightest overall area of the image, which also draws the eye. The lack of color also allows her wrinkles to become less prominent. As a result, her eyes and eyebrows become the most richly detailed part of her face, and the central aspect of the image. The only thing that eventually pulls our eye away from her face is the baby’s face in the bottom right, another relatively bright spot in the image that stands out and forces us to engage with such a young innocent face almost forgotten in the midst of the scene.

In the colorized version, all of that changes. Her winkles are more prominent, drawing the eye much more. The different color of her shirt compared to the childrens‘ clothing, and the fact that her shirt matches her eyes, creates an aesthetically pleasing look but also distracts us in that all-important first moment, and thereby weakens the scene’s immediate impact. The baby’s face, because the skin tone matches closely with the color of the clothing, is far less noticeable, while the contrast between the older kids’ hair and the skin on their necks makes those spots draw the eye much more than they do in black and white.

The original image centers powerfully on the mother’s eyes, the expression and intention in her gaze, and the contrast between her heavily-weighted form and the light, clean, innocent one in her arms. The older children serve almost as symbols of the idea of children, and the weight of responsibility on the mother’s shoulders, rather than focusing on them as individual people with their own stories.

The colorized image centers more on the mother’s age and situation (wrinkles and dirt) than on her expression/intention. The older children draw the eye a lot more, making one wonder more about them as individuals. The infant’s skin tone blends in a lot more with the color of the clothing, making it much less of a focal point and more a part of the background.

I am not intending here to pass a judgement on the story that either version of the image tells, but rather to illustrate how different those two stories and messages are. Even if colorization does make it easier for modern viewers to connect with these important historical images, can we afford to ignore the many ways in which the process inevitably alters the photographer’s intent? If these images are so important for us to engage with, should we not be more concerned with maintaining the integrity of the message/meaning/etc.?

I don’t have an answer to the debate itself, unfortunately. But it’s an important discussion to have, and I’m very glad to see you raise it here in this forum (even if this comment of mine doesn’t make the cut).

Honestly, I could go either way on this sort of thing, but I tend to fall on the side of thinking it's mostly a good thing.

One of the problems with history, especially the popular understanding of history, is the persistent tendency to view historical people as utterly different from contemporary people, often in crucial ways. And while there are differences, many of the perceived differences are due to filtering through the dirty lens of historical preservation. People of the past were no less sophisticated, diverse, or interesting than people of today, but a lot of that gets missed. This often extends to things like silliness and the ordinary petty details of daily life. Much of that is missing (and sometimes intentionally elided) from historical context, and the result is that it becomes very easy to view people of the past as less human and less relatable than they actually were. A perfect example of how using technology to "touch up" (including colorization) old video is this footage of a snowball fight in Lyon, France from 1896. Personally, I think it's hard to argue that particular work doesn't enable a greater sense of immediacy, a closer connection, and a humanization of the subjects of the video to modern eyes.

On the other hand, this type of work isn't without risks and downsides. The use of machine learning for upscaling and colorization (used on footage of Japan from the 1910s, used as an example in the article) is a perfect illustration of some of those problems. Those algorithms rely on learning from training datasets, and they will adopt biases and limitations based on those training sets. There are countless examples of how bias in machine learning works so I won't get too deep into it. However, in this case the major problem is that when you train an algorithm to upscale, sharpen, colorize, etc. video footage using modern examples the result is that you will bias historical footage to look more modern as a result. To use a simplistic analogy, if you were to start off with a textual description of a historical person that said simply "he wore a shirt and pants" you could erroneously fill in the details there with a modern context and create an image of that person wearing jeans and a graphic or logo t-shirt, which might not be accurate at all. In the case of upscaling the faces of people in 100+ year old footage the errors are smaller but no less real. The people of Japan of the 1910s had different levels of nutrition, experienced different diseases through childhood, had slightly different genetic makeup, had different standards of beauty, etc, etc, etc. all of which has an impact on the way their faces looked, and which would create subtle (but noticeable to the human eye) differences compared to modern Japanese faces. When you upscale footage like this you insert details that didn't exist, and that can create a false impression.

That said, I think there is some degree of fetishization of the "rough hewn past" that is at play here. Something akin to the desire to see Roman statues as pristine monochrome hunks of stone instead of the reality of the gaudy colors they were originally. Along with a desire to force everyone who appreciates history to have to go through the long and arduous path of suffering and hard work that "real" historians have to in order to properly "commune" with the past instead of just casually grabbing shiny baubles out of the past like a kid reaching into a candy bin. But honestly, I don't think there's a huge problem with people just consuming bits and pieces of the past that they find entertaining. I think there's a bit of a false dichotomy here, as the alternative to people consuming possibly slightly inaccurate views of the past isn't the entirety of humanity trudging up the steps to acquire a PhD in history, that's never going to happen. The alternative is as we have now, people filling in too much of history with eye-rollingly-bad garbage from pop-culture entertainment (television and movies especially). And sure, most of them will "know" it's fictionalized, but even then a lot of the fictionalized details still get stuck in people's heads through sheer repetition and osmosis. The average person's view of history is maybe 80% mythological, I don't think it hurts to expose people to more of the real stuff even if it ends up being a little bit inaccurate now and again.

Because they're often done badly or with very minimal research or historical knowledge and it's not always clear that they have been recoloured.

I work frequently with WWII images and it requires accurate colour interpretation of the images, there aren't a vast amount of images in existence but there are some from all fronts which due to age and techniques will display colours that are actually quite different to those used but knowing this allows interpretation with a good degree of accuracy.

Increasingly images are coming up of say a Spitfire in SEAC markings that may or may not have been coloured but without further information you cannot say if the camouflage was dark earth or a dark grey, both being used in the theatre and the A & B schemes being mirrors and inversions of each other you may have to end up guessing which is not exactly professional. Likewise was a Ju-88A4 pictured in North Africa a faded RLM70 or had it been repainted in RLM79 by the time of the photo? Sometimes there will be a photograph that is period and in colour which is genuine and provides further knowledge of what a unit was using (Luftwaffe is a minefield when it comes to markings anyway) but can you even rely on this if you know that it could have been coloured but the credit/watermark is absent and the original is not in the public domain or easy to locate for whatever reason.

Add to this film such as orthochromatic which throws up very unusual results of certain colours and again you're adding to the work of trying to accurately identify markings. But it will be an uphill struggle as an incorrectly coloured image will become accepted by many and I have seen these interpretations make it into print thus becoming 'fact'.

Colouring can be seen as harmless fun or allowing people to get in touch with the past but it also confusing matters as well and generates more work.

So I can help provided a bit of understanding from an archival standpoint. Full disclaimer I am not a practicing archivist but I am two and a half years into my graduate program on Archival Administration as the cert is called. I was just taking a class last Spring on audio and visual resources and we touched on this very issue in the class.

As archivists our job is primarily to preserve the primary documents that are entrusted to our care. Restoration is generally only done when an exceptional artifact is in danger of being destroyed completely or losing the important parts of context that make that artifact exception. This is also because restoration super duper expensive and time consuming requiring expert knowledge. So keep this in mind while thinking about the issue of colorization of photos, that archives primary service are to preserve documents.

Now a short stint into the history of photography. I don't want to get into too many specifics so I'll try and keep this brief and please trust me this is important to my main point. As the medium of photography was developing, not everyone was utilizing the same type of film, which leads to a rainbow of different print types and techniques from Daguerreotypes to Ambrotypes to the ever beautiful cyanotype and beyond. These types of film were generally impressions of an emulsion in a sheet of metal that would have a tint of color to them. To get a more complete history of early film checkout the Northeast Document Conservation Centers preservation pamphlets where they provide a plethora of information on film and other mediums. Particularly important to this conversation though is the common practice of adding color and blush to faces in these photos. Not only is the practice of adding color to images an old custom, but not a particularly uncommon one. Even in the early stages of the film industry the film was often hand painted after being shot for the audiences viewing pleasure.

But also common at this time was the alteration of photos, and this is where we get into muddy waters a bit. People have been altering photos since they have been able to make them, and though the techniques and results varied people bought it. There are many examples of historical photoshop but the best example has to be Stalin era Soviet photos such was the ones seen here. I don't want to insulate that colorizing historical photos is at all comparable to Stalin editing out his political opposition, but I would posit that the colorized image is not the same as the original.

There is no way that, when colorizing an image, you can guarantee that the colors being used are the "true" colors that were present in the scene that the photograph was taken. The fine folks who do this work, and this is my speculation, can look and surmise from the context, the angle of the shot, and the direction of the lighting what color may have been in place, but unless they were there on the scene they can't possibly have a 100% accurate coloring of the setting. And I don't want to knock the photo artists who work on these images, they do great work and can connect with people differently than we can in the archives, but it's just that.

Indulge me a minute and consider the scene in Schindler's list where Oskar Schindler over looks the liquidation of the Warsaw Ghetto and we catch a glimpse of color, a respite in a sea of black and white and grey. How much impact would that scene have if that girls dress wasn't colored pink? This was a deliberate choice for the film which total works in the medium, but it does make us worry that historical images could also be removed from their context and humanized to the point that we cannot historically analyze them anymore. What you are viewing is not actually the past, it is an artistic recreation of the past for modern eyes.

Remember my point at the beginning is that at an archives are meant to preserve the past as it came to us, so that future generations can appreciate the history that we have inherited. We present the primary sources as they come to us so that they may be critically analyzed to better understand the past, but colorizing images can take away our capacity to do this.

With all that said, I do quite enjoy the colorized images, but I often have to tell myself that they are not the actual image. There is so much more that can be said on the subject of photography and its importance as a medium in the archival world, but I need to stop somewhere. I hope that I have made a mostly coherent point and that someone can glean some information from this.

I would argue that the rejection of post-coloured photos and videos can be seen as part of a meta-debate that has been discussed in the historical sciences for ages, sometimes more, sometimes less openly.

The different opinions of the historians quoted in the Wired article and those of the people here in the thread show that there are different views as to which is the best way to achieve the greatest possible "understanding of history".

Elizabeth Peck, an employee of a company that makes colour algorithms, says:

[....] these upscaled and colorized ones give a new dimension to the experience. They’re fun. They make the past seem closer. “That is something that our clients and even the commenters on YouTube have pointed out consistently,” Elizabeth Peck of Neural Love told Wired. “It brings you more into that real-life feeling of, ‘I’m here watching someone do this’, whereas before you’re looking more at something more artistic or cinematic Peck compares Neural Love’s work to an installation at the Salvador Dali Museum in Florida which manipulates images of the artist to make him ask visitors for a selfie: “It's making something that's more palatable to a modern generation, which is used to interacting with media in a really different way.”

Non-historians do not know or do not care about the exact procedures of scientific work and what happens if they are not followed. Above foremost, watching a YouTube video or scrolling through an imgur gallery is not a scientific exercise for them or one that has to be carried out according to any rules at all (nor is it for me).

Dr. Luke McKernan has undeniably has a great deal of experience and expertise in the matter. He is a media historian and is particularly concerned with the issues of colour (in film/image) and how historical film material affects viewers. In his role as lead curator of news and moving images he will also have a lot to do with digitalisation. He has been involved in this field since 1993, when he wrote about the presentation of news material from WW 1 in TV/Film at the beginning of the nineties:

The lack of drama[in, or] the lack of any footage at all [...] leads to false close-ups of explosions actually filmed a mile away, scenes from fiction films presented as actuality, scenes of evident fakery. [Audio and video] from a wide number of sources of differing dates edited into a hodgepodge, montage falsely representing any one place or time. [...] Perhaps worst of all is the deliberate playing of silent film at the wrong speed." "[This indicates] that the producers found the material inadequate or ineloquent."

He goes on, it is clear that McKernan does not like the editing of old footage much - and considering things like "History Channel" documentation, I think everyone can understand that sometimes.

What does he think now, after two more decades of working on the subject, are there any new insights? In his blog post on "They shall not grow old" he writes 2018:

There is an argument for the colourisation of footage from the First World War. One can say that the original film is, of course, not reality, but a reflection of reality. Overlaying it with colour is only a further treatment of that reflection of reality, a way of looking at the past rather than the pretence of being the past itself. [...]
Such arguments can be made [...], but it is a nonsense. Colourisation does not bring us closer to the past; it increases the gap between now and then. It does not enable immediacy; it creates difference. [...]
If we want to encourage a new generation to understand [...] we should be inviting them to look at the films as they were made and through that effort to appreciate them for what they are, and what they meant in their time. It's the effort that creates the understanding. Without that there is no true sympathy, only false sentiment. Film that looks like it was shot last week belongs only to last week. [...]
Yes, on some occasions archive film can and should be manipulated for particular ends. It need not always be treated reverently in its original form alone - that way elitism lies. But using it to show what it is not does more damage than good. If we want people to understand the past, we should not be colouisring it.

I could not find a more detailed explanation for McKernan's rejection of colourisation in his publications. It does not seem to be based on a sociological, psychological or educational theory, but on his idea of what "an understanding of the past" creates.

Dr. Emily Mark-FitzGerald has a degrees in Arts Management and a PhD in Art History. Besides her research, she is also active in the field of museums and teaching.

"The problem with colourisation is, it leads people to just think about photographs as a kind of uncomplicated window onto the past, and that's not what photographs are."[...] "Getting people interested is one thing, but Mark-FitzGerald says there’s a need to critically assess what you’re seeing rather than passively absorbing whatever comes onto your Twitter feed. On the internet these images, she says, “come unmoored” from how and why they were made, and how and why they were changed."

This statement can be seen as an expression of a pedagogically oriented view of history. The problem is not necessarily that an untrained viewer misinterprets a single coloured photo/video, but that it subconsciously gives him/her a false image of them as a source. Precisely because it gives the impression that one can travel "into the past", it desensitizes an unsuspecting audience to the numerous pitfalls of obtaining information from historical sources.

All points of view are different because they are based on different basic assumptions. A great many people have (as already written) been dealing with this topic for a long time. How history is best understood and communicated is a question that probably cannot be answered (alone) by history scholars.

​

McKernans 1993 Paper on News Shows

His 2018 blogpost on the WK1 Movie

The wired Article: YouTubers are upscaling the past to 4K. Historians want them to stop