User:MIckStephenson/Preservation of historical images

This discussion is being started with the goal of establishing a consensus among project members regarding what kind and amount of retouching of historical images such as the example should occur for images to be used in articles here on the English Wikipedia. Sswonk (talk) 01:57, 13 September 2009 (UTC)

Related discussions and existing policy; please read first
{| class="navbox collapsible collapsed" style="text-align: left; border: 0px; margin-top: 0.2em;" ! style="background-color: #f2dfce;" | Original discussion from History of Photography talk page
 * style="border: solid 1px silver; padding: 8px; background-color: white;"|
 * style="border: solid 1px silver; padding: 8px; background-color: white;"|

"Cleanup" of Daguerreotype
I recently reverted the placement of the right hand image above in the article John Quincy Adams. While I appreciate the effort I have major problems with the intentions and results of the editors who did this. The original daguerreotype image was so significantly altered as to destroy any historical context it possesses. The left hand original image was originally from the Library of Congress as can be seen on the. These edits were made in the past two days. I am concerned enough about the potential for abuse to alert the member at this project, and I would like to know what your opinions about it are and if anyone has had experience dealing with such "cleanup" efforts in the past. Sswonk (talk) 01:37, 31 July 2009 (UTC)
 * Does everyone agree that such cleanup work has the potential for abuse? Maybe, but I am at least open to the possibility that there are two sides to this issue. --DThomsen8 (talk) 16:57, 1 August 2009 (UTC)
 * Agreed, there is always potential for loss of important and informative detail from the POV of a photo historian. OTOH there are two scenarios in which you might forgive a partial or full restoration. First, if the unretouched original scan is retained and prominently linked on the restored image file page. Pictorial value in the encyclopedia could (arguably) be improved with both versions available. Second, if the original scan came with a poor restoration attempt (as this one did with some weird, poor-quality smudge/blur thing employed to obscure large longitudinal scratches) that clearly would have been better left alone. This particular edit, had it been more accomplished, might have made a better encyclopedia image but it's only borderline passable but probably an improvement. It's worth mentioning the wider ethical point that anything done to retouch historical images risks adding non-authentic detail etc, ie we cannot possibly know what the missing emulsion on this image looked like. Cloning-in adjacent detail is the image equivalent of WP:OR and strictly speaking, you're creating a new image and attributing "original document" status to it. --mikaultalk 21:30, 1 August 2009 (UTC)

(outdent) I am a hard liner on this whole process, I think anything that is done in this manner constitutes original research and can only be viewed as an interpretation of the work. My feelings notwithstanding, there are now more examples of this work cropping up. Here is the original request at the Commons "Graphic Lab School" (!) &mdash;scrolling down on that page will reveal two newer requests. I think as long as these "restorations" continue to go unchallenged, many more will be done and the historical record will be diluted. I fear the consequences of an acceleration of this work once more people notice the photos that have been "fixed" so far. I don't know who to turn to or what to do to stop it, so please let me know if there is any recourse available. Sswonk (talk) 18:49, 4 August 2009 (UTC)
 * You could raise it at the Village Pump (policy) but I can imagine the response: that the image pool is "enriched" by the provision of retouched alternatives, while the historical record remains intact. I've argued this same point here in the past among sympathetic editors and found a lot of indifference and even opposition towards retaining damaged photographs in their original state within articles. Personally I think the originals have an indisputable place in articles on photographic process or the photographer but you'd have to fight for those, even, in the case of the examples you point out here. Articles about the subject would invariably favour retouched versions. What can you do? It might b a hard-fought battle, but you might get support for captioning the use of these restorations as "interpretations" or "artist impressions"  or somesuch. I wouldn't hold out much hope though. --mikaultalk 20:05, 4 August 2009 (UTC)
 * I would be fully supportive of the use of these images in articles about their subject, and would also be happy to see a caption indicating that it is retouched. On the other hand, for articles where it is the photograph or photographic method itself that is the subject, I would agree that the retouched version should not be used.
 * However I object strongly to this wording used by Sswonk: "I have major problems with the intentions of the editors who did this." I believe this to be an unfair slight against the character of myself and the other editors who have contributed retouched images of old photographs (albeit, in my case, not daguerreotypes), and would politely request that Sswonk retract that remark.
 * Speaking for myself, I performed work on the image of John C. Breckinridge. This was made entirely in good faith, with an honest view to producing a useful image. I in no way wish to destroy or corrupt the historic record - on the contrary I am very much in favour of it being preserved: the image's description page links directly to the | original source image at LoC, and also to a version of the  original hosted on Commons. ∙  AJCham   talk  21:34, 4 August 2009 (UTC)
 * I am not retracting the remark as requested. The edit summary of in his revert of my revert to the John Quincy Adams page stated: "I think the cleaned one represents the image more clearly, without distracting elements." I do not agree with those intentions. My statement said nothing about your character, or even whether I thought that the edits were not done in good faith. On the contrary, I am absolutely convinced they were done in good faith. I disagree with the use of modern techniques to alter the historical records in this way: I want to see the original photograph from the Library of Congress, not what your requestor Connormah views as better looking. In my view it is a case of good intentions but done without forethought as to the implications of what is being done. The image at John Quincy Adams was not properly cited when I originally posted this topic, and you have provided the Commons description of your work properly. Therefor, you should not take offense, and I will not retract. Sswonk (talk) 22:00, 4 August 2009 (UTC)
 * If the intent here is to illustrate for the reader what John Quincy Adams looked like, rather than what a particular now-damaged print of a photographic negative looks like in its present condition, then surely some photoediting is in order. For consistency, preserving the record, and all the other reasons named it would be best to develop a template that allows the reader to see either the "original" (which, in most cases is not an original at all, but somebody's scan of a print or negative) or the retouched / restored version, along with some explanation of what was done to the image and by whom.  Surely this is a broad, long-term issue in serious photography circles.  The world is full of botched art restorations, so many of them are misguided.  We will develop norms here.  WP:OR does not apply, that relates to sourcing of text.  Here it is a consensus / process for choosing which image to use.  What is the standard for scholarly works, exhibits, museum archives, and the like?  Why don't we follow their lead?  Wikidemon (talk) 02:45, 1 September 2009 (UTC)
 * The Library of Congress, the original source of the image which is shown unretouched on the left, set the standard by publishing the image as shown. We don't need to find anyone leading to follow, the leader is the highly professional provider of the work. The first part of your comment is what I am arguing against, that we can possibly know what John Quincy Adams looked like any better than the original photo shows. That is pure fantasy. He was the first president photographed. As for the part about WP:OR applying to text, I realized that and this is an area that I have yet to find a better guideline covering the issue, however stretching WP:SYN a bit this is an example of using source A, a photograph, combined with source B, software filters for digital editing, to come up with conclusion C, the opinion of an editor as to what the man looked like (really, what the editor thinks make him look his best). If there is a guideline covering photo manipulation, and I am looking for one, that obviously would be more specific than WP:SYN. OR is what I am using for now to stress my point that—like presenting text in an altered form as a quotation—retouching this and similar photos has the potential for all sorts of error and we are vastly better off and the reader better served by our using the image as is. To use an extreme example of how this is misleading to readers, imagine offering the reader a ride in a 1932 Ford like the hot rods pictured in the article, but telling them "this is how these cars ran off the assembly line!" No thanks, that is not what a good encyclopedia is about. Surely talented people can fix up photos just as experienced hot rod mechanics can juice up cars, no question. But that really isn't being responsible to the historical record in any explanation. We're an encyclopedia, not a manufacturer of collectors' items at souvenir shops. That is what the retouched photo reminds me of. Sswonk (talk) 04:18, 1 September 2009 (UTC)
 * Second that, although you're relying on a fairly weak interpretation of WP:SYN. I prefer a slightly stronger reading of WP:OR, specifically WP:OI, backed up with criteria written for Featured Pictures. Wikidemon raises a couple of valid points about general article use but I do think repaired-damaged historical images should be very prominently labelled as such, especially if the subject itself has been altered in any way, and the original file equally prominently linked on the description page. Even if there was a "standard practice" in other archives we should definitely think twice about presenting 21st century interpretations as 19th century originals. Sure, negatives are the true originals but raw scans are utterly faithful copies; even prints or other reproductions are open to manipulation of some form. Without getting too purist about it, we have a duty to present historical items as faithfully as possible within the bounds of quality illustration. This means if a photograph is severely damaged and would have exceptional encyclopedic value as a repaired image, we should offer a retouched version. If an image is slightly water damaged, missing emulsion from the edges, or perhaps unevenly developed as a negative, I would argue strongly for the original, unrepaired version to preside wherever it appears, as encyclopedic value would be lost, not gained, by correcting those flaws. --mikaultalk 09:00, 1 September 2009 (UTC)

[Bounce left] I disagree with the use of modern techniques to alter the historical records in this way: I want to see the original photograph from the Library of Congress, not what your requestor Connormah views as better looking. &para; You may be in the lucky position of being able to see the original, but the closest [or not] most of us will get is either the LoC's digital reproduction thereof, or a reproduction in a book. Yes I agree that one should at the least (i) think very hard before "restoring" an image, and (ii) acknowledge that this "restoration" has been done. [Let's skip the quotes around "restoration" and variants thereof: these become tiresome.] However, it's conceivable that a restored digital copy presents the original print more faithfully than does the pre-restoration digital copy. I'd agree that this is unlikely, but it's worth at least a moment's thought. I'd be particularly interested in a comparison between these two files and a reproduction in a meticulously edited and printed scholarly book. &para; Let's agree, though, that the unrestored version is more likely to represent the photograph. Now, is it desirable to process a photograph such as this in an effort to make the result more realistic or a better representation of what the camera was pointing at? (For the sake of brevity, let's make the dangerous assumption that the photographer aimed at realism, and put aside the question of what realism consists of.) &para; I'm no expert in photo restoration, old emulsions, or gimping, but off the top of my head I can think of three things that people might do. First, the "removal" (replacement) of dirt, cracks, and other "noise". Secondly, bending the whatever-it's-called curve to "bring out shadow detail" and "unblow" highlights and so forth. Thirdly, more local fiddling in an attempt to compensate for the pre-panchromatic sensitivity. Now, I'd guess that most people would rush to approve of at least the first two of these. After all, it is, or sounds like, the still equivalent of what they've come to expect (or been brainwashed to expect) in DVDs. Moreover, their own family collection of holiday snaps and so on from color negatives will provide abundant evidence of deterioration in photos over merely two decades. &para; So advising people, even thoughtful people, that restoration is not a good idea -- this is going to be a very hard sell. Can you point to a persuasive (thoughtful, not overly long or technical, non-strident) web page on the virtue of non-restoration? -- Hoary (talk) 09:35, 1 September 2009 (UTC) However the collodion negative provided a very good reproduction, for its day; making a very large format copy of a very small format original is basically what we're doing when we make hi-res scans of these things. The LoC scan should, therefore, be considered a very close approximation to the original image and a more faithful reproduction than, say, a print made at the time. It may not provide the perfect example to illustrate a discussion on restoration ethics, but it does introduce two important and distinct "classes" of reproduction: faithful and enhanced. Among the faithful, there should be a further subdivision of original "seen" and "unseen". Copy negs and even raw scans may appear to provide veracious reproduction but they invariably change tonal compression, colour hues (including monochrome tints) and subtleties like the true depth of blacks and clarity of highlights are often lost unless the archivist takes time to properly calibrate the process. Film copies are really tough to calibrate; when I did this sort of thing years ago I'd often run as many as a dozen 10x8 sheets of Ektachrome through, using a range of filtering and processing tweaks, just to get something that closely approximated the original; it was extremely expensive and frankly unjustifiable for all but the most important works. Even though the same care taken with a digital copy is much quicker (and cheaper!) I really don't know how often it is used. Certainly the sheer volume of work archived at the LoC I know to have been a major barrier even to digital calibration, let alone perceptual comparison and correction; many later scans are made alongside colour control strips and this can allow later editors achieve relatively accurate colour correction, but crucially it's not done in sight of the original and those subtleties are lost. I'd argue that only a copy that is visually calibrated in this way (original sighted) can be considered a first-class veracious reproduction. In practice, certainly with LoC scans, we're stuck with "unseen" copies, a kind of second-class faithful reproduction. This should both temper our zeal slightly when we compare image editing to WP:OI guidelines, and provide a cautionary note to editors when making assumptions about restoring original tone and hue. It's from this perspective that we have to examine the necessary changes to repair damage, enhance and "clean up" subject visibility and generally improve illustrative value (or EV – encyclopedic value) per comments above. I also think we need to be very careful with words like "restoration" when what we more often mean is "repair". These distinctions made, we can look at best practice and other ethical issues. Without them we're likely to start at different ends of the elephant and get nowhere fast. --mikaultalk 00:54, 3 September 2009 (UTC)
 * (edit conflict!) I'll admit to having a difficult time finding anything that supports exactly what I am saying. However, to find something that covers the concept of Wikipedia's policy against original research is going to be difficult. I was referred to WP:OI which is very clear about what is considered original research in this area. Quoting:"Images that constitute original research in any way are not allowed. It is not acceptable for an editor to use photo manipulation to try to distort the facts or position being illustrated by a contributed photo. Manipulated images should be prominently noted as such. Any image that is found to have manipulation that materially affects its encyclopedic value should be deleted from the article and a note should be posted at the file page informing users that the file contains Original Research. It is also suggested that the file be posted to Wikipedia:Files for deletion."
 * Since that statement is contained in WP:NOR, policy, I think my attempt to tie the term "original research" to visual, rather than text, information is validated. The question then becomes, what constitutes something that "materially affects its encyclopedic value"? That is a loaded question and is where the argument stands in my view. I have also discussed this elsewhere on a couple of talk pages, and am currently attempting to gather a group, especially those commenting here, who will be interested in participating in a centralized discussion that will expand on that single sentence in the policy, probably by creating a guideline that emphasizes minimum retouching. Regarding what you wrote about what people might do, I couldn't support broadly allowing any of those. It should be made clear that photographic restoration must be extremely faithful to the original and any manipulation should be done with extreme care. I hope this discussion can be centralized somewhere and would like to read some opinions on where that might be. For now, my suggestion is at the WP:NOR talk page, where the statement quoted is found. However, there may be a much more logical and appropriate venue that I haven't thought of, so please weigh in on that. For amusement, readers can visit http://photoshopdisasters.blogspot.com, which I found when trying to support the virtue of non-restoration. Sswonk (talk) 03:04, 2 September 2009 (UTC)
 * I realise this project might not be the professed home for this kind of thing but as it's here, I'd suggest pointing interested parties to this discussion, at least for the time being. --mikaultalk 01:00, 3 September 2009 (UTC)
 * This particular image is an especially ill-suited example to form the basis of any serious discussion about Wikipedia's parameters for image editing. Yes, John Quincy Adams posed for a daguerrotype.  Several months ago I downloaded the highest resolution version of the Library of Congress original and determined that it was not suitable for restoration; I suspected it was not the original  Upon review, the unrestored version above is definitely not the original.  Wikidemon's and Hoary's mentions of "negatives" are a tipoff that something is seriously wrong: have a read of the daguerrotype article.  Upon review, the bibliographic notes make it clear: this image was not digitized from the original.  It was digitized from a glass colloidon negative copy made by either Matthew Brady or Levin Corbin Handy long after Adams's death.  Adams died in 1848; this image was made between 1855 and 1865.  The unedited image could not have been the original.  Jklamo's edit brought out the wispy character of the glass plate negative.  When something doesn't look like a daguerrotype, please check to see whether it actually isn't.  The only possible way to reproduce the daguerrotype was rephotography, with a consequent loss of detail.  I'll take a wild guess and venture that the negative may also have been physically retouched.  Durova  308 02:43, 2 September 2009 (UTC)
 * I noticed that in the notes from LOC as well. The term Daguerreotype as used in the title of this thread can be thrown out and replaced with "historically significant photograph". Thank you Durova. Where is jklamo in this? I see that name in the file description but beyond that I see Julielangford. Sswonk (talk) 03:11, 2 September 2009 (UTC)
 * No idea; I don't know Jklamo. The main thrust of this point is how that particular image only looks like a daguerrotype at low resolution.  The resemblance is artificial, and attempts at restoration enhance its artificiality.  That isn't necessarily the fault of the digital editor.  Durova  308 04:50, 2 September 2009 (UTC)
 * Yes, good point, I confess I didn't even look at the description page of the original upload. The glass neg copy would have provided the only way possible to reproduce the original image and may even be the only extant version of it. That notwithstanding, what we started off with is a daguerrotype that may or may not have been retouched (quite a common practice, despite their small size) that was then copied, presumably in order to make a number of reproductions, or possibly to apply more retouching on the subsequent print, this being an important image even then. Worth noting that collodion emulsions are extremely difficult to retouch – it was always done on the resulting print – so that crude repair on our upload is almost certainly digital.

Discussion at Jimbo Wales talk page
 * }

Existing policy on image integrity

Examples
Please feel free to add examples in support of your argument here, using the

I recently reverted the placement of the right hand image, or "JQA2", above in the article John Quincy Adams. While I appreciate the effort, I disagree with the intentions and results of the editors who did this. The left side image, or "JQA1" was so significantly altered as to destroy any historical context it possesses. JQA1 was originally also retouched from the Library of Congress (LOC) as can be seen on the. The LOC image is directly linked below.

LOC Digital ID: (digital file from original neg.) cwpbh 02619 http://hdl.loc.gov/loc.pnp/cwpbh.02619

LOC higher resolution JPEG version (129 kilobytes)

LOC uncompressed archival TIFF version (18 megabytes)

LOC full resolution TIFF version (79 megabytes)

--Sswonk (talk) 01:57, 13 September 2009 (UTC)

Open discussion

 * You realise, of course, that the vast majority of historic images are not photographs, but include engravings, lithographs, paintings, and other such works. Shoemaker's Holiday Over 205 FCs served 00:11, 14 September 2009 (UTC)
 * Actually the vast majority of historic images are not glass plate rephotography of daguerrotypes. Hard cases make bad law, as they say.  Durova  318 00:45, 14 September 2009 (UTC)
 * The discussion here will hopefully lead us to a benchmark for all historic images. The Adams image just happens to be a photograph, one with an interesting precedence in terms of reproduction that should serve us well as one example, with others added as & when. It's a complex issue and rather than start it as it appears here I'm planning to reference and summarise the original discussion that sparked it off, along with links to related discussions, explanatory notes etc which will (hopefully) save loads of confusion and tangental arguments. --mikaultalk 02:18, 14 September 2009 (UTC)
 * You do realise there's a fundamental difference between an engraving, and, say, a painting or daugerrotype that would make rules appropriate to the latter appallingly stupid for the former: There is no such thing as "historic damage" in a mass-produced work, since engravings, lithographs, and the like were meant to be printed hundreds or thousands of times. Any damage on such a work cannot be considered "historic". Shoemaker's Holiday Over 205 FCs served 02:51, 14 September 2009 (UTC)
 * I do realise you didn't read the discussion I linked to, where I pontificate in tiresome detail as to the likely veracity of digital copies we have here compared to their analogue originals. The distinction you're making there is relevant only to verifiability. The sort of rules that apply to this project relate to digital reformatting, almost irrespective of original media types, in which case all of our scanned images can be assessed from a common perspective. Within this there are likely to be a variety of standards, depending largely on the type and quality of calibration, but the only fundamental differences are those between born-digital images and scanned analogue images. The argument is that image editing of the former requires a slightly different set of ethical guidelines to editing of the latter. The objective here is to establish if that is the case and, if so, what those guidelines should be. --mikaultalk 07:22, 14 September 2009 (UTC)