Wikipedia:Featured picture candidates/Prothonotary Warbler (Audubon)

Voting period is over. Please don't add any new votes. Voting period ends on 7 Jul 2014 at 18:53:16 (UTC)

 
Original – Painting of Prothonotary Warbler by John James Audubon
Before processing
After processing
 
Painting of Ivory-billed Woodpecker by John James Audubon (Duke of Portland's edition in near mint condition)
 
Pseudo detoned image (half size)
Reason
This image has both historical and encyclopedic interest, being an Audubon engraving. The resolution is excellent, extending to individual marks of the engraver.
Articles in which this image appears
The Birds of America, Prothonotary warbler
FP category for this image
Wikipedia:Featured pictures/Artwork/Others
Creator
Mturtle
I've been trying to get a better threshold mask. The problem is that the underbelly of the lower birds and the wings of the upper bird are in part painted in the same colour as the background, so that you're confronted with some quite delicate repainting. As for that background tone there's no prospect of recovering the original and I can't find a reference image on the web. Coat of Many Colours (talk) 21:05, 1 July 2014 (UTC)[reply]
  • The foxing has been removed. What remains might well be an artefact of inking (I mean I don't know). You're examining it at a very high magnification to notice it. In any case what part of WP:FP? says every little bit should be clean? Note there is some license granted for historical images. I should be very sorry to see the text go over a little shadow. Coat of Many Colours (talk) 20:56, 27 June 2014 (UTC)[reply]
  • Comment - I'm not sure that every bit should or needs to be clean (in fact the background looks artificially clean, no texture of any kind) but the restoration of the image should be consistent. Some places have been cleaned, others remain original. There is at least one area in the design where the restorer has made a few hesitant strokes to clean, but then abandoned the area (directly underneath the upper bird's leg). It's a very nice image, but the restoration is not finished, in my opinion...--Godot13 (talk) 23:01, 27 June 2014 (UTC)[reply]
Looking at the numerous image uploaded, I would say that the background was entirely "clean" (I mean many of these images are mark resolved, so one can clearly discern toning artefacts), and that's confirmed by facsimiles such as this. I can see what you mean by the hesitant strokes. As far as I know you can't deal with foxing (toning) in even very sophisticated image processing suites (perhaps Crisco can confirm?) You just have to get in there with a rubber (as Juliet remarked to Romeo) and WP:FP? does make a concession here: "Exceptions to this rule may be made for historical or otherwise unique images. If it is considered impossible to find a technically superior image of a given subject, lower quality may sometimes be allowed". Coat of Many Colours (talk) 23:47, 27 June 2014 (UTC)[reply]
Please don't get me wrong (as you are preaching to the choir regarding an understanding for historical or unique images). There are cases for no restoration, there are cases for light restoration, and there are sometimes cases for heavy restoration. This however, in my opinion only, is incomplete restoration. That doesn't take away from the value of the image, but it does make it much more difficult to become featured.--Godot13 (talk) 00:17, 28 June 2014 (UTC)[reply]
Absolutely not and there you go! Coat of Many Colours (talk) 00:36, 28 June 2014 (UTC)[reply]
Oppose
  • The "restorer" has ignored the fact that the paper on which the image was printed was not white, and the paper on which the image was created would not have been stark white either. This shows a blatant disregard for the intentions of the artist who would have taken the colour of the paper on which the artwork was created into account.
  • The subtle shading of blue and green has been diminished so that the areas that in the "unrestored" version are slightly iridescent, in the so-called "restored" version have lost their sheen.
  • The loss of detail to the borders of the individual feathers of the wing are clearly apparent from these two details.
My recommendation is that the colour of the background is restored to the image, (and by "restored" I don't mean "removed") and that the worst spotting and damage is cleaned up, and possibly the background lightened a little in tone to counteract the effect of aging, but with the integrity of the work maintained. Amandajm (talk) 08:36, 28 June 2014 (UTC)[reply]
First of all I have to defer to your expertise about background colour (see also my further remarks below). I have no idea what the prints originally looked like and assumed the restorer knew what they were about. I did look at this virtual presentation of an edition and they simply looked foxed to me (the whole page). As I mention above there was an earlier image that made it to "Featured". That too has a white background (whereas the original has a mustard yellow background) and the issue wasn't raised at its discussion. Finally facsimiles such as this confirm a white background.
But that is all justification after the fact. I readily confess I just assumed the paper was white without thinking about it. Whoops. But the whole set tinted mustard yellow? It does sound slightly implausible to me. On the other hand this Christies set does indeed have tinted backgrounds.
Obviously if the community can't reconcile itself this time round to a white background, then we can't feature it. Coat of Many Colours (talk) 11:07, 28 June 2014 (UTC)[reply]
I'm tempted to rerestore from the original. Adam Cuerden (talk) 07:22, 29 June 2014 (UTC)[reply]
Since I initiated all this I suppose I ought to offer, but the fact is I haven't the faintest idea how to go about it. Looking at the original featured image, the cleaning seems to have been carried out at pixel level. I saw a web source about a tecnical process for removing foxing using a technique called inpainting here. I've withdrawn my nomination for this image incidentally. Coat of Many Colours (talk) 08:39, 29 June 2014 (UTC)[reply]
Been researching these editions. One modern facsimile, the Openheimer Field Edition does indeed use white backgounds. The Commons category contains a complete collection of prints from the University of Pittsburgh edition. That is an example of the original Havell edition. These engravings were aquatints (explaining the speckling Kaldari noticed above and which puzzled me - incidentally Mary Cassatt was a master of the coloured aquatint - this is one of hers I uploaded to Commons) which were then hand painted with water color. Comparing the examples on the Audubon Society pages with the Pittsburg edition, you can see that the Pittsburg edition is very heavily toned and that it would be quite impossible to restore the original aquatint wash, thus our restorer's (I think we can bin the two-finger quotes in the circumstances) decision I expect to provide a white background on her restoration. The Christie's set I mention from the Duke of Portland is much better but they aren't reproduced on their site (after dezoomifying I mean) in such high resolution, but it should be possible to recover the aquatint tone from their reproductions. I will upload at least the Prothonotary Warbler image into a Duke of Portland subcategory on the Commons page should anyone like to have a go. Coat of Many Colours (talk) 11:25, 29 June 2014 (UTC)[reply]
Prothonotary Warbler isn't among the 40 or so Christie's images. I'll upload the version of the featured image mentioned above. Coat of Many Colours (talk) 12:33, 29 June 2014 (UTC)[reply]
Done (thumbnailed) Coat of Many Colours (talk) 20:40, 29 June 2014 (UTC)[reply]

@Kaldari:, @Amandajm: I've uploaded a pseudo detone image for the Ivory-Billed Woodpecker (thumbnailed), complete with the naive Mathematica algorithm I used. It's only half-size because I've only got my laptop with me and it doesn't have enough memory for the full-size. I'll upload the Prothonotary Warbler later. Might be useful, but I shan't nominate it for "Featured". Coat of Many Colours (talk) 19:14, 2 July 2014 (UTC)[reply]

@Coat of Many Colours: Defoxing isn't something that can be properly done by algorithm, IMO. It requires careful work in the 'levels' and 'hue/saturation' interfaces in Photoshop. In particular, you want to completely remove ("dodge") the range of tones that comprise the page color without negatively affecting the tones in the artwork. Sometimes this isn't even possible, but Photoshop definitely makes it easier. Sometimes you can get a pretty clean de-foxing just by using the input levels graph in the levels interface and the set whitepoint tool. You'll almost always need to do some manual clean-up afterwards though. Kaldari (talk) 23:32, 2 July 2014 (UTC)[reply]
Also, I should mention that I can notice the artifacts of the incomplete defoxing of the warblers image even at thumbnail size. If it isn't noticeable to you, you may have your monitor contrast too high. Kaldari (talk) 23:35, 2 July 2014 (UTC)[reply]
Yes, I'm sure that's right (and I do make that clear on the Commons page). However I'm frankly surprised that this naive algorithm does so well and I'm planning some tweaks based on Mathematica's HistogramTransform function (taking the virtually pristine Duke of Portland's edition as a reference work, but I'm constrained by RAM limitations). I think it's worth persisting because it's very unlikely high resolution images of the Audubon engravings other than the Pittsburg ones, sadly toned in my opinion, will become available. But I shan't maintain the debate here. Thanks for your comments. Coat of Many Colours (talk) 23:49, 2 July 2014 (UTC)[reply]

Not Promoted -- — Crisco 1492 (talk) 22:46, 7 July 2014 (UTC)[reply]