User talk:Jheald

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Archives[edit]

Photoshopping old prints[edit]

Hi, I see that you are involved with handling scans of old prints on Commons and thought you may be able to give me some guidance on how I should process images that I wish to use on en:Wiki.

When uploading scans of prints from old books I'm unsure as to how much I should manipulated the image in Photoshop. Should I make the background white and convert to a grayscale image? How tightly should I crop the image? Should I try to correct imperfections and brown stains? As an example, I recently took this picture from the Internet archive: http://archive.org/stream/voyagedanslesoud00mage#page/211/mode/1up and uploaded File:Palais d'Ahmadou à Ségou.jpg to Commons (replacing an earlier upload). Would it have been better to keep the yellow background? Was it a good idea to crop the image and remove the original caption?

With the uploading of images to Flickr by the British Library, some of the images that I've uploaded in the past are now available at higher quality than I could obtain from Google/Gallica/Internet Archive. The above image is here: http://www.flickr.com/photos/12403504@N02/11085655303

When using the BL Flickr images how should I proceed? Should I upload them to Commons as is and then make a separate derivative that is rotated, cropped and perhaps colour corrected? Many thanks Aa77zz (talk) 14:14, 20 January 2014 (UTC)

Hi! Many thanks for your very good questions.
I don't know if I know all the answers. There are probably forums on Commons that would be good places to ask these questions also, and other users who may be able to give better more authoriatative advice. But here are some thoughts anyway.
Looking at my Commons user page, the material that I think I worked hardest to clear up (at least from a uniform source) was the Commons:Le Keux, Memorials of Cambridge series.
If you look at the image description page for one of the images in that series, you can see that I did some quite aggressive compensation for the background of the image, followed by further bleaching out of the background, straightening, and finally cropping to a standard size of border. I was using Gimp v2.8 for software, on images taken at the Internet Archive, and found that a useful tool was the 'divide' filter -- a characteristic of the University of Toronto scans was that the middle of the page was lit more brightly than the edges. I found if I found a 'blank' page from the book, and applied a blur filter to it, that gave me a good baseline reference that I then could use to 'divide' the scan by, to get rid of most of the effect of the background (and deal with most of its non-uniformity). That may be less of a problem with BL images though -- they seem to have been illuminated more evenly.
I also did then usually use the 'curves' tool to further adjust the contrast, and I think I usually did convert the images to greyscale. (I can't remember if I usually did so on the basis of 'luminosity' or 'lightness' -- Gimp gives both options, and I'm not sure that I was necessarily consistent.
I'm not at all sure I got it 100% correct though -- looking at the page of thumbnails, they don't look quite "right" to me. Perhaps they're all just a bit too dark, suggesting I overdid the contrast enhancement; or perhaps they're just a bit too even, too smooth, or too silvery grey. I can't quite put my finger on it (and I'd welcome anybody's input or thoughts), but to me there's some attractiveness of the original that I've lost, and I'm not quite sure why. So this may be an example of over-cleaning of images.
Some more recent work was on Commons:Category:Ackermann's Repository - London views. There are probably mistakes here as well, but I think the results are a bit happier (though dirtier, because of the quality of the source material). Colour images are more fogiving in a way, but also you can't push the cleaning as far because (a) you're not trying to get to pure black and white, so you have to be more protective of the colour; and (b) I found that, at least with Gimp, if I did push it too far, I found I'd start getting very unrealistic pinks and blues starting to appear, particularly in the background and around the borders. (Though that may just because I don't know how to fly the software well enough, or perhaps strictly speaking because I ought to be working in a different colour space, eg something like CIELAB, cf HSL_and_HSV#Disadvantages. But Photoshop perhaps works in that more advanced colour-space anyway?).
One thing I did do, having again used the divide trick to reduce the effect of non-white backgrounds (this time just by sampling a point in the background, and then turning into a layer), was then to multiply by a uniform layer of off-white colour -- looking at an old workfile I think I used #fbf7ea -- which makes the image a bit warmer, and seems to correct the overkill of the colour-correction itself. I think this probably is a good idea, though I don't know the best off-white colour to give black-and-white images.
But "your milage may vary" as they say. I think it probably really depends on how you want the image to look on any eventual Wikipedia page or gallery page.
In terms of filenames, I think it is okay to use the same filename from original upload to what you consider is your "best version" of the original image. One good thing about the Wikimedia upload system is that it makes it very easy to see previous versions, and I think it is good to deliberately leave an audit trail of the major changes you have made, so people can see what the original version of the image looked like, and what treatments you have applied to it (in case they may to try different choices).
As regards cropping, there are different points of view. Myself, I quite like to know what the original caption text was, and especially if there is any information there about the artist and the engraver. So I tend to leave any caption, plus a little white space below it, and a similar amount of white space around the image. But on the other hand, it's undeniable that larger images make more impact, and you will get an appreciably image in a thumbnail on a Wikipedia article page if there is less surrounding white space. So a lot of people like to crop right on the boundary of the image -- and I would quite likely use one of their crops if I was selecting the image to use on the WP page. So I think this is a situation where there can be a case for having two images -- one cropped to leave in any caption, and then a second cropped variant that is more tightly cropped just to the actual picture.
But in all of this, it's entirely up to you as to how much work you want to do, and how far you yourself feel you want to go -- with an eye to how you want to use the image, or how others may want to use the image if they find it for example through Google Search. For example, looking at some of my most recent uploads,
File:Heinrich Barth's route through Africa, 1850 to 1855 (Deutsch).jpeg,
File:Routes of European explorers in Africa, to 1853.jpg, and
File:View of Constantine, Algeria, 1899.jpg,
the first two are exactly as I found them -- one from the BL digital collection, the other off the internet -- as really is the third one, apart from my having cropped it.
So I guess my answer is, do as much as you feel you need to do to make the image seem as useful and usable as it can be; but it doesn't have to be perfect (and probably never can be).
Finally, you asked about over-writing existing versions of files. I think the answer, again, is to follow your instincts. File:Harlaxton Manor Morris.jpg is an example of a file where somebody has done just that, entirely reasonably I think, and also with the advantage that they don't have to edit Wikipedia pages in 5 different languages to point to a new filename. On the other hand, both Commons:Category:Bury's Coloured Views on the Liverpool and Manchester Railway, 1831 and Commons:Category:Pyne's Royal Residences include some different variants of the same view, and that's also fine I think. If there's something identifiably different about the other view, eg a different caption, or a different 'state' of the engraving, or a different colour scheme, or perhaps if it is from a different source, then it may well be worth having two different versions. Equally, if somebody else has done a lot of image enhancement on the other image, or simply if it doesn't fit into your preferred naming scheme, then it's probably uploading your own new version. But otherwise it's up to you I would say.
And now I seem to have written an awful lot, without really having given you very many very clear answers. Again I should caution that there may be people that are far more knowledgeable than me on Commons, both as regards best methods/use of technology and best practice. But thank you for giving me the chance to try to put my own thoughts in order, and I hope you can find at least some counsel of use in the above! All best, Jheald (talk) 19:18, 20 January 2014 (UTC)
Wow - many thanks for your detailed reply. How much to "improve" an image is always going to be a tricky decision. Here is an old photograph that I uploaded: File:Edouard Baldus Tour Philippe le Bel c1862.jpg. If you zoom-in you'll see that there is a brown/orange stain to the left of the tower - it would be trivial to remove (photoshop healing) but on this occasion I resisted the temptation. Aa77zz (talk) 13:29, 21 January 2014 (UTC)

File:Hornby early logo.png missing description details[edit]

Dear uploader: The media file you uploaded as:

is missing a description and/or other details on its image description page. If possible, please add this information. This will help other editors make better use of the image, and it will be more informative to readers.

If you have any questions, please see Help:Image page. Thank you. Message delivered by Theo's Little Bot (opt-out) 04:42, 22 January 2014 (UTC)
Fixed. Jheald (talk) 15:06, 22 January 2014 (UTC)

Le Keux images[edit]

Honestly, I'd say most of the issues with these come down to:

  1. The originals have a lot of JPEG artefacting.
  2. The effective resolution of the originals isn't enough to show the microscopic lines that make up steel and copper engravings, giving them a photogravure-like appearance that doesn't really match engravings as we know them
  3. The white point is set to remove all paper texture, which can look a little unnatural with this kind of image.
  4. They are completely desaturated. Most paper and ink has a slightly warm tone to it, whereas desaturated grey tends to appear somewhat on the cool side.

In all honesty, it's rather good work with a mediocre source. This is why I prefer to self-scan engravings as much as possible; they're much harder to get good sources for. Lithographs are far easier. Though, I'll be honest: I still struggle with getting colour right, pretty much every restoration. It helps a lot to work from originals so you can compare, though this is, of course, impractical in a large number of cases.

As for your advice, I'm honestly a little surprised you use layers. I've never seen much need to go beyond the Levels, curves, clonestamp, healing brush, hue-saturation, and select tools. But then, I'm often rather unhappy with my background borders, so I suppose I can't say that's sufficient.

Oh, one bit of advice I'll add: You'll often need to deviate from it by the end of the series - especially if the paper in the book varies a lot - but note down exactly what you do to the first image of a set, saving any settings you used, and you often have a big start on making a consistent set. Don't force everything to match those instructions - you'll screw yourself over that way - but it's a start, and helps. I did that with, for example, the Puck of Pook's Hill set. In fact, I have those instructions still. They're applied after the cleanup is finished.

* 320 px border all sides but bottom. Bottom ~476.
  • 3332x5140

Settings :

  • Saturation: -30
  • Colour Levels: Puck of Pook's Hill
  • Sat -30
  • Curves: Puck 2
  • Text: Curves: Puck text Text often looks better if darkened more than the image. A good way to do this is to use curves to keep the values that relate to the paper exactly the same, while darkening the levels related to the printing a bit.
  • Final adjustments as needed


Alt: Sat -50, Puck 2 levels, Sat -30, Puck 3 curves Puck Part 2, Puck Text, -35 sat, adjust white point.

Those are applied in that order; most of those are linking to saved, pre-done adjuments. ("Levels: Puck of Pook's Hill" means "Go to levels, appy pre-made setting 'Puck of Pook's Hill'"). You'll probably note that a lot of the images don't actually fit the dimensions given. And I did need to adjust the curves and levels for some images (extremely for some more faintly-printed ones in the middle of the book.)


I think the biggest obstacle is the quality of scans. A bad scan - one where the page isn't pushed flat onto the scanner, and where the resolution is too low, or the image saved at low-resolution - can screw you over. Chromatic fringing leads to the pink and blue spotting you mention - a GOOD scanner will avoid that. I find that I need to replace scanners every few years, when that problem develops. Adam Cuerden (talk) 18:58, 22 January 2014 (UTC)

Oh, one more thing: I think the amount of cleanup justifiable depends on the medium. When you're cleaning up an engraving, you're basically trying to create - of the hundreds, thousands, or even hundreds of thousands copies out there - an ideal example. With a painting, all I could really see fixing was fixing the image to better resemble the original (hue/saturation, levels, curves, crop, maybe, perspective adjustments if the photograph was taken at an angle), and perhaps - at the most - combining a few copies with different lightings to remove excessive light reflections. The painting is a singular object; the engraving multiple. Adam Cuerden (talk) 19:02, 22 January 2014 (UTC)

Huh. Though, mind, saying that, I don't obey that rule all the time, as there's a competing interest. With old photographs, there's a type of common damage - the albumen being scratched off the glass plate - that is particularly distracting, as in, it makes it difficult to focus on the image itself. In that case, a restored image that removes the damage and thus puts the focus back onto the subject of the image can be very helpful to the encyclopedia. Paintings are generally notable as paintings; photographs for their subjects. Adam Cuerden (talk) 19:08, 22 January 2014 (UTC)

If I had to guess about the pink and blue - particularly if the image started off a little yellow or the like and got adjusted to white, it's not hard to get something slightly off - a white that's, say, 253, 249, 247, or the like. If you upped saturation on that enough, it'd look pink, while appearing white before that. And a lot of people are going to be using auto settings, which have a high chance of getting those slightly off colours. There's natural variations int the image, after all. And particularly with curves adjustments. It's not really something to worry too much about. In a worst case scenario, I'd just completely desaturate the image, then add a little yellow back in by manually adjusting the levels (there's a box where you can set the white point for blue at, say, 250, adding a bit of space on the right.

As for rotation - honestly, I've never been able to notice a difference. But I wouldn't worry too much about making very slight adjustments if it looks alright already. Your bigger nemesis is saving as JPEG before the very last moment - repeatedly saving as JPEG will degrade the image rapidly. I tend to use PNG. Lossless, but smaller file size and better supported than TIFF.

Oh, and save the JPEG at at 99 or 100 quality. Adam Cuerden (talk) 23:36, 22 January 2014 (UTC)

By the way, I've actually updated Durova's instructions a bit. Some of what she did came back to bite her at times, like cropping and then having someone object to the crop, after she spent hours restoring. Adam Cuerden (talk) 23:41, 22 January 2014 (UTC)

An RfC that you may be interested in...[edit]

As one of the previous contributors to {{Infobox film}} or as one of the commenters on it's talk page, I would like to inform you that there has been a RfC started on the talk page as to implementation of previously deprecated parameters. Your comments and thoughts on the matter would be welcomed. Happy editing!

This message was sent by MediaWiki message delivery (talk) on behalf of {{U|Technical 13}} (tec) 18:27, 8 March 2014 (UTC)

gwtoolset handling of special characters[edit]

Just a heads up and to keep you in the loop, I filed bugzilla:62909 for the issue of gwtoolset converting apostrophes and parenthesis to dashes. Bawolff (talk) 05:44, 21 March 2014 (UTC)

Merge discussion for LBC[edit]

Merge-arrows.svg

An article that you have been involved in editing, LBC, has been proposed for a merge with another article. If you are interested in the merge discussion, please participate by going here, and adding your comments on the discussion page. Thank you. Khairul Islam 00:43, 3 June 2014 (UTC) — Preceding unsigned comment added by Khairul Islam (talkcontribs)

Disambiguation link notification for July 12[edit]

Hi. Thank you for your recent edits. Wikipedia appreciates your help. We noticed though that when you edited Transatlantic Trade and Investment Partnership, you added a link pointing to the disambiguation page Anti-dumping. Such links are almost always unintended, since a disambiguation page is merely a list of "Did you mean..." article titles. Read the FAQ • Join us at the DPL WikiProject.

It's OK to remove this message. Also, to stop receiving these messages, follow these opt-out instructions. Thanks, DPL bot (talk) 09:00, 12 July 2014 (UTC)

TTIP and GMO crops[edit]

Are you sure that the EU negotiators have a red line on GMO? I thought the issue was that that the EU members only want GMO to be labeled as such (and the products could then be offered for sale - making it down to condumer [and retailer] choice to buy or not) but the US disagrees. I don't have a source for either position [(a) ban or (b) label]. --John Maynard Friedman (talk) 16:20, 22 July 2014 (UTC)

OK, tyvm. I knew that growth hormone beef was a red line. I'll look at the text again to see if it was the phrasing that misled me. --John Maynard Friedman (talk) 16:52, 22 July 2014 (UTC)