User talk:Andy Dingley/Archive 2011 February

← Archive 1 | Archive 2 | Archive 3 →

Category:Scans from 'The Book of the Motor Car', 1912

I'm glad you appreciate this work, however there is no need to thank me, I do not want anything there to optimize the image through, because I use a good editing program, if you want to take off even the captions at the bottom of the images to me is not a problem.

It must be said however that such scans are not bad, I've seen much worse around.--A7N8X (talk) 12:48, 16 February 2011 (UTC)

Drawings in JPEG

You are doing the right job by scanning old books and uploading the illustrations here. But, unfortunately, there is a significant flaw: the drawings are uploaded in JPEG format, which is perhaps the worst imaginable for them. Please read "Wikipedia:Preparing images for upload" and, if you have the originals in a lossless format, please upload them as PNGs.

I can't give you a good reference to read about scanning techniques, but the "standard" is bitonal (also called "line art") with 300 dpi resolution for text and 600 dpi for drawings with fine detail. TIFF with CCITT Group 4 compression is an excellent storage format (although not suitable for online use, as mentioned in w:WP:PIFU).

Mikhail Ryazanov (talk) 05:40, 13 February 2011 (UTC)

It's OK, I've stopped doing scans for Commons anyway. Thanks for your criticisms though, I've added them to the pile.

Late to the party here, but seriously, get a clue, Mikhail. Your suggested scanning format is "not suitable for online use"? Uhh, what the fuck do you think the Commons is, buddy? Shut up and stop biting the ankles of people who are doing real work here. Lewis Collard! (lol, internet) 00:16, 18 February 2011 (UTC)

Pleas read the sentence again and try to understand it. The keyword is "storage", it refers to "Keep the source" advice from w:WP:PIFU. The last sentence of the first paragraph tells explicitly about what should be uploaded.

And, if you did not understand, the intention of my communication was the suggestion how the "real work" people can improve the quality of their results (without any additional effort, in fact).

BTW, look here and in future please talk in a more civilized way.

Mikhail Ryazanov (talk) 07:34, 19 February 2011 (UTC)

Andy, I have found (guess how ;-) ) a reference about drawings and PNG which you may find more convincing than my own words... -- Mikhail Ryazanov (talk) 07:34, 19 February 2011 (UTC)

Mikhail, you're just wrong on this one. PNG is a good format for things that are line art (assuming they're not even better as SVG), but old scans of line drawings don't make for line art. Your insistence on converting greyscale images to images with only a handful of grey tones leads to some very ugly artifacts and problems with jaggies when rescaling. This is particularly a problem when scanning from hundred year old books with brown paper and faded ink. We might like these to be line art, but they're now a mushy grey at best. Automatic and arbitrary conversion of their levels to force them into being pure black & white lines does not help. What about my 16th & 17th century stuff from vellum? They aren't even greyscale - are you going to convert and tag those for deletion next?

The Pearson Scott Foreman project guidelines have long been one of the best practical guides to image handling on Commons (and should be wider known), but they're working off paper in quite good condition, not antiquarian sources. Andy Dingley (talk) 11:38, 19 February 2011 (UTC)

I still don't see anything except emotional arguments. This is not the best way to support your point of view. ;-)

Now my response to your allegations:

• I've never insisted on converting to few grey tones. Some of the images I've made have 16 shades of grey only to make the files smaller. 16 was chosen because it is the lowest number (power of 2, for convenience) which does not make visible difference from the 8-bit greyscale original. FYI, television standards rarely guarantee more than 8 shades of grey, so even that number is apparently considered "good enough". Additionally, what is the real SNR of the original? The scanners usually have 8- or 10-bit intensity resolution, but the noise (of the scanner itself and, mainly, the paper original) surely exceeds the LSB level many times. Therefore, the reduced number of intensity levels simply reflects the decision not to store the noise.
• You "jaggies" claim still remains unsupported. In fact, when I compare the use of my PNG version you complain about with the use of the original JPEG, I see less jaggies in the PNG. At the same time, the JPEG has ringing artifacts, which are completely missing in the PNG. Text legibility is equally poor in both small images. What exactly do you mean by "some very ugly artifacts"?
• Scanning of old books with faded ink on yellowish paper can yield reasonably good results when done properly. You can check an example from 1649. With proper processing it could be even better. After all, the books were printed with black ink on white paper, and that was the intended appearance; the condition of one particular copy is usually of no interest here. In principle, 600 dpi lineart is considered acceptable even by modern high-quality publishers (APS guidelines). It means that bitonal scan at 600 dpi can reproduce the original exactly. In practice, bitonal 300 dpi is enough for good legibility (although it will look a little jaggy when printed), and this is the usual representation of scanned books, providing better quality and smaller file size than JPEG format could. If you look at DjVu format, which was developed specially for scanned images, you may notice that it also tries to use bilevel representation as much as possible.
• I have not seen any of your "16th & 17th century stuff from vellum", and thus can't give any comments about that.
• PSF guidelines have an example of a "not quite good conditions" drawing. I think, it was chosen on purpose to show what to do in that situation (the other 4 drawings are much easier to extract).

P.S. Concerning your complains about "description of my book scans as 'the worst imaginable'". This is simply a slander: I only pointed that they are uploaded in inappropriate format. Hopefully, this will clear the misunderstanding...

Mikhail Ryazanov (talk) 08:09, 20 February 2011 (UTC)

Over-categorization of File:Chainsaw cutting tree.jpg

What exactly did you find unhelpful about this edit? Actually you yourself should have removed the supercat in the first place as you are the creator of Category:PPE (lack of) - and you put it into Category:PPE. Please get yourself acquainted with COM:Cat#Over-categorization. Regards, --Burkhard (talk) 00:29, 17 February 2011 (UTC)

The image shows both use of PPE, and mis-use of PPE. Owing to lack of better images, it's widely in use on WP as an illustration of PPE use. However it's not a good use of PPE - the visor is left open, hence the PPE (lack of) categorization.

Taxonomy n the real world is rarely a matter of simple exclusive tree-structures, where membership of a parent is wholly implied by membership of a child. Andy Dingley (talk) 10:37, 17 February 2011 (UTC)

Well, you are free to categorize any image in your real world at your discretion, but Commons - like any other wiki project - has a clear cut policy on avoiding over-categorization for good reasons. Your ad hoc arguments from above just show that you either haven't understood or that you don't care about the rule and the reasoning behind it. Why don't you try to understand it first before going onto some edit war? Especially the chapter Why is over-categorization a problem is worth considering. --Burkhard (talk) 21:14, 17 February 2011 (UTC)

:)

It makes me giggle that you made Category:PPE (lack of) and it's either me or my stepdad making up a majority of the pictures. Teehee. Also thanks for all your hard work! Lewis Collard! (lol, internet) 00:09, 18 February 2011 (UTC)

Well I'm not showing you my angle grinder injuries! (suffered on a very similar Land Rover). Andy Dingley (talk) 00:24, 18 February 2011 (UTC)