Commons:Graphics village pump/August 2020

GIF current version doesn't loop properly

I've tried to upload new versions of the GIF at File:Intersecting_ellipsoids.gif, but no matter what I do, the current version does not loop indefinitely as it should. As you can see, I've reverted it to previous versions which loop properly in the history, but the current version does not. Any advice would be appreciated. --Student298 (talk) 23:19, 9 August 2020 (UTC)

Extraneous text in SVG; can't be seen/deleted in Inkscape

So I translated File:ܬܪܥܣܪ ܫܒ̈ܛܐ ܕܐܝܣܪܐܝܠ.svg from the English version. All the translated text was changed to 'path' and everything looked good but, when I uploaded it, there is some extra Syriac text--not an image/path--that I mistakenly put in there somehow (overlapping where 'Judah' is on the English version). This extra text only appears in browsers; it does not appear in Inkscape so I can't click on it and delete it. Any ideas as to how to fix it? --334a (talk) 07:42, 14 August 2020 (UTC)

I have plenty of old scientific books (post 1830s) filled with engravings I can scan in PNG. Anyone up for a conversion collab so we can pour a lot of unique free SVGs on commons?

Hey, I'm not sure if this is the right place but I am not a wikicommoner, at most I've fixed a couple articles on EN and IT wiki and I don't know very well how all of this works.

Thing is, I'm a chemist with a bunch of old/antique books (1830s onwards), and they are filled with beautiful engravings, mostly related to chemistry, biology, animals, plants, scientific equipment... Now, old engravings and illustrations can be found in high res or SVGs on stock websites for a substantial price even though they are public domain material created by unknown people (very few are signed, most are anonymous and some were commonly recycled and cross-used by editors). As I dislike most paywalls, especially those that claim ownership of material they do not own, I'm looking for someone willing to convert engravings such as those on my personal page here (User:Valeg96) into SVGs. Should anyone be interested, I can get to work and provide them lots of scans, and with some clarification, add the proper metadata and make them available for free to everyone. Should anyone want to also upload the SVGs and just mention me, it's ok; I am not doing this to accumulate a portfolio or something, I just want to provide some material if anyone is interested.

I apologise if any of this (including my profile page) violates any WikiCommons guideline; let me know if anything needs to be fixed. --Valeg96 (talk) 16:35, 22 August 2020 (UTC)

• 
  Steam locomotive (fornari 1903) PNG
• 
  Steam locomotive (fornari 1903) SVG

User:Fæ should know. Alexpl (talk) 13:29, 23 August 2020 (UTC)

Thank you. I just wrote on their talk page. --Valeg96 (talk) 14:58, 23 August 2020 (UTC)

I'm flattered, but I know very little about SVG creation. It's possible to automate this, possibly with acceptable results but not 'smart' results, so large files but little volunteer time needed per file.

Using the free potrace from a command line, this created the above SVG version. It's a large file, but looks okay when you zoom in.

So, I guess all you need to do is get on with uploading the collection of scans, and later on someone with a bit of programming know-how can batch process them to create SVGs. There's no harm in retaining the PNGs, as the SVG conversion is not a "lossless" process. --Fæ (talk) 16:02, 23 August 2020 (UTC)

Why should we trace PNG of high resolution and store SVGs? Is there a need to do so? -- MaxxL - talk 16:18, 23 August 2020 (UTC)

Thanks a lot User:Fæ! Even though I definitely won't have the time to also convert them with that tool (It's hard enough for me to get around this platform, talks, edits and whatever) and I'd need way more time to learn how to use it, it's kinda an amazing find. User:MaxxL from what little experience I've had so far, PNG images with that much detail are absolutely awful when inserted in a printed document (especially since you need to resize them), while SVGs (or eternal objects, like ChemDraw structures) keep a decent look; at least for the work I've had to do in the past, PNG or GIF scanned engravings looked awful when you actually print or export the PDFs. Then again, this is not my field and it could also be caused by something else. On a side note, what is the proper resolution I should use when scanning them? I can go from 100 to 1200 dpi, and I'm usually around 400 or 600. And do you think the metadata/info on the few I've uploaded so far are correct? Again, I'd rather do the work decently once, rather than have people correct mistakes later, so that's why I'm asking. Thank you all!--Valeg96 (talk) 16:37, 23 August 2020 (UTC)

Suggest going for highest resolution scans, even if any extra information is slightly arbitrary. Keep in mind that Commons routinely copes with 500MB files and is never going to run out of space, plus images are displayed in easy to handle sizes, so the full size is not forced on anyone. The info is fine, but if you can find it, the publication year or even OCLC ref, like OCLC 797489580 is very helpful for others to discover and add more metadata later. For finding OCLCs, worldcat.org is your friend. If alternative crops or SVG versions are created, the image pages should cross-link to each other, preferably using the <gallery> tag. --Fæ (talk) 18:59, 23 August 2020 (UTC)

@Valeg96 and Fæ: You may also find dFX helpful for the uploading and crosslinking.   — Jeff G. ツ _{please ping or talk to me} 16:58, 24 August 2020 (UTC)

Thank you both @Jeff G. and Fæ: . Another question: who should I add as author? The person who scanned the images (me) or the author of the book? Since the actual artwork was done by neither of us, what's the best option? For the sample ones I uploaded already, I added my name.--Valeg96 (talk) 17:07, 24 August 2020 (UTC)

As the intent is to display the illustration, the 'author' is the illustrator, which may not be the book author. In the alternative template {{Book}}, there are fields to add book authors, illustrators, engravers, publishers etc. In the standard information template any or all of these might be added to the author field. You can add your account name as the scanner, but it's not that important and you are automatically in the file history as the uploader and making scans does not introduce any new copyright that needs mentioning. If in doubt have a look at other examples (COM:IA books is a very big current book project) and see if you like the way they are laid out and if you fancy setting up similar levels of information. There are good practices, but no hard rules apart from ensuring there's sufficient to understand copyright and sourcing.

If you want to add a tracking category to see the progress of your uploads, you can create something like Category:Scans uploaded by Valeg96 as a maintenance category. I tend to create tracking categories like this for most of my projects. --Fæ (talk) 19:53, 24 August 2020 (UTC)

Colours missing when rendered

filters removed

An odd one, and I thought I was good at this. I created, using Inkscape: Aksai Chin Sino-Indian border map.svg

It was based on an existing map (Kashmir location map.svg) and I just drew additional lines, text etc (taken from a PNG version). In the first upload the text did not appear, so I reduced the number of layers. Now all the layers and objects appear but the lakes appear white (they should be stroke #0978abff and fill #c6ecffff). Double-click for the 'Upload' page and it appears as it should, but when rendered on the file detail page or in Wikipedia and the lakes are white. Has anyone come across this before? Hogweard (talk) 19:07, 18 July 2020 (UTC)

Hi @Hogweard: I'm unsure why two SVG filters which didn't seem to do anything useful were applied to these elements. After removing them, the lakes appear blue. Did you create these filters?

By the way, I've noticed that the thumbnail renderer sometimes makes lighting filter effects lighter e.g. desert and water patches on File:Catan_Universe_fixed_setup.svg and turbulence filters darker e.g. other patches on File:Catan_Universe_fixed_setup.svg compared to Web browsers, so if you need the filters, you'll have to compensate for this. Cheers, cmɢʟee ⋅τaʟκ 00:52, 19 July 2020 (UTC)

That's brilliant - thanks. I didn't knowingly add a filter and I have not come across filters, so I will try to learn something about them and work out if it is in Inkscape or some deeper coding. It does make me wonder if it is a similar thing, in reverse, that went wrong with some changes I tried to make in BlankMap-World-162E.svg: there, some unwanted areas appeared that had been invisible, and the change that made them appear was to click 'ungroup'. Hogweard (talk) 10:06, 19 July 2020 (UTC)

 Done My pleasure! Here's a brief tutorial on SVG filters. Cheers, cmɢʟee ⋅τaʟκ 11:41, 19 July 2020 (UTC)

P.S. SVG filters unexpectedly created in Inkscape has also be reported on en:Wikipedia:SVG_help#Poor_rendering_of_image.

@JoKalliauer: , @Redrose64: , @Glrx: and @TilmannR: have you come across this issue? Thanks, cmɢʟee ⋅τaʟκ 07:01, 30 August 2020 (UTC)

Hatch fill lost in some PNG

Scanning tunneling microscope

Even though my SVG file passed "Commons:Commons SVG Checker" checks, and there are no problems with the SVG file in any of my browsers, the hatch fill in one or both rectangles disappears in some PNGs converted by wikimedia's buggy converter. Namely, resolutions: 320 × 201 pixels and 640 × 402 pixels look OK, 800 × 502 pixels is showing only the left rectangle, and 1,024 × 643 pixels and 1,280 × 804 pixels none. I shouln't be asking, but until there is a decent converter (command line inkscape converter for inkscape svg files, perhaps?), which features do I need to avoid so I don't waste time on trying to fix these things and administrators' time on deleting bad uploads? I've learned that blurred objects get strange cut-offs, that scaling cloned objects results in a disaster, text (especially math formulas) should better be converted to paths... Ponor (talk) 09:34, 29 August 2020 (UTC)