This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
Hi. Thanks for noticing the errors.
After seeing the author's call for help at the Graphics Village Pump, I simply opened the file in a text editor to see what's wrong with it and edited it down to the current version with regex substitutions.
Of course regexes can be a bit tricky and the older version of the file is so slow to render that I didn't check for deviations as thoroughly as I should have.
Using a WYSIWYG editor like Inkscape didn't seem appropriate due to the size and complexity of the image
and I'm new to the whole SVG editing thing, so I haven't made or found any good tools yet.
Regarding the size of the circles: What are the guidelines for acceptable differences between versions of a file? Yesterday I fixed some problems with Extragalactic_Distance_Ladder.svg and while doing so removed a bunch of gradients and broken shadows to make the file more easily editable. If such changes aren't acceptable, I better revert and reupload as a new file.
TilmannR (talk) 06:18, 19 May 2018 (UTC)
thx for reducing the size. Your result is great. Could you describe the detailes of you reducing procedure ? What text editor, what patterns, ... I would like to put it on the image page at the section : "postprocessing." Thx again --Adam majewski (talk) 17:23, 19 May 2018 (UTC)
I'm glad you liked my edit. I'm new to Commons and getting this kind of feedback is quite encouraging. (The text editor I used was Notepad++[1], but that's not particularly relevant.) I don't remember each individual pattern I used, but the main culprit definitely were lines like <use xlink:href='#gpPt6' transform='translate(600.3,998.2) scale(0.45)' color='rgb(190, 190, 190)'/>. There are 12141 of those, each about 100 characters long, so approximately 1.2 MB in total. A short regex for that is e.g. <use[^>]*190, 190, 190\)'/>, but the real solution to superfluous elements is to not generate them at all. Apparently it's possible to avoid generating points by setting their point_type to -1. If I was more familiar with Maxima, I'd fix the script, but I'm not, so I won't. TilmannR (talk) 21:03, 19 May 2018 (UTC)
In general if I would optimise a SVG I would not accept such visiual differences, as in your edit.
Generally editing files not by the original author should have a better reason, than just minor file reduction, or removing unproblematic errors.
I think Empty bucket.svg should show on purpose the difference between Tool-made-svg and Text-editor-made-svg (compare with Full bucket.svg).
I think it is case dependent, and there are no guidelines for optimisation, therefore everyone has different onions on it. Generally SVG simplification techniques and SVG files to be drawn by hand should be used to have easily editable files.
I think the examples might not have been the best, maybe File:Dojikko2.3.svg is a better example of what might get reverted (see Perhelions's edits). (My edits did some visible changes on purpose, because the file was corrupt and some elements didn't render before.) Previews above 600px might won't work in all previous versions.
Generally if you are unsure: Sarang has more experience than I have, maybe ask Sarang. But as said before you did a great work. I just wanted to explain why did't improve more in my edit. JoKalliauer (talk) 14:37, 20 May 2018 (UTC)
regex
Thx for describing how you have used regex.Can you describe what was the idea ?
If I understand correctly :
finding the most frequent pattern : "There are 12141 of those, each about 100 characters long, so approximately 1.2 MB in total"
and remove or group them : "real solution to superfluous elements is to not generate them at all."
"Can you describe what was the idea ?" What exactly do you want to know? Could you please rephrase the question.
"Is it general method for all svg ?" No, it's not useful for all SVGs. Removing elements only makes sense, when those elements don't contribute to the image. For example: File:Dashed_line_p6.svg is unnecessarily large. In fact it could be more than 200 times smaller! None of its elements is inherently superfluous though, so deleting them would destroy the image, but the dashed strokes are encoded inefficiently. Such a problem cannot be solved with regexes, it cannot be solved with simple optimizers. It's a problem that requires a certain kind of intelligence.
I don't think I have anything to add to the "Reducing the file size" section, but personally I wouldn't recommend the removal of code formatting. Line breaks and indentation make SVGs much more readable and they usually have a negligible impact on file size and rendering time.
@Watchduck: I had uploaded all stellations of the icosahedron and exactly one of them happens to also be a Kepler–Poinsot polyhedron. I just uploaded the dodecahedron stellations (and as a bonus the stellated octahedron). Thanks for the suggestion. TilmannR (talk) 23:51, 9 September 2018 (UTC)
@Fma12: @Yann: Hi! "USD Bitonto Calcio.svg" was originally uploaded by AltHis_En (Log: [2]). I merely fixed a rendering issue and validation errors. Marking me as a copyright violator for this seems unfair. I don't remember the file having a deletion request or missing a license. (I cannot view the info of a deleted file. Maybe I overlooked something? If I did, please tell me, so I can avoid future mistakes.) Being held responsible for somebody else's copyright violation would severely discourage me from fixing broken files in the future. (Honestly, I probably wouldn't complain, if I wasn't so proud of keeping a spotless record throughout all my work for the Illustration Workshop. And thank you very much for doing all this administrative work. ) TilmannR (talk) 21:26, 20 September 2018 (UTC)
I don't have as much time for illustration requests as I used to.
Your requests tend to take a lot of time and effort.
I have already accepted more request from you than from any other user.
It is quite unfortunate that the other graphists seem to be disinterested in your requests, but there's not much I can do about that. TilmannR (talk) 10:26, 24 October 2018 (UTC)
Please note that the copyright of the design of a coat of arms and the copyright of the specific image that depicts a coat of arms are different things. The design is often non-copyrightable, but the image might be. In this case the original file had the cc-by-sa-3.0 license. You should honor that by naming Pianist as one of the authors and adding the appropriate license tag. TilmannR (talk) 16:29, 26 October 2018 (UTC)
@Ronhjones: Wildpark Müden has no contributions other than the upload. They didn't even add their logo to the relevant article. I predict that they won't fix the source & author and the image will be deleted.
Would you recommend that I continue to fix broken SVGs like this one or should I leave them alone? Getting my talk page flooded with warnings isn't fun. TilmannR (talk) 03:24, 10 November 2018 (UTC)
I can only suggest to see if you think the image is valid enough to stay. An obvious logo with an "own work" is probably doomed. If it's not an original file that you upload, I would remove any warning with an edit summery like "original file not my upload". It's the script that is used for the placing of banners on the file page, also does the warning automatically. Ronhjones (Talk)11:28, 10 November 2018 (UTC)
I wanted to highlight the difference between embedded-PNG and plain-PNG: embedded for PNG-24, PNG-32, PNG-48, PNG-64 and plain for PNG-8 (palette variant: 2-256 colors). Correct me if I'm wrong. This clarification for future image uploads. Waiting for your reply. Hello!--Grasso Luigi (talk) 10:56, 16 November 2018 (UTC)
@Jeromi Mikhael: It's great to see how much progress you've made. I replaced the sun with a circle and filled the hands with white. I used a clip path to make sure that everything effortlessly fits into the red circle. TilmannR (talk) 15:26, 26 November 2018 (UTC)
Thank you for your minimum SVG vectorial. I had a question: why does the black rectangle appear in the middle of the files (pure vectorial but not as minimal) uploaded by me? Is it necessary to change which part of the source SVG to avoid it? Waiting for your reply. --Grasso Luigi (talk) 09:17, 5 December 2018 (UTC)
@TilmannR: nein, ich hab aber einige fragen - von welcher wikipedia kommst du? von der deutschen? wieso hast du dort weder benutzer- noch diskussionsseite?
ich habe mehr durch zufall dateien von dir gefunden:
beide sind gut gezeichnet und mich würde interessieren mit welchem programm du sie gezeichnet hast?
die wichtigste frage ist aber, wieso diese beiden (und vermutlich noch andere) von dir erstellten dateien in den artikeln nicht ersetzt sind? nicht mal in den deutschen artikeln
@Mrmw: Ich komme tatsächlich von keiner Wikipedia. An meinen Statistiken lässt sich leicht ablesen, dass ich fast ausschließlich auf den Commons tätig bin. Ich habe Freude am Erstellen von Vektorgrafiken, nicht so viel Freude am Schreiben von Artikeln. Ich habe keine Diskussionsseite im Wiki, weil dort noch niemand mit mir diskutiert hat. Ich habe keine Benutzerseite, weil ich nicht glaube, dass ich eine brauche. (Und so ein Redlink in der Signatur sieht doch auch schick aus. :) ) Meine Grafiken sind üblicherweise in Inkscape erstellt, mit SVGOMG vereinfacht und im Texteditor per Hand weiter optimiert. Große Ausnahme sind Overwrites von kaputten SVGs, die sich in wenigen Sekunden allein im Texteditor fixen lassen. Meine Dateien sind nicht immer verwendet, weil ich nicht aufdringlich erscheinen möchte, und weil das Verwenden von Grafiken (im Gegensatz zum Erstellen von Grafiken) etwas ist, was die Autoren auch selbst hinbekommen, falls sie meine Grafiken für hinreichend hochwertig erachten. (Dafür gebe ich mir auch Mühe mit der Kategorisierung und dem Setzen von {{Vva}}-Templates.) TilmannR (talk) 14:41, 22 December 2018 (UTC)
vva ist mir geläufig - aber ist es nicht so, dass ein autor von sich aus hinter die grafik sehen muss (kontrollieren) ob eine vektor-grafik verfügbar ist? verstehe mich nicht falsch, ich finde es nur schade, dass deine arbeit, die ich gut finde, nicht gleich zur anwendung kommt - sowas ist doch schade
ich selbst benutze auch inkscape und habe zum teil auch schon mit svgomg vereinfacht - nur bin ich dabei recht unsicher, welche vereinfachungen am sinnvollsten zu verwenden sind
zum teil waren im w3c-check nach der vereinfachung fehler im test - wie testest du deine svgs bevor du sie hochlädst? es gibt wohl auch fehler/warnungen die man ignorieren kann?
gern wüsste ich auch mehr darüber wie man per hand am sinnvollsten die svgs editiert und auf fehlersuche geht - wie du eben schreibst im texteditor - hast du dazu literatur oder tutorials im netz die du empfehlen kannst? wie hast du dir das wissen angeeignet? try and error?
Ja, Autoren werden nicht automatisch über Vektorversionen informiert, wenn sie nicht gerade die Rastergrafik in ihrer Watchlist haben.
Du kannst meine Bilder gerne in Artikel einfügen, wenn dir das wichtig ist.
Zu SVGOMG: Die Vereinfachungen, die die Dateigröße am meisten beeinflussen ohne das Aussehen des Bildes zu beeinflussen, sind am wichtigsten. Ich schalte immer "Compare gzipped" aus um die tatsächliche Dateigröße zu sehen und schalte "Prettify markup" und "Sort attrs" ein, damit die Datei im Texteditor lesbarer wird. Ansonsten sind die Standardeinstellungen schon ganz in Ordnung.
Zum W3C Validator: SVGOMG lässt Dinge, die es nicht erkennt, einfach in der Datei um nichts kaputt zu machen. Dazu gehören dann manchmal auch Dinge, die der W3C Validator nicht erkennt.
Inzwischen checke ich meine Dateien kaum noch mit dem W3C Validator. Wenn man sich lange und intensiv mit SVG auseinandersetzt, ist relativ klar was "erlaubt" ist und was nicht.
Letztendlich sind alle Validationsfehler irrelevant, solange man sich sicher ist, dass das SVG trotzdem richtig gerendert wird. Wenn man sich nicht sicher ist (oder aus Ehrgeiz null Fehler möchte) sind natürlich alle Validationsfehler wichtig, aber alle Warnungen trotzdem irrelevant.
Wie man SVGs am sinnvollsten per Hand editiert: Sorge erst dafür, dass die Datei klein und lesbar wird (also z.B. SVGOMG mit "Prettify markup" verwenden) und nutze dann dein Verständnis über den Aufbau von SVGs um zu erreichen was du erreichen möchtest. Mehr habe ich dazu eigentlich nicht zu sagen. Um mehr über SVGs zu lernen, einfach googlen. MDN ist eine recht zuverlässige Quelle. w3schools geht auch. Und auf jeden Fall Help:SVG und Wikipedia:SVG_help durchlesen, denn manchmal werden auch korrekte SVGs nicht korrekt gerendert.
@Mrmw: Es gibt Fehler und Warnungen die man absichtlich drinnen lasst. Beispielsweise: Ein Hilfsgitter oder Hilfslinien sind laut W3C-SVG1.1-DTD invalid. Sie zu entfernen verändert zwar nicht das Rendering, aber der nächste der bearbeitet hat diese Elemente nicht mehr, daher sollte man das mMn drinnen lassen. User:Glrx spricht sich auf User_talk:JoKalliauer#Re:_if_you_scrubb_a_svg_please_make_the_file_valid_on_your_own dafür aus, dass aria-label (Wenn ein Text zu Flächen konvertiert wird, dass der Textbuchstaben noch als Text vorhanden sind), und data-name (Bennenung von Elementen, dass man es wiederfindet) erhalten bleiben, auch wenn beide zu Fehlern führen.
Ich bin oft zu perfektionistisch und lösche zu viele Fehler.
@TilmannR: Es gibt auch wichtige Warnungen, beispielweise wenn man eine Datei ohne DoctypeDefinition (DTD) schreibt kommt die Meldung: Consider adding a lang attribute to the html start tag to declare the language of this document., wenn man DTD gemäß User:Quibik/Cleaning_up_SVG_files_manually hinzufügt, dann wird die Datei iA. "strenger" geprüft, hier wird auch darauf hingewießen, wenn type="text/css fehlt.
I saw your edit on the Australian 2016 map, first off, thank you so much for fixing that, so quickly after I uploaded it as well! I've been having issues with saving my files as SVG ever since I got a new laptop, and its causing corruption issues and black boxes in some of my older maps.
How did you go and fix the issue exactly? I'd be curious to know so I can prevent further situations!— Preceding unsigned comment added by DrRandomFactor (talk • contribs)
@DrRandomFactor: Hi. A lot of the broken SVGs on Commons have the same problem: The program, which turns the SVGs into PNG thumbnails, ignores all <style> tags, which don't have the type="text/css" attribute. Technically that attribute isn't necessary. It's just one of the quirks of the particular version of librsvg used by Wikimedia. You can fix the problem by opening the SVG in a text editor and adding the attribute manually. The SVG editing program you're using probably has a setting to save SVGs entirely without CSS, which would also solve the problem. You can learn more about various SVG rendering problems by reading Help:SVG and Wikipedia:SVG_help. (Please remember to sign your posts with ~~~~.) TilmannR (talk) 11:57, 29 December 2018 (UTC)
@DrRandomFactor: TilmannR is correct, but this issue can be detected without understanding SVG-code:
Line 3, Column 11: required attribute "type" not specified ... Typical values for type are type="text/css" for <style> and type="text/javascript" for <script>.
ERROR in <style>: It will not be rendered properly by Wikimedia's SVG renderer. As workaround add attribute type="text/css" to <style>. See https://phabricator.wikimedia.org/T68672 for details.
@JoKalliauer: That's interesting. Based on Help:SVG#cite_note-2 I had assumed that a missing type attribute wasn't a validation error, because it MUST default to "text/css" anyway. Recommending the Validator and Checker to newcomers is definitely a good idea. Thanks for reminding me. TilmannR (talk) 19:43, 29 December 2018 (UTC)
@TilmannR: You are right, that's why I said you are correct, and I know thats a (in my opinion helpful) Bug of Validator: Accoring to https://www.w3.org/TR/SVG11/styling.html#StyleElementTypeAttribute (which is as far as I know the official guideline): If a ‘type’ is not provided, the value of ‘contentStyleType’ on the ‘svg’ element shall be used, which in turn defaults to "text/css". Therfore not specifying should be allowed, but maybe because it might lead to problems (like Rendering with librsvg), they gave out an Error, with a wrong explantation, that beginners know what to change. But if you check the same file without &doctype=SVG+1.1: W3C takes the result of validator.nu and clames it as its own, which does not mention anything (related to style): https://validator.w3.org/check?uri=https ... (Simple_Map).svg
User:Sarang told me on User_talk:Sarang/Archive/2017#Autor-Feld: Im Author= sollten tatsächlich nur Autorangaben stehen, alles andere stört bei der maschinellen Auswertung; das Auswerteprogramm verkraftet die verschiedenen Schreibweisen.
specifying who did what should not be used for example:
I don't know why it should be computerreadable, maybe there are bots to find possible copyright-violation depending on the author, or for statistiks.
@Sarang: Can you explain why the autorfield should be "maschinenlesbar"?
@JoKalliauer: I was using something like *derivative work: [[User:TilmannR]], because that's what derivativeFX generates, but apparently that tool is obsolete. {{AutVec}} is an easy case, once you've been made aware of its existence. (Is there an Introduction to Templates or Template Overview page somewhere? E.g. I only found out about FakeSVG this Thursday, because you used it in the Illustration Workshop. That's a rather inefficient way to learn.) It'd be nice if there was a similar template for arbitrarily many authors with arbitrary roles.
I'm not a fan of the summary of Poutre definitions de.svg. The authorship information is all over the place: You're the "author", Bbanerje and Cdang are "sources", you and Bbanerje made "modifications" ("Changed french to english", "Changed english to german"). It would be very difficult to write a program, which understands all of this. Therefore the authorship information is arguably not machine readable. Is it really correct to not list Bbanerje and Cdang under "author"? I've been searching through the guidelines and policies and didn't find much about this issue. In theory you could add the names of other authors as a parameter to the license template (e.g. {{Cc-by-sa-4.0|Example Name}}) or you could add a {{Credit line}} to the image information, neither of which appear to be commonly used. Is there no official recommendation this? TilmannR (talk) 03:14, 30 December 2018 (UTC)
I now changed the Autor according to scientific papers: I named all authors of them starting from the most important author to the last important author, that should be botreadable.
I don't think there is a guideline, since everyone uses as they like and it not a copyright-violation, and therefore it is "no problem", to make it differently, also different Wikipedians have a different point of view.
As somebody already asked earlier #Sprache, naming the language skills helps others to contact you in the correct language. I personally would prefer Esperanto per default on Commons, also I cannot speak a Esperanto at all, since commons is a multilingual project, and should be fair between languages. (Just my personal (idealistic) point of view.)
I think Commons:Guide_to_adminship#Initial_setup (de) might be a good guideline also for not administrators, which recommends to use language- babels (you could also use them on your talk-page). There are many Admins without babel. (And I think it is also there right to keep the language-skills private.)
Since you are (one of) the most active User on Commons:Graphic_Lab/Illustration_workshop, this might be helpful for other. (But don't feel pushed to do anything, just a side remark, because I don't know if you prefer English/German.)
@JoKalliauer: That's the main problem with constructed languages: Nobody knows how to use them. And if any constructed language was used as much as a natural language, it would probably evolve, shift and adapt like a natural language, so you'd end up with multiple natural dialects of your constructed language.
I'm aware of Babel, but since I prefer to use English, and since this preference should be obvious to anyone, who ever read anything I've written on Commons (except for that one section in German), I don't think I'd get anything out of using Babel. I also don't like the idea of putting a numeric label on my language proficiency. In every language I know there are things that are easy to talk about and things that are difficult to express. Distinguishing between a native language and a native-level second language is just silly. TilmannR (talk) 03:39, 30 December 2018 (UTC)
PS:You know the librsvg-bugs very well, and you are a good drawer. (Thanks for all your works they are great.) In Category:Pictures_showing_a_librsvg_bug_(unsolved) are several files I cannot get fixed, maybe you want to take a look at it.
You seem to be very engaged, therefore I just want to inform you about things in you area of interest that should be done.
@JoKalliauer: I'm aware of the other workshops. Is there a way to receive notifications for new requests? Checking all of those sites, only to find that nobody requested anything, is mildly annoying. Am I really supposed to fix those files? Aren't they needed to demonstrate the bugs? Or is there a different category for the broken SVGs that aren't supposed to get fixed? I vaguely remember something like that. TilmannR (talk) 04:04, 30 December 2018 (UTC)
how to recive Mails when new request are comming
You could add the related sites to your watchlist and change the Preferences in de:Special:Preferences and en:Special:Preferences, that you recive an E-Mail when the site get changed after your last visit.
I know it is not the best solution.
I "can't" use it in de.wikipedia since I am watching to many pages, therfore I use the Watchlist (Beobachtungsliste).
Files in Pictures showing a librsvg bug (unsolved) are SVGs without workaround (only PNG or SVGs with reduced features). Finding a workaround will be helpful for other pictures that have the same bug. If the librsvg is needed (f.e. Pictures demonstrating a librsvg bug) you should upload to a new file, otherwise you can overwrite it.
Hi TilmannR, thanks a lot for your attempted fix in this file.
My end goal is a gallery of the situations in 2600 BCE, 2500 BCE, ... 400 BCE, 300 BCE; according to Middle Chronology - but for that, I need a master file. This here is it, a container with 24 maps in one.
However, it looks like the layering was seriously disturbed in your fixing attempt: I have each century in its own layer, 18th century BCE was the one that was currently put on visible. After your edit, all elements of the visible layers were in a group, per layer; all elements of the invisible layers where gone; there was no evidence of any layer structure left. So while the fix worked on the surface; it made things worse.
What you call "flowed text" is really disturbing and I'd like to edit it out of the file eventually. I have tried several approaches, but it seems once I put a text beyond a font-size of ~20, the Commons-Display turns it into an ugly box very unlike the Inkscape view. Loaded directly in my browsers just doesn't display these same texts at all. So, what's the trick behind that and how do I achieve "regular text" instead of "flowed text"? I'd appreciate your hint so I can apply it there. --Enyavar (talk) 14:13, 10 January 2019 (UTC)
@Enyavar: Sorry for breaking your file. I should have been more careful during optimization. In Inkscape's "Text" menu there's a "Convert to Text" option, which converts flowed text to regular text. TilmannR (talk) 15:02, 10 January 2019 (UTC)
Thank you for your helpful tips; I converted the texts following your hint and now they look fine! Also, don't be sorry for nothing - there was no harm done. I have kept the local files; the upload is mostly to have a version history if I manage to break or delete my offline files. Also, my file description might have suggested that this is already a completed work.
Once I produced a final version, I think it might be prudent to delete the working-versions of the master file. Do you agree? --Enyavar (talk) 20:47, 10 January 2019 (UTC)
@Enyavar: I have no strong opinion on the matter, but let me play devil's advocate:
The amount of storage taken up by the files is negligible in the grand scheme of things.
The works-in-progress document how your images were made, which is nice.
The time that an administrator would take to read and execute your deletion request could be spent on deleting harmful files instead.
Tilmann, you put a FakeSVG notice on my file about the "Supreme Isosceles Trigon". I was having trouble keeping the text aligned with the geometric construction so, in Inkscape, I converted the image into a bitmap (Inkscape has a command which does that). The conversion seems to have solved the text alignment problem. Should I have used some other solution for the text alignment problem? Scott Gregory Beach (talk) 14:19, 16 January 2019 (UTC)
ERROR in <tspan>: Text element found with list of coordinates. It will not be rendered properly by Wikimedia's SVG renderer. See https://phabricator.wikimedia.org/T35245 for details.
This means <tspan x="2914 3100">ab</tspan> should be replaced with <tspan x="2914">ab</tspan> or with <tspan x="2914">a</tspan><tspan x="3100">b</tspan>
Thank you for fixing the 2003 Scotland Election map I uploaded today! I also uploaded another one for 1999, and that one uploaded with the colours intact but it messed up the spacing on the ridings and the map looks wonky now as a result. Would it be possible for you to take one of the earlier uploads and fix it?
Also, how exactly does one go about fixing these? Illustrator always makes my maps look like this when I upload and its annoying! Is there a step-by-step on how to fix this? I know there is code involved, but Im garbage at that!
@DrRandomFactor: Hi. Simply open the SVG in a text editor, look for <style> (it's usually near the beginning of the file) and replace it with <style type="text/css">. You can read more about SVG rendering quirks on the SVG Help page. I believe Illustrator has an option for saving SVGs without CSS, which would also fix the problem. You can check whether your SVG will render correctly by using the Commons SVG Checker.
Thank-you for File:Strouhal_de_corps_2D,_Blevins_et_autres.svg
thanks, dear TilmannR, for having changed my file so that the texts appear. I wonder what mistake I made in Inkscape (with which I'm starting) ... Regards, Bernard de Go Mars (talk) 20:52, 27 January 2019 (UTC)
Hello TilmannR! Thank you for this tip! Obviously, after converting to text, you have to get the text to put it back in place (especially for the vertical text that disappears from the screen). Thanks again for the link! Another thing I'm wondering is how to make a multilanguage svg image (like we can place multilanguage notes with annotator). That's my goal, but I do not know if it's possible ... Regards, Bernard de Go Mars (talk) 10:32, 28 January 2019 (UTC)
@Bernard de Go Mars: You can make a multilingual SVG by using the <switch> tag.
I don't know an easy way to create switch tags with Inkscape. You might have to create them by editing the SVG in a text editor. TilmannR (talk) 11:43, 28 January 2019 (UTC)
Thank you for your answer. I did some tests with the Switch; but in fact I do not know where to place the text between switches; according to https://tools.wmflabs.org/svgcheck/index.php, there are a lot of errors and nothing is displayed ... We must understand that we, non-computer scientists, are disabled people ... Sincerely, Bernard de Go Mars (talk) 14:17, 28 January 2019 (UTC)
As you can see, you can place text elements with systemLanguage attributes anywhere between <switch> and </switch>. If you upload your image, I can tell you exactly why it doesn't get displayed correctly. TilmannR (talk) 14:38, 28 January 2019 (UTC)
@Bernard de Go Mars: The problem is in line 74 to line 86: There are a <text> element and two <tspan> elements, which never terminate. Every <text> must have a matching </text> and every <tspan> a matching </tspan>, just like any opening parenthesis in a mathematical expression needs a matching closing parenthesis.
Great. I'm reviewing my code and I'm translating! (I do not know where the lack of parity of <text> and <tspan> came from) (but I tinker so much that it is not surprising ... See you later ... Regards, Bernard de Go Mars (talk) 20:52, 28 January 2019 (UTC)
Hello, TilmannR! Lost in my coding prowess, I did not notice that you had improved the design of my sections. This shows that you are a master of SVG and the result is much better. But there is a small problem with the green "profile" of the top. His curious leading edge did not come from a drawing error on my part. You will see in the file http://perso.numericable.fr/gomars2/aero/strouhal_sections_non_circ_blevins1999.png that this form (of aspect ratio 5) is indeed such that I had drawn it. I have no more information on this section than this drawing. As this graph is meant to be universal, I hope you can correct the shape of this "profile" (the others are very good) Note that there is no urgency ... Regards, and thanking you again for your nice contributions, Bernard de Go Mars (talk) 13:53, 29 January 2019 (UTC)
Hi Tilmann. I was wondering can you vectorize this emblem, until i found a modern graphic representation. I was hoping that you could made that, but with several adjustions (e.g. the shield and the wheat ears should be wider, the ribbon should be white). Inscription : ПРОЛЕТАРИИ ВСЕХ СТРАН, СОЕДИНЯЙТЕСЬ!, ПРОЛЕТАРІ ВСІХ КРАЇН, ЄДНАЙТЕСЯ! Thanks.--Jeromi Mikhael (talk) 01:08, 2 February 2019 (UTC)
I tried to dash off a quick svg with Inkscape, failed to fix the flowed test, and noticed only once I'd uploaded it that it was a mess. I was going to upload a fixed version today, only to discover you'd beaten me to it, not ten hours after the original upload. It's things like this that make editing collaboratively a pleasure. Thank you very much! HLHJ (talk) 03:08, 2 February 2019 (UTC)
@Jü: I don't use Mac OS, but apparently there's a Mac version of Inkscape: Just select the background rectangle and delete it. If it's part of a group, you can double click the group to step into it. (Assuming that Mac has double clicks. I dunno. Use Google, if necessary.) You might need to step into groups multiple times, if the background is in a group in a group in a group.
If you don't consider installing a WYSIWYG SVG editor a "simple method": SVGs are just XML files. You can open them in your favorite text editor, figure out which node corresponds to the background, and delete it. If you want to change a lot of files, you can write a regex, which matches everything that looks like a background, and replace those matches with nothing. Such a regex might look like <[^>]*fill:(#ffffff|rgb\(100%,100%,100%\))[^>]*>. It just means "look for a < followed by arbitrarily many characters that aren't > followed by fill: followed by either #ffffff or rgb(100%,100%,100%) followed by arbitrarily many characters that aren't > followed by >". Note that this assumes that all white objects belong to the background, have self-terminating tags, and have their color specified in a style attribute as opposed to a fill attribute. In the case of Wikipedia:Grafikwerkstatt#Hintergrund_bei_SVG-Strukturformeln_entfernen all these assumptions were true. TilmannR (talk) 08:46, 9 February 2019 (UTC)
If you created this file, please note that the fact that it has been proposed for deletion does not necessarily mean that we do not value your kind contribution. It simply means that one person believes that there is some specific problem with it, such as a copyright issue. Please see Commons:But it's my own work! for a guide on how to address these issues.
Please remember to respond to and – if appropriate – contradict the arguments supporting deletion. Arguments which focus on the nominator will not affect the result of the nomination. Thank you!
Thanks for uploading that SVG of the LA bike share logo. I'm still pretty new to graphics editing - I tried using an online SVG converter to convert the original PNG file of the logo to an SVG, but that didn't seem to work. Thanks again for your help, and hopefully I can learn how to create SVG files on my own in the future. ECTran71 (talk) 07:49, 26 October 2018 (UTC)
You're welcome. I recommend starting off by familiarizing yourself with your tools. Open up Inkscape (or whichever program you're planning to use) and go through every single button and menu you can see. Just experiment and try to understand what they are there for. You should also consider using optimizers like SVGOMG to reduce the file size and clean up some of the superfluous junk that SVG editors tend to generate. Once your SVG file is optimized, it becomes much much easier to edit it "by hand" in a text editor, which is sometimes necessary to circumvent rendering errors or to make very precise changes that would be more time-consuming to do in a WYSIWYG editor. Additional info: Help:SVG. TilmannR (talk) 08:32, 26 October 2018 (UTC)
Logos are much more likely to be missing type="text/css" than other SVGs (presumably because they are just copied from their source instead of being created specifically for the Commons). So when I see one that is completely black, doesn't have "black" in its name, and looks like there should be some color in it, I view it directly in the browser. E.g. Laudamotion.svg is just a black rectangle with black "audamotion" text. What kind of company would use a black rectangle as it's logo? Sometimes you can see the hairline cracks between elements of different colors like in Dapp Life logo.svg.
In theory I could write a bot that automatically downloads all new SVG thumbnails, filters out the completely black ones, downloads their SVGs, searches for <style>, fixes and overwrites it, but I don't feel sufficiently motivated to do so. TilmannR (talk) 11:03, 3 March 2019 (UTC)
@JoKalliauer: Apparently &sort=last_edit_desc doesn't show most recent uploads, but most recent edits to a file's page. &sort=create_timestamp_desc might be what you're looking for. TilmannR (talk) 15:26, 3 March 2019 (UTC)
You maybe know Large SVG files, some can be compressed (f.e. File:E8out96g16.svg 72MB->5MB), but files with many <circle cann't (except compressing as svgz). There is a similar picture: File:1_42_polytope_7-cube.svg ( 483.840lines and 7.855 circle ) I merged all lines to one path, to reduce filesize, but librsvg seem to not like such long path-data and fails (see history)
I think this RTVE Testcard.svg should be rendered by librsvg and then the PNG should be traced to svg
@JoKalliauer: Ha ha! That's an excellent question. I had lots of fun with this particular file. :)
I used Inkscape's GUI, which unfortunately converts flowed text in such a way that the resulting regular text has to be left-aligned in order to be in the correct position. It's not really the correct position, when you take librsvg's text rendering quirks into consideration, but I didn't have the patience to fix that.
One of the flowRoots contains three different flowParas. I guess that that's where both Inkscape and your regex fail. If I remember correctly, I used Extensions > Text > Split text... to turn that into individual lines of text.
All in all I feel terrible about the job I did on that image, but at least it's presumably legible now. Of course I don't actually know whether the text is correct, because I don't speak മലയാളം. Thanks to STC19 we'll get to see plenty of "fun" SVGs over the next month. TilmannR (talk) 22:10, 28 February 2019 (UTC)
Thank you for the answer. Actually I had a different problem with Inkscape GUI: Converting changed the fontsize to ~12px (smaller). I now retried and the font-size was constant (just replaced as you said), therfore preprozessing with scour or with svgo, (most likely) leaded to the wrong behaviour in Inkscape (normally preprozessing with optimizer make Inkscapes behaviour better). — Johannes Kalliauer - Talk | Contributions07:42, 1 March 2019 (UTC)
@JoKalliauer: Map solar system.svg is a tough one. 225 texts, many of which are transformed, centered, multi-line flowed text. Fixing this by hand is possible, but way too tedious even for me. Maybe we need to write a more sophisticated script that actually calculates the correct positions? TilmannR (talk) 12:17, 3 March 2019 (UTC)
for text-anchor="start" therefore the text will be replaced for text-anchor="middle".
single <flowPara
inkscape ./${file}f.svg --verb=EditSelectAll --verb=ObjectFlowtextToText --verb=FileSave --verb=FileClose --verb=FileQuit works for all flowRoots that can be converted by Inkscape, but as soon as one flowRoot cannot be converted inkscape fails completely. https://bugs.launchpad.net/inkscape/+bug/1748725
@Level C: Yes. If you open that file in a text editor, you'll see that this "SVG" consists entirely of a JPEG image. It would have been better, if this image had been uploaded as an actual JPEG file instead. TilmannR (talk) 20:42, 10 March 2019 (UTC)
@Level C: As far as I know, being a fake SVG is not a sufficient reason for deletion. But if you upload a JPEG version of the file, the fake SVG can be speedy-deleted as an inferior duplicate. TilmannR (talk) 20:54, 10 March 2019 (UTC)
@Level C: There is no rule how to handle fake-Svgs and different admins will have different opinions on that.
I would distinglish between two cases:
If the file should be vectorized like File:AibonitoFlag.svg, some say you should keep this file and wait till someone vectorizes it properly. (There are different opinons on that.)
I'm with you (I follow). Today I was thinking I would try to learn more about .svg and 'vectorizing' and then reupload proper files. If I can't learn it then I'll delete those that can't be vectorized. Thanks.--Level C (talk) 22:03, 11 March 2019 (UTC)
If you created this file, please note that the fact that it has been proposed for deletion does not necessarily mean that we do not value your kind contribution. It simply means that one person believes that there is some specific problem with it, such as a copyright issue. Please see Commons:But it's my own work! for a guide on how to address these issues.
Please remember to respond to and – if appropriate – contradict the arguments supporting deletion. Arguments which focus on the nominator will not affect the result of the nomination. Thank you!
Thanks very much for designing the logos for FindingGLAMs, we've decided to go with number 7. Could you email me your address (J.Cummings@UNESCO.org), we would like to send you a thank you. John Cummings (talk) 13:38, 14 March 2019 (UTC)
Tilmann: I found your message about The Great Sethahedron.svg. I changed the copyright to "public domain" as you requested. You asked, "Could you give me its vertex coordinates?" I do not have those coordinates. A brief video about TGS is posted on YouTube at The Great Sethahedron. Seth was my great grandfather and that is why the name TGS includes "Great". Scott Gregory Beach (talk) 22:10, 16 February 2019 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 03:16, 16 July 2020 (UTC)
I uploaded Mars elevation 2.stl yesterday. It's not quite finished, but I already like it a lot. It's generated via an embarrassingly brute-force method: I use a normalized subdivided Deltoidal icositetrahedron as a base mesh, which leads to more uniform triangle sizes than using a normalized subdivided cube. Then I took nearest-neighbor-interpolated (i.e. not at all interpolated) altitude samples from megt90n000fb.img for a 207.4 MB STL. Finally I decimated the file to a tenth of its original size in Blender. Originally I intended to use an adaptive subdivision approach, but decimation (while being extremely slow and memory intensive) gives excellent results, so I don't feel motivated to do that much additional programming. My texture coordinate calculation seems to be off by one, which causes a small hole in the south pole. A "south hole", if you will. I'm thinking of doing iterated subdivision and decimation to reach an even better mesh. And I should probably use a smarter sampling method for the altitudes.
Hi @TilmannR: , Great that you managed to improve on my model; I like it a lot too. I'm no expert at STL or planetary models, and didn't know Blender can optimise STL files – that's a great resource. If I understand correctly, triangulated irregular networks are far more efficient with polygons. Are you familiar with their use in 3D modelling?
I also don't fully understand what you mean by "the positions of the peaks of your little pyramids". Can you please elaborate?
The "more interesting" side of Mars
P.S. Just a minor nitpick: would you mind turning the planet around (keeping the equator horizontal) so that Tharsis and Valles Marineris are visible in the thumbnail, as on the right? Thanks! cmɢʟee ⋅τaʟκ13:32, 26 March 2019 (UTC)
@Cmglee: Hi. Let's continue the conversation over here. I still don't receive any pings from your talk page.
Yes, Blender is a surprisingly powerful tool. One of the most successful open source projects of them all.
As far as I can tell "triangulated irregular network" is just geography jargon for "triangle mesh". So, yeah, using an arbitrary triangle mesh is more efficient than using a regular grid, because a grid puts unnecessarily small triangles in smooth regions of the planet and it fails to put vertices at the very edge of sharp cliffs or at the very tip of sharp peaks.
By "little pyramids" I mean what you called "subdivide facets with local minima/maxima" in the file history. The center points of these subdivided facets end up far away from where they should be. This is particularly noticeable in the corners of the top and bottom face of the cube when smooth shading is applied, because it makes some triangles face towards the planet, which causes dark lighting artifacts.
I'd prefer, if there was a way to control the camera parameters of STL thumbnails, but I guess adjusting the planet's orientation is the way to go for now.
I'll create a new Earth and Moon once I've found a more efficient way to create these meshes. If I haven't come up with a better method by Sunday, I'll use the "Blender decimation method" again. TilmannR (talk) 21:29, 26 March 2019 (UTC)
My plan to iteratively subdivide and decimate in order get a better mesh didn't quite work out. The rough parts of the planet require a huge amount of geometry in order to be represented accurately, which leaves only very sparse geometry for the smooth parts and that just doesn't look very good. TilmannR (talk) 03:52, 27 March 2019 (UTC)
Got your ping (wonder why you don't get mine...) Anyway, thanks for trying. If Blender automatically does the equivalent of TIN, perhaps it's simpler just to use Blender for that. By the way, I've since found a better DEM for Earth: the ETOPO1 datasetcmɢʟee ⋅τaʟκ13:53, 27 March 2019 (UTC)
Thank you so much for uploading File:Earth_dry_elevation_2.stl and File:Moon_elevation_2.stl and rotating File:Mars_elevation_2.stl. They are much more detailed than mine and are simply stunning! Earth is an example where Blender's optimising may be suboptimal though: it loses the faint coastlines visible in mine, such as north of Australia, that makes it easier to identify countries. If only it's possible to tell Blender to weight the coastlines...
And sorry I'm nitpicking again, but would you mind rotating the globe as you did for Mars to show a more interesting thumbnail? I'll leave what to show to you (though I'm partial to showing both the Himalayas and Marianas Trench myself). Thanks in advance!
Lastly, I read that STL unofficially supports a colour for each polygon, but haven't found any viewers that support this. This is something which put me off making more STL models, as they all end up some bland grey. If colour and preferably other material properties like specularity, opacity and reflectivity were supported, I'd be tempted to add more to my list. Cheers, cmɢʟee ⋅τaʟκ22:38, 27 March 2019 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 03:16, 16 July 2020 (UTC)
WAM logo in Slovak
Hi TilmannR,
I saw that you created WAM logo in Malay and I wanted to ask how did you change the handwriting. I want to make Slovak version of logo with "ÁZIJSKÝ MESIAC" instead of "ASIAN MONTH" and I wrote to 林立云 who wrote the original one if he would be able to make Slovak name but he didn't reply. If it would be possible for you, would you be please able to make the Slovak version?
As I said, "ASIAN MONTH" should be changed to "ÁZIJSKÝ MESIAC" and the "WIKIPEDIA" to "WIKIPÉDIE". The order should be reversed, therefore "ÁZIJSKÝ MESIAC" should be first and "WIKIPÉDIE" below.
If you cannot do it, it is okay, I can still wait if he replies but I wanted to speed things up a bit so I can prepare this year's edition of WAM.
@Luky001: I copied most of the letters from the original image. The "L" and the "U" in "BULAN" are just slightly modified versions of the "T" and the "M" in "MONTH". If I remember correctly, I made the "B" out of two copies of the "O". I don't have any experience in creating calligraphy from scratch, so I recommend waiting for 林立云 to reply.
@TilmannR: Hi, I have the handwritings from Li-Yun Lin, could you please send me an email (via Email this user) so I can send them to you? I don't want to upload them to Commons since those files shouldn't be here permanently and they are not my work. --Luky001 (talk) 11:17, 19 July 2019 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 03:16, 16 July 2020 (UTC)
Any islands that aren't touching the grid will be unprintable. Generating support structures by hand would be tedious. Especially Tasmania and various parts of Southeast Asia are problematic.
Can your 3D printing software take the union of multiple intersecting parts? Blender's Boolean operations are too unstable.
The mesh currently has a radius of approximately 64 arbitrary length units. Unless you give me a conversion factor between arbitrary units and millimeters, it's mathematically impossible for me to complete this task. ;-)
Turns out that not every place with an elevation below 0 is underwater. And some places are temporarily wet/dry depending on the tides. Shall I use File:NASA_bathymetric_world_map.jpg or something else?
Thanks for your very quick response! You raised many good points I haven't thought about, especially the first. I haven't done 3D printing myself; thought I'd send it to a printing service. I'll get back to you after thinking about how to resolve the issues. Thanks again, cmɢʟee ⋅τaʟκ14:39, 27 October 2019 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 03:16, 16 July 2020 (UTC)
I know you solved the librsvgBug relating pattern on small thumbs somehow with scaling. Can you link an example? (I know you did it at some immage(s), but I would be quite slow finding it.)
@JoKalliauer: Oops. I should have read your message more carefully. I already uploaded a fixed version. Sorry.
The problem seems to be that the rasterization fails, when the size of a single tile of the pattern is smaller than one pixel.
The solution is to increase the size of the tiles by manually repeating them (e.g. using two stripes in the pattern definition instead of one).
File:Ducktales2017Logo.svg for example has sixteen circles in the pattern (four clones of a group of four circles), when a single circle would have theoretically been enough.
Hi Tilmann, do you feel like making an STL version of this object? It's a rather obscure representation of this graph, but I think it looks interesting. (And it should be easy to make.) The coordinates and colors I used are in this file. A matching STL of the concertina cube would also be nice. (The coordinates are here.) Greetings, Watchduck (quack) 12:06, 13 June 2020 (UTC)
Danke. In grau ist das Baum-Gebilde natürlich etwas unübersichtlich. Ließe sich das mit den Farben so umsetzen, wie es in dem verlinkten Bild zu sehen ist? Grüße, Tilman mit sechs Buchstaben a.k.a. Watchduck (quack) 19:12, 16 July 2020 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 06:17, 18 September 2020 (UTC)
Flag alterations
@TilmannR: Hello, and sorry for the intrusion, but since you've worked on flag-related files on WC before, I was wondering if you'd be able to help with the improvement of a certain two SVG flag files that I think could do with some adjustments? Snow Lion Fenian (talk) 15:29, 17 July 2020 (UTC)
I say that the edition you made in Periodic_table, good SVG.svg It was wrong, Elements 95, 96, 97 and 98 exist in nature even if it is by reactions in uranium mines from other isotopes or from contamination by atomic bombs and nuclear accidents. (if I'm the wrong one, correct me)
Thanks Once4O4 (talk) 17:32, 14 September 2020 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 07:09, 11 February 2021 (UTC)
SL Police logo
Hi @TilmannR: , Could you make svg image pls and instead of SLPF write Somaliland Police Force and on both sides instead of RU write SL.
and I want to Thank you for making Logo of Somaliland Police Force svg image
Many Thanks 🙏🏽 --
— Preceding unsigned comment added by Siirski (talk • contribs) 20:05, 27 March 2021 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 14:59, 16 August 2021 (UTC)
Colors of the flag of Palestine
I don't have any special knowledge in this area, but a few years ago, someone went around changing all the SVG files of flags of Arab countries to use a uniform standardized shade of green. Further back, there was a clash of strong opinions about the shade of blue to be used in Chinese-related flags... -- AnonMoos (talk) 22:32, 20 April 2022 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 02:32, 21 April 2022 (UTC)
If you created this file, please note that the fact that it has been proposed for deletion does not necessarily mean that we do not value your kind contribution. It simply means that one person believes that there is some specific problem with it, such as a copyright issue. Please see Commons:But it's my own work! for a guide on how to address these issues.
Please remember to respond to and – if appropriate – contradict the arguments supporting deletion. Arguments which focus on the nominator will not affect the result of the nomination. Thank you!
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 12:21, 25 April 2022 (UTC)
Your barnstar awarding
Danke Tilmann, für den barnstar. Du hast sehr richtig beobachtet dass ich mich vor allem um technische Verbesserungen bemühe.
Da ich wegen dieser Verleihung aktuell wieder auf dich aufmerksam wurde, komme ich dir hier nun mit einigen Anmerkungen.
Du erzeugst sehr gute SVG. Gelegentlich wird von dir auch die Beschreibung von Image generation eingebaut, gelegentlich unterbleibt diese Anmerkung. Ich würde dir die generelle Verwendung des script empfehlen, trotz aller dort noch vorhandenen Māngel kann damit viel Aufwand eingespart werden.
user name redlinks: In einer eigenen User page sehe auch ich keinen Vorteil, und wie unzählige andere nicht einmal eine Erfordernis dafür; hingegen sehe ich (im Gegensatz zu dir?) keine Attraktivität an redlinks: meine Beteiligung sehe ich nicht als so wichtig dass mein username derart stark farblich hervorzuheben ist, und ganz generell sollte IMHO jeder link direkt irgendwohin führen, ohne weitere links zu erfordern. Deshalb fände ich es besser wenn du eine Pseudo-Benutzerseite einrichtest mit #REDIRECT [[User talk:TilmannR]]. Aber das ist natürlich eine autonome Entscheidung von dir! Übrigens, an vielen Stellen - Vorlagen, script, cleanup - werden von mir solche redlinks gebläut, voll automatisch.
Barnstar - technical works.svg: ganz typisch Inkscape, voll von Datenmüll (opacity:0.50000000000000000); von Grund auf neugeschrieben würde sinnvoller Code sicher weit weniger als 10% erfordern. Auch wenn es ohne Nutzen wäre, eine Herausforderung, zumindest eine nette Übung. PX! -- sarang♥사랑08:15, 8 May 2022 (UTC)
Danke für das Kompliment! Soweit ich mich erinnern kann, habe ich noch nie {{Igen}} in meine eigenen Uploads eingebaut. • Die Parameter sind nicht menschen-lesbar und ich bezweifle, dass derartig komprimierte Templates besser sind als eine Aneinanderreihung mehrerer einzelner lesbarer Templates. • Automatisierbare Tasks (z.B. W3C validation) sollten automatisiert ablaufen. Circa 0 % der Benutzer kennen die entsprechenden Templates/Scripte/Guidelines. • Spätestens nach der Optimierung ist irrelevant welches Programm ich für WYSIWYG-editing verwende.
Gute Argumente. Nur habe ich mich vier Jahre lang an meinen Redlink gewöhnt. Und solange ich keine (Pseudo-)Benutzerseite habe, ist genau dessen Abwesenheit ein Ausdruck meiner Individualität.
Tja. Da war wohl jemand hinreichend dankbar um dir etwas in den User Talk zu schreiben, aber nicht dankbar genug um dafür ein neues, handoptimiertes SVG hoch zu laden. Vielleicht nächstes Mal. TilmannR (talk) 14:47, 8 May 2022 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 08:40, 14 September 2022 (UTC)
Update - File:Kandler 1998 Early diversification of life and pre-cell theory.jpg ?
Bascially I'm trying to build my PhD on improving the representation and access to information. I saw your SVG and wanted to build more like this, wanted to understand how it works and extend the usage of this representation to other fields, hopefully semi-automatic.
@TimBorgNetzWerk: Well, what makes this SVG special is that it contains a <script> node, which overwrites itself with static SVG content. It's a compromise between
wanting to use scripting for improved maintainability
without requiring the installation of any compilers, interpreters, or libraries
while the direct uploading of SVGs with scripts is disallowed for security reasons.
It's a solution to a very specific problem. Your information sharing network doesn't have the same constraints. You can just feed plain old data into an off-the-shelf data visualization framework. And based on your tweets it seems like you're already familiar with some data mining and visualization techniques. TilmannR (talk) 16:54, 10 January 2023 (UTC)
Thanks alot again for the reply, I'd like to continue that train of thought though:
If i have an overview over these visualizations, or have generated such a visualization myself - wouldn't it be nice to extract where what dot is, save it to an xy map, feed that back through your script and have low-weight map of anything you'd want on any type of wiki, with the advantages you decribed?
In any case, I'd love to learn what you did exactly and how I would replicate it. Maybe you can just give me some starting points, like "take a look at my scribt and read xy documentation" or whatever, I'd appreciate any input :)
If you are already familiar with JS, then you can just read the script. It's a bit old-school, but I don't think that diminishes its legibility too much.
It's mostly arithmetic and string processing anyways. As you already know tree_text contains all the text and the names of Wikipedia articles. The function generate takes the data, computes the positions for the text and the handles of the bezier curves, and converts it all into a string to be inserted into the SVG.
wouldn't it be nice to extract where what dot is, save it to an xy map, feed that back through your script
It is not universal. It cannot generate visualizations that are substantially different from what it currently generates unless it's heavily modified. And while the circular arrangement is visually pleasing, I don't think it's a good idea to make people read vertical or diagonal text. So you're much better off using a more powerful data visualization library instead of trying to reuse this script. TilmannR (talk) 18:57, 15 January 2023 (UTC)
Thanks, I guess that's more than I could have ever asked for :)
I was and still am just so impressed about the potential of having a graphic that is searchable. With this explanation, I'll be able to work it out once recreating or utilizing this is on the agenda for my project. I'll pin this conversation and thank you again for your help ;) TimBorgNetzWerk (talk) 19:59, 15 January 2023 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 20:31, 15 January 2023 (UTC)
@Mrmw: Die Bilder sind in Inkscape per Hand getraced, dann mit SVGOMG optimiert. Die Vorlage steht jeweils als "source" in der {{Information}}. Das "M" logo und einige der Ziffern sind aus den "other versions" kopiert, damit alle Versionen einigermaßen konsistent sind. TilmannR (talk) 11:46, 21 February 2023 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 18:56, 3 March 2023 (UTC)
Rollback right given
I have granted rollback rights to your account; the reason for this is that after a review of some of your contributions, I believe I can trust you to use rollback correctly by using it for its intended usage of reverting vandalism, and that you will not abuse it to revert-war. For information on rollback, see Commons:Rollback. If you do not want rollback, just let me know, and I'll remove it. Good luck and thanks. —Mdaniels5757 (talk • contribs) 00:11, 4 March 2023 (UTC)
For the record: Rollback rights turn out to be less useful than I hoped they would be, because one can only rollback the most recent user's edits, even if it's just a minor edit by a bot account. Therefore all vandalism that gets categorized by a bot (e.g. here) is un-rollback-able. TilmannR (talk) 20:41, 4 March 2023 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 02:48, 4 March 2023 (UTC)
If you created this file, please note that the fact that it has been proposed for deletion does not necessarily mean that we do not value your kind contribution. It simply means that one person believes that there is some specific problem with it, such as a copyright issue. Please see Commons:But it's my own work! for a guide on how to address these issues.
Please remember to respond to and – if appropriate – contradict the arguments supporting deletion. Arguments which focus on the nominator will not affect the result of the nomination. Thank you!
I do not recommend posting such requests in multiple places at the same time. Imagine if Goran tek-en and I had immediately started working on this request. One of us would have wasted their time.
Assuming that the image is based on File:Somalia location map.svg, you need to adjust the source and license information of your file, because the original file is under a license, which requires that derivative works use an equivalent license and mention the original author. TilmannR (talk) 19:26, 26 March 2023 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 20:08, 21 May 2023 (UTC)
@Glrx: Humility requires me to declare that I didn't do anything impressive. But I do appreciate the compliment. It really brightened my day. TilmannR (talk) 18:53, 6 April 2023 (UTC)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 20:08, 21 May 2023 (UTC)
If you created this file, please note that the fact that it has been proposed for deletion does not necessarily mean that we do not value your kind contribution. It simply means that one person believes that there is some specific problem with it, such as a copyright issue. Please see Commons:But it's my own work! for a guide on how to address these issues.
Please remember to respond to and – if appropriate – contradict the arguments supporting deletion. Arguments which focus on the nominator will not affect the result of the nomination. Thank you!
I was wondering, seeing as you’ve already started :), do you think that you would have the time to also create the same type of 3d-renderings of Venus and of Mercury? That way, you would be providing a full set for people to marvel at!
(If you don’t have the time to do both, I think Venus would be the more significant of the two, considering its many similarities to our own planetary home.)
Thanks in advance, and I look forward to hearing back from you!
I am aware that the necessary radar data for creating the models does exist, likely on the same nasa/wu pages that you have used in the Mars model, such as perhaps here :
I'm not actively participating in the Wikipedia Commons at this time. I'm busy with various other things.
I've searched for the script I used to generate the Earth and Mars models, but apparently I no longer have it.
As far as I know the high-quality topographical data of Mercury and Venus is incomplete. Mixing it with low-quality estimates would lead to distracting visual artifacts unless a lot of effort is invested into generating fake data.
They are prints of 3D relief maps of Venus and Mercury (Mercator projections), which seem to be extremely detailed. So even if the datasets that you would need are not to be found at the WU-StL links that I originally provided, they certainly seem to be available somewhere.
I wish you every success with whatever it is that you are working on currently! :)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 15:36, 3 January 2024 (UTC)
Self-generating SVG with JavaScript
Hi TilmannR,
I've just come across File:Cognitive_bias_codex_en.svg and am intrigued by your idea to generate a static SVG from one with JavaScript. I have been generating SVG with Python. As you wrote, the advantage of your method is that one doesn't need any software besides a web browser.
I'd like to experiment with your idea but somehow embed the JavaScript as a comment in the SVG itself, so that the source is never detached from the file. If one saves my SVG files generated from Python as a .py file and deletes the first line (I would've preferred not having to do this but couldn't find a way), one gets a valid Python script that generate the SVG.
Might you be interested to further your experiment in a similar manner? I suppose the editor has to replace
@Cmglee: Hi! The cognitive bias SVG has the following structure:
<?xml version="1.0" encoding="UTF-8" standalone="no"?><svgxmlns="http://www.w3.org/2000/svg"width="600"height="600"viewBox="-100 -100 200 200"><gid="generated_content"><scripttype="text/javascript"><![CDATA[ var result = ['<circle r="100"/>', '<circle r="50" fill="#fff"/>']; // Just an example document.getElementById("generated_content").innerHTML = result.join(''); ]]></script></g></svg>
so the line that makes the result visible simultaneously removes the script. With a slight modification...
<?xml version="1.0" encoding="UTF-8" standalone="no"?><svgxmlns="http://www.w3.org/2000/svg"width="600"height="600"viewBox="-100 -100 200 200"><gid="generated_content"></g><gid="generator_script"><scripttype="text/javascript"><![CDATA[ var result = ['<circle r="100"/>','<circler="50"fill="#fff"/>'];//Justanexample
document.getElementById("generated_content").innerHTML=result.join('');
vargenerator_script=document.getElementById("generator_script");
varscript_as_comment=generator_script.innerHTML.replace(/<scrip[^>]+>/,'<!-'+'-scrpt>').replace('</scr'+'ipt>','</scrpt-'+'->');
generator_script.innerHTML=script_as_comment;
... you can keep the script as a comment. Then the person using the script only has to replace <!--scrpt> with <script type="text/javascript"> and </scrpt--> with </script>, which is easily automated.
Hi TilmannR, many thanks for explaining how your method works and devising a way to embed itself. I'll experiment with it the next time I've to upload an SVG needing scripting. If it works well, would you mind if I write it up as a how-to, of course crediting you? Cheers, cmɢʟee ⋅τaʟκ16:47, 27 December 2023 (UTC)
@Cmglee: I wouldn't mind at all. I'm just not sure whether it's wise to invite other people to distribute arbitrary source code via Commons. It's not what Commons is for, and we have no way to protect people, who just want to translate or fix typos, from executing malicious code other than JS being somewhat sandboxed by default. TilmannR (talk) 08:01, 28 December 2023 (UTC)
That's a very fair point. Guess any language we use to generate the SVG has risks of someone exploiting it. At least Javascript is sandboxed, though due to potential browser exploits, even an external tool like Toolforge has risks. Cheers, cmɢʟee ⋅τaʟκ13:10, 28 December 2023 (UTC)
The analogue of Gabriel's horn in two dimensions has an area of 2 but infinite perimeter
I've made my first SVG using your script, modified and commented to help other editors as much as possible. Would you have some time to review my changes? Thanks in advance, cmɢʟee ⋅τaʟκ07:07, 31 December 2023 (UTC)
Converted the loop into a function (easier to reuse & replace)
Using string interpolation instead of + (asymptotically faster, but doesn't matter for a small number of short strings. P.S. I wouldn't be surprised, if modern JS interpreters automagically optimize string concatenation.)
Replacing var with const (slightly nicer semantics)
piecewise_linear_approximation could be replaced with a cubic_bezier_approximation function, but I'm too lazy/tired to implement that right now. I wonder whether there's a better way to implement number_to_string i.e. Number.toFixed() but without superfluous trailing zeros.
Thank you very much for your feedback. You seem an expert in JS! I've implemented most of your recommendations.
Though you are right that a const array (out) can still be pushed to, I wonder if that might mislead editors (as it did me until I looked it up).
One nitpick I have is that non-ASCII characters specified thus: ± are replaced: ± so the file is no longer ASCII. When opened in ASCII-only text editors or fonts, the characters are garbled. Even if it works, it becomes difficult to distinguish different types of dashes. Is there a way to force the saved file to be ASCII?
I don't consider myself to be a JS expert, but thanks for the compliment.
It's all a matter of perspective: Not the array is constant, but the reference to the array is constant. So in a way const makes the code less confusing, because you know that (within the current block) the given identifier always refers to the same object.
UTF-8 is a very popular standard, so I wouldn't worry about limiting the character set to ASCII. Imagine a file with Chinese or Arabic or Russian – any kind of non-latin text: If all those characters are converted to escape sequences, the text is no longer human-readable and becomes much more difficult to edit. TilmannR (talk) 11:43, 1 January 2024 (UTC)
😊 I see your points. I've updated the code and added a comment to avoid confusion (SVG may be pushed into the array even if out is a constant.)
It's true for text in different languages. It can lead to confusion for different types of dashes or spaces, though. Cheers, cmɢʟee ⋅τaʟκ14:38, 2 January 2024 (UTC)
Hi TilmannR, in case it interests you, I've begun porting my US choropleth map to this model.
I also adapted your number_to_string function to strip trailing zeros as follows:
Note that .replace(/\.0*$/, '').replace(/(\.\d+)0+$/, '$1') turns 0.1000 into 0.100, because the \d+ will greedily match all but the final 0. You can fix this by either explicitly asking for a nonzero digit (i.e. \d*[1-9]) or making the + lazy by appending a ?, which is shorter, but also less legible.
Thank you very much, TilmannR, and also for telling me why I should use syntaxhighlight; I've updated the JavaScript.
Would you also be interested to have your invention (generating SVG with JavaScript) presented at Wikimania 2024? If you'd rather not present it, would you mind if I did, giving you full credit, of course? I think that it is a superb way to programmatically generate SVG and that the community would benefit from more widespread use of this technique.
Oh, that's a tricky question. On one hand giving a presentation might help me become more confident and better at giving presentations. On the other hand I have very little to say about this. The key insight is that browsers execute script tags in SVG files. Everything else is just implementation details, which are (in my opinion) not particularly interesting.
🤔
P.S. Instead of using the Levenshtein distance to fuzzy-select an abbreviation I'd output an error message, using the Levenshtein distance to suggest the "correct" way of spelling a given name. Otherwise you can get unexpected results, if someone for some reason names a territory that isn't in your list. TilmannR (talk) 12:19, 17 January 2024 (UTC)
Thanks for your support. If you think it's too short, we could split the session into your generation with JavaScript, and I could talk about my prior generation work with Python, Perl and Google Sheets.
The deadline for scholarships has passed and presentation submissions are due in June so we have plenty of time.
) TilmannR (talk) 21:26, 20 September 2018 (UTC)
.svg)
Done. TilmannR (talk) 20:06, 29 January 2019 (UTC)
.svg)
This section is resolved and can be archived. If you disagree, replace this template with your comment. TilmannR (talk) 03:16, 16 July 2020 (UTC) 
Moved from User_talk:Cmglee 
-- AnonMoos (talk) 22:32, 20 April 2022 (UTC)
TilmannR (talk) 14:47, 8 May 2022 (UTC)
.png)
? Suthernorthernwestby (talk) 16:50, 26 March 2023 (UTC)
_=_z*z+0.25.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
_%3D_z*z%2B0.25.svg){kind=link}
_=_z*z+0.25.svg){kind=link}
_=_z*z+0.25.svg){kind=link}
_=_z*z+0.25.svg){kind=link}
![[2]](https://commons.wikimedia.org/w/index.php?title=Special:Log&page=File:USD_Bitonto_Calcio.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
![[3]](http://heraldicum.ru/uzbekstn/towns/images/tashkent3.gif){kind=link}
.png){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.png){kind=link}
.svg&action=edit&redlink=1){kind=link}
.svg){kind=link}
.svg){kind=link}
