Help:SVG

This page in other languages:

Official logo for the SVG file format

SVG (scalable vector graphics) is an accepted file format on Wikimedia Commons. It is suited to diagrams, maps, and other non-photographic images. SVG is highly appreciated on Commons because it describes graphical objects with a sourcecode and thus currently provides the best way to modify and improve content.

SVG files are used extensively on Wikimedia Commons. Here are a few tips to get started.

What is SVG?[edit]

An example of an SVG image — note that at whatever scale you view the image, its curves always appear smooth, never "jagged"

Scalable vector graphics (SVG) is an XML specification and file format for describing two-dimensional vector graphics (as opposed to a "bitmap" or raster graphic) in both static and animated forms. It is stored in either plain text or compressed (binary) format which is much more size efficient than most other image formats such as JPEGs or GIFs. SVGs also allow embedding of raster graphics and editable text.

Creating SVG images for Wikimedia Commons[edit]

Programs you can use to create SVG images are shown in section editors hereafter. Alternatively you can edit the SVG code directly with any text editor or IDE.

The following sections explain specific features of SVG.

Before you upload: validation and checking image appearance[edit]

Validation[edit]

Although not a requirement, it is often a good idea to run your file through the SVG validator (validator.w3.org) before you upload. This can tell you about possible problems in your SVG file which might cause the file to render incorrectly in different platforms. Unfortunately this validator cannot handle RDF or other metadata, should you wish to include it, but it can still find errors in your SVG code. It supports various Doctype declarations, but this is not a requirement in SVG.^[1]^[2]

See #Tagging SVG files for how to tag a file as validated.

Image appearance[edit]

Different software programs and different browser platforms may render the exact same SVG file in slightly different ways. The way you see your final image in Adobe Illustrator may not reflect how that image will look once it is uploaded to Commons and is viewed by others. Also, all SVG images are automatically converted to PNG files when used in a Wikipedia article, and this may affect how the final image looks as well. As an example of the former, in an image containing a series of very fine strokes, those strokes will probably look much more bold and thick in Illustrator than they will once the image has been converted to SVG and viewed in Google Chrome. Those who create SVG files should consider saving their work in SVG format and then opening that same image using two or three different browsers to verify that it is rendering as intended. A similar check can be done by opening the file in other image editing programs as well.

To preview how an SVG will be rendered to PNG on Commons, go to Commons:SVG Check.

The reasons why the same file becomes rendered differently depending on context is a result of how these contexts (browsers or programs) interpret the file's SVG code and use that code to generate the image. Aim to create a "least common denominator" SVG that renders correctly no matter what browser is used to open it, and never assume that your own image editing program is showing you what the image will really look like to other viewers.

Also, some programs will allow you to insert raster-based components into your file without warning you that if the file is converted into SVG these components will not simultaneously convert to vectors, meaning that you may need to be very careful about the sorts of filters and effects you include in an image and should check for any extraneous embedded "links" within an image before you upload the SVG to Commons. How to do this will vary from program to program

However, if you are using Adobe Illustrator, here are examples of things which will cause Illustrator to convert a path into a raster component:

Any pattern applied to any fill, including patterns that are automatically included in Illustrator. You can sometimes work around this by converting any patterns you have used into rasters yourself, and then using the Image--> Trace function to convert these back into vectors.
Any path which contains more than one of Illustrator's "Effects"— e.g., a fisheye warp and a freehand distort. Illustrator cannot handle two of these in the Appearance menu of any single path and will rasterize them.
Any path which has been altered with an Envelope-distortion. You can sometimes work around these in the same way you can work around the pattern problem mentioned above, but the results are usually disappointing.
Any path which has been assigned the Raster effect under the Effects menu. To retain these as vectors, you may only need to undo this effect.
Illustrator allows users to apply gradients to strokes in three ways: gradients within strokes, gradients along strokes, and gradients across strokes. However, only the first of these, gradients within strokes, is supported by an equivalent SVG code; any strokes containing gradients along or across them will be converted into rasters when the image is saved in SVG format.

These are just a few examples of things which will cause Illustrator to produce a raster when a vector was intended. The simpler these paths can be, the more likely Illustrator will be able to retain them as vectors.

For more information on how to check SVG files for MediaWiki, see #How SVG files work in MediaWiki.

SVG document declaration[edit]

<?xml version="1.0" encoding="utf-8"?>
<svg
xmlns="http://www.w3.org/2000/svg"
xmlns:xlink="http://www.w3.org/1999/xlink"
width="400" height="300"
viewBox="0 0 400 300"
>

The source code of every SVG file should begin with the basic information of its dimension and namespace. Document type declaration or DOCTYPE is entirely unnecessary. W3 insists that character encoding declaration is essential, and its validator warns (instead of reporting an "error") for its absence; in the absence of all other errors, the validator returns a verdict of "tentative pass" instead of "pass". The important thing about character encoding is that the software with which you create or edit the SVG file should provide the option of choosing character encoding. UTF-8 is RECOMMENDED for internationalization as opposed to ANSI or Unicode, but avoid the use of byte-order mark (BOM) because of poor compatibility with older applications.

Inside the SVG element, width, height and namespace attributes are required. Add viewBox attribute if you want to change the scale and position of the image. Of all XML namespaces, xmlns="http://www.w3.org/2000/svg" should at least be present, otherwise the SVG will not be rendered at all. If you have used the linking function in your SVG, xmlns:xlink="http://www.w3.org/1999/xlink" must be included too to enable it.

Bitmaps[edit]

There are some valuable uses of bitmap images within SVG files, such as annotating a photograph (e.g., labeling the parts in an anatomical photograph). However, most uses of bitmaps in SVG files are not necessary and bitmaps are often better converted or redrawn as vectors. Bitmaps should be included within an SVG file if for some reason this is the best solution, but should be redrawn as vectors if this yields an overall better or more informative image. Further, other editors who come across the image on Commons may decide to apply the tag {{BadSVG}} to an SVG image containing undesirable raster-based elements.

Please think carefully about using bitmaps in SVG files, and consider converting them or redrawing them as vectors when incorporating them into an image rather than including a bitmap directly. See the talk page for further discussion of this issue. If you do decide to use a bitmap image, please be sure to embed and not link to the image; otherwise it won't render.

Animation[edit]

SVG images can have animation built in with SMIL, but because the SVG is converted to raster PNG format for use in Wikimedia projects, the animated SVG file is only visible from the image source page.

SMIL, however, is getting diminishing support across browsers. Internet Explorer does not support it altogether; Google has planned to deprecate it in April 2015 with the possibility that its component of SMIL may be removed completely from Google Chrome someday later.^[3]

See #Tagging SVG files for how to mark an SVG file as animated.

Title[edit]

The "title" element right under the SVG element serves as the title name of the SVG image itself, but it can be also applied further inside of the descendant elements within the SVG image. Most modern desktop browsers render these titles into popup tooltips which help readers to identify the meaning of the object on mouse-hover even if there is text label by the object. Similarly, title can also be applied on text too for collapsing message but it may be a good idea to give some hint of the existence of the title element by using text decoration like underline. XML comment is more suitable for technical information which should not be displayed to the reader normally but other users who may edit the SVG source code.

Currently most mobile browsers can not display any title content innately except for the one right under the SVG element because there is no equivalent of hovering the pointer on mobile devices.

<?xml version="1.0" encoding="UTF-8"?>
<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" width="50" height="50">
	<title>Title of the whole SVG image</title>
	<g id="shapes">
		<title>Title of the "shapes" group</title>
		<rect x="0" y="0" width="10" height="20" fill="#f00">
			<title>Title of the rectangle</title><!--XML comment-->
		</rect>
		<circle cx="5" cy="5" r="3" fill="#0f0"/>
		<text x="2" y="30" text-decoration="underline">Text
			<title>Title of the text</title>
		</text>
	</g>
</svg>

Scripting[edit]

MediaWiki does not accept SVG files containing scripts.

External files[edit]

Before saving the final version of an SVG image you should remove any references to other files you used to make the SVG (this is what is meant by "linking"); if these references are left in the SVG text, then the render will not find the external file(s) and will fail.

Tidying up[edit]

It's also a good idea to remove or "vacuum" any unused definitions (defs), as these can needlessly bloat the size of the file. Scour is an open source Python script that cleans and optimizes SVG files. It has integrated in Inkscape as of version 0.47; you can use it by saving as Inkscape's "Optimized SVG (*.svg)".

Plain SVG, compressed SVG, generic specifications[edit]

To optimize compatibility across browser platforms it is generally best to save most SVG files in "plain SVG" format whenever possible. This option usually appears in the program's Save As dialog box as an alternative file format (as opposed to saving the file in "SVG" format, which might be the program's own version of SVG).

In Adobe Illustrator or Inkscape the default file format includes information (such as user preferences) that is not needed in uploaded files. This increases the file size unnecessarily and sometimes causes thumbnails to not render properly.

On the other hand, saving in "plain SVG" format may result in the loss of some useful information from the file. In Inkscape, for example, plain SVG format removes layer information. For this reason you may wish to create a plain version just for uploading to Commons and retain an alternative ‘fancy’ master version for your own purposes, or tag the image with {{Created with Inkscape|IMPORTANT=yes}}. See Help:Inkscape for more.

Inkscape and other programs also allow users to save files in SVGZ (compressed binary SVG) format. This format is not supported on Commons and should not be applied to any images uploaded here. Such images will not render.

Fonts / text[edit]

A limited number of fonts can be rendered using the <text> tag. You can find a list here, and sample renderings here. Note that some fonts are not available in Regular style, only in Bold or in others such as Italic; consider alternatives like Liberation. Please consider using these fonts and not converting fonts/ text to paths for the following reasons:

Saving as a font rather than paths can greatly reduce the overall file size, particularly if the image contains large amounts of text.
Typographical or grammatical errors in the image can be fixed by other editors without having to redraw the entire set of text.
Text can be readily translated into other languages, an important quality since Commons is a multilingual project. Paths require much more work to translate. See also Template:Translate.
Text can be easily searched with search engines, which may not recognise path outlines.

Text/ path problematic SVG files may be tagged with {{Path text SVG}}. For linking text modifiable (what we have been calling "text" or "font") and text non-modifiable (i.e., path-based) versions of an SVG file use {{Vector text versions|…}}.

If you still insist converting text to path for reasons like the lack of support of textPath function or rare glyph which is not present in most typefaces, you should duplicate the raw text group to overlap the converted text and add the attribute fill-opacity="0" stroke-opacity="0" to this group in order to hide the raw text. This way the text in the original SVG file can still be searched, clicked and highlighted by readers manually. opacity="0" may lead to some misunderstanding because it will make the highlight invisible while still selectable. display="none" prevents readers from searching the text without accessing the source code.

Black rectangle (Flowed Text bug) [edit]

Inkscape supports a feature known as "Flowed Text" which automatically wraps text to fit into the bounds of a given text box. Unfortunately this feature is not part of any SVG standard (it's only contained in a working draft of SVG 1.2) and its usage will almost certainly result in compatibility problems.^[4] On Wikipedia, a flowed text box does not render at all or renders as a black rectangle (phab:T43424 – depending on font size and color).

However there are some easy ways to avoid and/or fix this issue:

The easiest solution is to only use a single click (without dragging) to create text with Inkscape's text tool. (In contrast when holding the mouse button and dragging the mouse to draw a text box, you're specifically telling Inkscape to use Flowed Text).
If you already created a Flowed Text box you can convert it to normal text using Text → Convert to Text.

If you accidentally created a Flowed Text box which is empty (i.e. contains no text) it's not visible or selectable in Inkscape but it still shows up as a black rectangle on Wikipedia. To delete it there are two possibilities:

Use Inkscape's built in XML editor (Edit → XML Editor) and delete all Flowed Text boxes (search for all nodes which are called <svg:flowRoot id="flowRootXXXX"> in the tree view).
Use a simple text editor (e.g. Notepad) and search for the text "flowRoot". Delete all Flowed Text boxes manually by removing their opening XML tag (<flowRoot ...>), their closing XML tag (</flowRoot>), and all text between two matching tags.

Text transform limitations[edit]

Font scaling currently does not work as expected if horizontal and vertical scales (according to the effective transform matrix) are not equal (the horizontal stretch will be ignored: only the font height, computed from the matrix is currently honored); this can prevent the correct rendering in narrow/condensed or width-extended styles from another font, with text taking unexpected width and flowing out of the expected box (the alternative is to select an alternate free font family which is already narrowed/condensed).

Beware also with font metrics differences: an SVG file may look good when rendered directly in a SVG-capable (all modern) browser under Windows or Mac OS (which can use common, but non free fonts such as Courier or Courier New), but will look wrong when restricted to the list of free fonts currently supported by the MediaWiki's SVG-to-PNG renderer or when viewed under OSes like free distributions of Linux.

Writing mode limitation[edit]

A diagram in Chinese which could benefit from the top-to-bottom writing mode but turns out not quite right in librsvg…

Unlike Inkscape and Adobe Illustrator, the librsvg software used by Wikimedia and some browsers does not properly support the writing-mode property. When this property is given any non-default (left-to-right writing mode) value, all characters will clump together in librsvg (phab:T65236) or the property is simply ignored in Mozilla Firefox. If the alternative writing mode is still desired (e.g. top-to-bottom mode for Asian languages), align all characters separately instead of relying on this property.

Fallback fonts[edit]

When the Arial font is found in an SVG file, the Liberation Sans font is currently (February 10, 2011) being used instead of Arial for generating the PNG files. See the following File:ArialSubstitutionTest.svg for test results. To future proof the SVG file in case Wikimedia changes the Arial substitution font, change all occurrences of ‘font-family: Arial’ to ‘font-family: Liberation Sans, Arial, sans-serif’ (leave out the quotes). Adding Arial and sans-serif as fallback fonts will help the SVG version of the file be displayed properly on machines that do not have the Liberation Sans font installed. Some other fonts are also interpreted, such as the Times New Roman font, which is interpreted as the current default font Liberation Serif.

If you are using non-Western characters, you should define those typefaces preceding Liberation Sans. For example, if your text contains Chinese characters, font-family definition should be: ‘ Liberation Sans, Arial, WenQuanYi Zen Hei, SimHei, sans-serif’. In this case, western fonts are put in the front because the quality of Western charachers in Chinese fonts is often low, and we want to use the higher-quality ones first.

Inkscape begins supporting fallback fonts from version 0.91.
Adobe Illustrator version 18.1.1 or earlier (probably in later versions as well) does not support fallback fonts, so the SVG file has to be manually updated in a text editor after it is saved.

Layout with text and tspan elements[edit]

Although you can use only the text element to render simple text in the image, the tspan element can be useful to handle text which the position of the word is dependent to the other.^[5] A tspan element must be nested within the text element, and then more tspan can be nested within this tspan, all styling defined in the parent text element will be inherited by its child tspan elements. For example:

<g style="font-size:12px">
  <text font-weight="bold" x="10" y="10">Wikipedia hosts knowledges
    <tspan fill="blue"> contributed
      <tspan text-anchor="end" dy="12"> by volunteers</tspan>
    </tspan>
  </text>
  <text font-style="italic" x="15" y="30">Wikimedia Commons hosts free media files for all sister projects.</text>
</g>

The primarily difference with the "g nesting text" structure is that few important text layout behaviors are exclusive to tspan: When multiple text elements are lined up consecutively, they still position themselves independently; if multiple tspan elements are lined up consecutively but the second and subsequent tspan elements are not given absolute x coordinate, text of these tspan elements will line up horizontally right next to the end of the character of prior tspan in the same text element.

However, tspan should be used sparingly even if you define the absolute x coordinate for every single tspan element in the text element. This is because under the default xml:space definition, any line break (within text or tspan) with an extra trailing white-space is converted to a blank space character automatically.^[6] If there happens to be multiple consecutive line breaks, they are treated as one space still. This would not be a problem if the line break is followed by a tspan element with absolute x coordinate and is align to left (text-anchor=start, the default value). If xml:space is given the 'preserve' value, librsvg will translate the indent to a long white tab character equal to 8 space characters. Because the auto-formatting or re-indentation (indent style) of most XML editors creates indent for every new descendant XML tag automatically, use re-indentation carefully if your SVG file contains such usage of tspan.

The behaviors in different SVG renderers to this issue varies much and is usually inconsistent to the W3C specification (also the 'preserve' value of xml:space will be deprecated in SVG 2.^[7]). The only problem with librsvg is that it never converts the line break alone into space no matter which value of xml:space is given if there is no trailing space or indent before the tspan tag. Except for librsvg, nearly all modern browsers always add the converted space character right after the preceding text, this is the reason why the end or middle values of text-anchor property always align the text 1 space leftward because the new line and/or the trailing tab or space after the closing tspan tag is converted to an unexpected space character.^[8]

Re-indentation	XML Code	Result
Before	<text> <tspan>O</tspan><tspan>n</tspan><tspan>e</tspan> </text>	One
After	<text> <tspan>O</tspan> <tspan>n</tspan> <tspan>e</tspan> </text>	O n e

Inkscape is buggy to render tspan, for example, inheriting text-anchor property from the first tspan element incorrectly for the subsequent tspan elements even if the property is specifically given a value in all tspan elements.^[9] The only way to redeem it is group tspan with the same text-anchor value within the same text element and refrain from applying text-anchor in tspan. This fix is entirely cosmetic for appearance in Inkscape. Ignore this issue and use tspan as you please if you are not going to open or edit the SVG file in Inkscape.

If you are generating an SVG using Adobe Illustrator, you will be given options to "Output fewer <tspan> elements" and "Use <textPath> element for Text on Path" during the save dialog. To maximize the consistency of the appearance of your font across browser platforms, you should place a check mark next to the former and make sure there is no check mark next to the latter. If you do not do this, the kerning of some of your letters may cause them to overlap (depending on your browser).

Line wrap

As of version 1.1, SVG does not provide any form of line wrap (either hard or soft). Any line wrap should be done manually by repositioning new text or tspan elements with the same absolute x coordinate and a new y coordinate. For tspan this can be easier organized by using relative positioning with dy property. Although textPath can handle the similar situation,^[10] this element is not supported by librsvg/Wikimedia (phab:T11420).

Font-weight value[edit]

You can change the boldness of the text using font-weight property. The W3 SVG specification states that the default value is normal^[11], i.e. no bolding or thinning at all, but librsvg does not recognize this particular value. If you want to revert the text to normal weight but the font weight has already been defined in the parent element, use the value 400 in place of normal.

Font styling class[edit]

To ease the process of updating font family for every text or tspan elements, define the font CSS class(es) within style element and apply the class property in elements like this:

<style type="text/css">
text {font-size:12px; font-family:Liberation Sans, Arial, sans-serif} <!-- General font styling-->
.small {font-size:9px}
.special {fill:#f93500; font-weight:bold}
.title {font-size:14px; font-weight:bold; font-family:Liberation Serif, Times New Roman, serif}
.italic {font-style:italic}
</style>
<text class="title" x="10" y="20">Qridfs</text>
<text x="10" y="30">Rfnkl fgkj qljf fgk gskla</text>
<text class="small">
  <tspan x="15" y="40">Dfj fdmnkl</tspan>
  <tspan x="15" dy="12" class="italic">Akgfld fkdngf mna</tspan>
</text>
<text class="special" x="10" y="60">Tcjgh xlij qpfj</text>

Even if one particularly style is defined once in the SVG, it is still best advised to replace the style with class for a tidier layout of XML codes. But it is also good to use SVG groups instead of CSS classes (these can also produce a smaller code by lesser librsvg limitation, see note.

Note

phab:T68672 – CSS style element get ignored if the default value type="text/css" is omitted.
phab:T43423 – CSS child selectors are not supported by librsvg.
phab:T68551 – CSS classes on text elements get not inherited on child tspan elements. If you want this, you need to add a separate class for tspan as well, or use simply groups (but not in the text element as it Scour unfortunately does).

Text with background edge[edit]

Sometimes the text might be unreadable because the background color has low contrast with the font color or the pattern is too complicated that confuses the reader. This can be solved by duplicating the text in exactly the same position, but the back text rendered with stroke and the fore text without it like this:

<text style="font-size:12px">
  <tspan x="10" y="20" style="stroke:white; stroke-width:3px; stroke-linejoin:round;">Placeholder</tspan>
  <tspan x="10" y="20">Placeholder</tspan>
</text>

"stroke-linejoin:round" is defined to avoid sharp corner of the text stroke.

The problem with applying stroke and fill to the same text element is that the stroke will be rendered over the main body of text, making it otherwise unreadable if the font is too small but requires a larger stroke to be distinguishable from the background.

If you have lot of text that needs to be rendered with a stroke background, you can manually include the text within <g>, <text> and <tspan> elements. The <g> element which will be rendered underneath, apply the stroke, then <use> the text group of <text> element without a stroke and fill the font with contrasting color:

<g style="stroke:white; stroke-width:3px; stroke-linejoin:round">
  <text id="text_group" style="font-size:12px">
    <tspan x="10" y="80">Placeholder 1</tspan>
    <tspan x="10" y="95">Placeholder 2</tspan>
  </text>
</g>
<use xlink:href="#text_group" style="fill:black; stroke:none"/>

W3C includes paint-order property in SVG 2.0 which allows the order to render the attributes of "fill", "stroke" and "markers" within one element arranged by the user manually, instead of following the default mandatory order.^[12] Mozilla Firefox also includes this attribute experimentally but it defaults to disable in the stable release until SVG 2.0 matures. This feature can be enabled in the about:config entries of Firefox.^[13]

Gradients[edit]

Users of Inkscape version 0.46 may notice that images with "Fill gradients" display perfectly on Inkscape, but display "messed up" in Opera or Firefox browsers. One possible cause is that the last "stop" on a gradient has opacity set to 0. Inkscape appears to ignore this last stop, but both Firefox 3.0.6 and Opera 9.36 will render it as a white border in the filled object. Solutions are either remove the last stop (because it was probably created in error) or change the opacity (and colour) so that the image renders correctly in browsers.

Users of Adobe Illustrator CS5, CS6, or CC and who also use Firefox may notice that an Illustrator "symbol" containing a gradient will render normally as a Commons PNG file but if the SVG file itself is opened in Firefox these symbols will appear vacant. Other browsers will have no difficulty— Google Chrome, even Internet Explorer will render Illustrator's SVG symbol gradients correctly, but because of a known bug that the designers of Firefox have been aware of since 2005 (Mozilla's Bugzilla #376027 and other reports that lead to it such as #353575) but have so far failed to adequately fix, Firefox will display such symbols as empty shapes (symbol strokes and symbol paths without gradients are unaffected).

Adobe Illustrator users should also note that any Illustrator "symbols" containing paths with both a gradient and a filter will be rendered as a PNG in Commons only within the first instance of the use of that symbol. Subsequent use of a symbol containing these paths will render the paths (though not the entire symbol) as blank (neither gradient nor filter nor fill nor stroke will be applied).

Blurring[edit]

Comparison of gaussian blur filter effect between Google Chrome and Wikimedia. (Source SVG file)

Wikimedia's SVG renderer currently supports various filter effects, but some effects such as gaussian blur (feGaussianBlur) behave differently when the blurred object exceeds the edge of the actual image. The current version of RSVG will treat the cropped edge as the actual edge of the object to apply the effect (as if the object has applied clipPath before applying filter effect), creating an undesirable filter effect against the image edge.

Converting to SVG[edit]

Given an image file that is in a format other than SVG, there are various ways to convert it to SVG; see #Converters below for programs.

Note: For vector images where the vectorization was done by a third party, there is the issue where the vectorization code may have its own copyright even if the depicted image itself is out of copyright or below the threshold of originality. See #Copyright for more info.

Generally, there are 3 ways to produce an SVG file, given an existing image:

Regenerate it
Convert it
Create a new one

If you are unable to convert an image yourself, please add the tag {{convert to SVG}} to flag that the image should be converted. Some images are not well-suited to SVG, notably continuous-tone images such as photographs. These generally should instead be stored as JPG.

Regenerate it[edit]

If the image was generated by a program (rather than drawn by hand) — for example, a chart from a spreadsheet — it may be possible to set the output format to SVG, rather than PNG or other format, and thus produce an SVG image from the same source. If possible, this is generally easiest.

Convert it[edit]

Many vector file formats can be easily converted to SVG simply by changing the format (repackaging the existing data), by using a converter program. This may be as simple as opening the file in a vector image editor and then saving it in SVG.

For raster (bitmap) graphics, this cannot be done because the image is a set of individual pixel values, not a set of strokes (vectors). For these images (notably PNG images), one can convert to SVG via "vectorization" or "tracing" — automatically generating a vector graphic from a raster image.

Many SVG editors have tracing functions, and dedicated converters exist. In Inkscape, this is done by the potrace program.^[14]^[15] Concretely: open the bitmap (File → Open… (Ctrl-O)), then select Path → Trace Bitmap… (Shift-Alt-B). There are various options; after using, please delete the underlying bitmap, and consider simplifying the path to reduce the number of nodes (Ctrl-L).

Adobe Illustrator also allows for the conversion of bitmap images into vectors via its Image Trace function. Care must be taken when doing so that the resulting vector image is neither ponderously large (often the result of too much conversion fidelity with regard to color and shape) nor so simple that it loses faithfulness to the original image. Converted images almost always require some manual correction after conversion. As in Inkscape, the underlying bitmap should be removed after conversion is complete.

It should also be noted that when Illustrator converts one of its files into an SVG file, it tends to convert a number of shading effects such as "inner glow", "outer glow", and gradients applied along or across a path into undesirable raster images which it then embeds in the SVG image using links. Such embedded images can often be immediately detected by viewing the image in a web browser under high zoom where they will appear cloudy or have visibly jagged edges (neither of which is ever true of a correctly rendered vector image). As these effects were originally created without regard to the W3C consortium, Adobe does not offer and is never likely to have a solution to the problem.

In order to remove raster components from an Adobe Illustrator-generated SVG prior to uploading the image to Commons, the artist must first be able to identify where Illustrator has put them. This can be done in the following way: First finish editing an image file and save in Illustrator's AI format (*.AI), then save it again in SVG format. During the second Save dialogue, choose to "embed" the links (for now). Then close out the image (File=>Close or Ctrl+w). Next, open the SVG version back up again, click on "Window" from the right side of the top menu bar, and scroll down to "Links". This will open up a small window containing information about each of the links (if any) which currently exist in the SVG image (whether or not Illustrator produced them). You can click any of the entries in the top half of the window and then select the "Go to link" button located in about the middle of it-- this will take you to that particular raster so that you can identify it and either remove it or replace it with a non-raster based alternative. Note that this SVG version of your image, when viewed in Illustrator, may look very different from the same SVG file you thought you had saved just a few minutes prior: this is because Illustrator has made a set of codes during the creation of that file which it then has difficulty reading (SVG is not an Illustrator-proprietary format, AI is, and so while Illustrator will never have difficulty correctly viewing an AI (Adobe Illustrator) file, it will sometimes struggle to correctly render an SVG, even one that it itself produced). Fortunately, even an SVG that looks confusing in Illustrator will usually at least appear to render correctly in most browsers (with the occasional and already noted exception of Firefox) and in Inkscape.

Create a new one[edit]

If an image cannot be regenerated or converted, the remaining option is to recreate it — draw a new SVG image, using the existing image as a model. This is manual and the most time-consuming option, but may ultimately yield the highest quality.

In all cases, it may prove useful to touch up the resulting image in an SVG editor — perhaps adding details which are hard or impossible to produce in the originating program, or simplifying or refining a complex conversion (such as simplifying a path described by 100 points to one described by 5 points, if extra points add unnecessary detail).

Once you have created an SVG image, upload it (under a new name), and tag the original raster image with {{vector version available}}, but please do not delete the raster image or list it for deletion, as it still serves some purposes — see Commons:Superseded images policy for details.

Uploading & categorizing[edit]

Every SVG file uploaded to Wikimedia Commons should show

how it was made: use template {{Inkscape}}, {{Adobe}}, {{HandSVG}}, etc.
whether it is W3C valid or invalid: set the appropriate parameter to the template, e.g. {{Inkscape|v}}

You should check the appearance and the validity before you upload the file; use the tool SVGcheck

If you are not sure or want to see first how librsvg will render your file, upload it to Test.svg

Image description pages[edit]

Uploading is described at Commons:First steps/Upload form.

Make sure to fill out the {{information}} box as fully as you can. Often, it's a help when an image has a complete information box as this makes categorization by others easier and allows others to make translations.

If your file has passed the SVG Validator, you can put the {{validSVG}} tag on the image page.

See the #SVG software tags section for detailed information on a SVG media.

Categorizing SVG files[edit]

SVG

SVG graphics by software used

Vector images using elements from other vector images

► SVGs demonstrating filter effects

SVG diagrams

Examples representing SVG

► File description pages with SVG source

SVG files

► SVG galleries

Icons representing SVG files

► Interactive SVG

SVG marker templates

SVG media for cleanup

► SVG posters

SVG printing art

Scripted SVG

SVG Simplified

ß

All SVG graphics should be sorted into the appropriate subcategory or subcategories of Category:SVG. Please do not place images directly in the parent category as it will become overcrowded and useless (we have ~~thousands and thousands of~~ more than 100,000 SVG files).

Sort subcategories which rely not on subject but on technical aspects ("created with…", "animated", "valid", and so on) by '*' to keep them outside the alphabet listing.

The SVG categories tree[edit]

The main tree is Category:SVG, as a part of Category:Images → Category:Media types.

Concerning the creation of topical subcategories, it is unnecessary to rebuild the complete hierarchy tree of Category:Topics. Normally, it is enough to gather some topics into one. For instance, Category:SVG colors could contain all color-related SVG files of Category:Colors, so it's pointless to create Category:SVG red, Category:SVG blue and so on. Maybe some subcategories will become useful later, but stay close to the existing topics tree, and do not put any SVG files only into a SVG category: Categories are primarily for finding media, not for hiding them. Do not over-categorize.

See Commons:Categories for general information.

Category naming conventions[edit]

Category names start with "SVG" followed by the topic in lowercase (unless it's a proper noun). For example, a category containing SVG files related to chemistry would be called Category:SVG chemistry. There was no naming convention for a long time, thus you may see titles that do not match this format. These will be converted over time.

Tagging SVG files[edit]

See Category:SVG marker templates for all SVG markers available.

Transcluded from Commons:SVG marker templates