Appendix J: Minimizing SVG File Sizes
This appendix is informative, not normative.
Considerable effort has been made to make SVG file sizes as
small as possible while still retaining the benefits of XML and
achieving compatibility and leverage with other W3C
specifications.
Here are some of the features in SVG that promote small file
sizes:
- SVG's path data definition was defined to produce a
compact data stream for vector graphics data: all commands
are one character in length; relative coordinates are
available; separator characters do not have to be supplied
when tokens can be identified implicitly; smooth curve
formulations are available (cubic Béziers, quadratic
Béziers and elliptical arcs) to prevent the need to
tesselate into polylines; and shortcut formulations exist for
common forms of cubic Bézier segments, quadratic
Bézier segments, and horizontal and vertical straight
line segments so that the minimum number of coordinates need
to be specified.
- Text can be specified using XML character data — no need
to convert to outlines.
- SVG contains a facility for defining symbols once and
referencing them multiple times using different visual
attributes and different sizing, positioning, clipping and
client-side filter effects
- User agents that support styling with CSS can
use CSS selectors and property inheritance to define commonly
used sets of attributes once as named styles.
- Filter effects allow for compelling visual results and
effects typically found only in image-authoring tools using
small amounts of vector and/or raster data
Additionally, HTTP/1.1 allows for compressed data to be
passed from server to client, which can result in significant
file size reduction. Here are some sample compression results
using gzip
compression on SVG documents [RFC1952]:
Uncompressed
SVG | With gzip
compression | Compression
ratio |
---|
12,912 | 2,463 | 81% |
12,164 | 2,553 | 79% |
11,613 | 2,617 | 77% |
18,689 | 4,077 | 78% |
13,024 | 2,041 | 84% |
A related issue is progressive rendering. Some SVG viewers
will support:
- the ability to display the first parts of an SVG document
fragments as the remainder of the document is downloaded from
the server; thus, the user will see part of the SVG drawing
right away and interact with it, even if the SVG file size is
large.
- delayed downloading of images and fonts. Just like some
HTML browsers, some SVG viewers will download images and
WebFonts
([CSS2], section 15.1)
last, substituting a temporary image and system fonts,
respectively, until the given image and/or font is
available.
Here are techniques for minimizing SVG file sizes and
minimizing the time before the user is able to start
interacting with the SVG document fragments:
- Construct the SVG file such that any links which the user
might want to click on are included at the beginning of the
SVG file
- Use default values whenever possible rather than defining
all attributes and properties explicitly.
- Take advantage of the path
data data compaction facilities: use relative
coordinates; use h and v for horizontal and
vertical lines; use s or t for cubic and
quadratic Bézier segments whenever possible; eliminate
extraneous white space and separators.
- Utilize symbols if the same graphic appears multiple
times in the document
- For user agents that support styling with CSS,
utilize CSS property inheritance and selectors to consolidate
commonly used properties into named styles or to assign the
properties to a parent ‘g’ element.
- Utilize filter effects to help construct graphics via
client-side graphics operations.