potrace(1)potrace(1)NAMEpotrace - transform bitmaps into vector graphics.
SYNOPSISpotrace [options] [filename...]
DESCRIPTIONpotrace is a utility for tracing a bitmap, which means, transforming a
bitmap into a smooth, scalable image. The input is a bitmap, which
means, a pixel-based image composed of the two colors black and white
only. The default output is an encapsulated PostScript file (EPS). A
typical use is to create EPS files from scanned data, such as company
or university logos, handwritten notes, etc. The resulting image is not
"jaggy" like a bitmap, but smooth. It can then be rendered at any reso‐
lution.
potrace can read bitmaps in the following formats: PBM, PGM, PPM (col‐
lectively known as PNM, see pnm(5)), as well as BMP (Windows and OS/2
bitmap formats). The input image should only use the two colors black
and white. If other pixel values appear in the input, they will be con‐
verted to black and white using a simple threshold method.
potrace can currently produce the following output formats: EPS, Post‐
Script, PDF, SVG, DXF, PGM, Gimppath, and XFig. Additional backends
might be added in the future.
OPTIONS
The following options are supported. Dimensions (arguments of type dim)
can have optional units, e.g. 6.5in, 15cm, 100pt. The default unit is
inches (or centimeters, if this was configured at compile time, see
COMPILE TIME CONFIGURATION below). For pixel-based output formats such
as PGM, DXF, and Gimppath, the default unit is pixels.
General options:
-h, --help print help message and exit.
-v, --version print version info and exit. This also shows the
defaults that were compiled into this version of
potrace.
-l, --license print license info and exit.
Input/output options:
filename Each file can hold an input image, or multiple concate‐
nated input images. If filename arguments are given,
then potrace will by default create one output file for
each input filename given. The name of the output file
is obtained from the input filename by changing its suf‐
fix according to the chosen backend. If changing the
suffix is impossible because the names of the input and
output files would be identical, then the output file‐
name is created by adding the "-out" suffix to the name
of the input file. If no filename arguments are given,
then potrace acts as a filter, reading from standard
input and writing to standard output. A filename of "-"
may be given to specify reading from standard input.
-o filename, --output filename
write output to this file. All output is directed to the
specified file. If this option is used, then multiple
input filenames are only allowed for multi-page backends
(see BACKEND TYPES below). In this case, each input file
may contain one or more bitmaps, and all the bitmaps
from all the input files are processed and the output
concatenated into a single file. A filename of "-" may
be given to specify writing to standard output.
-- End of options. Any remaining arguments are interpreted
as filenames. This also disables filter mode, even if no
filenames are given. This is useful for shell scripts,
because potrace-- $FILENAMES will behave correctly even
for an empty list of filenames. However, -- with an
empty list of filenames is not permitted in conjunction
with the -o option, because this would generate a docu‐
ment of zero pages, which none of the backends permit.
Backend selection:
For general information, see also BACKEND TYPES below.
-b name, --backend name
Select backend by name, where name is one of eps, post‐
script, ps, pdf, pdfpage, svg, dxf, pgm, gimppath, xfig.
Backend names can be abbreviated by a prefix as long as
it is unambiguous. Backend names are case insensitive.
-e, --eps, -b eps, --backend eps
EPS backend (default). The output is an encapsulated
PostScript file. This is a single-page, variable-sized
backend.
-p, --postscript, -b ps, --backend ps
PostScript backend. The output is a PostScript file.
This is a multi-page, fixed-size backend. If the input
consists of multiple bitmaps, they are each rendered on
a separate page.
-b pdf, --backend pdf
PDF backend. The output is a file in the Portable Docu‐
ment Format. If the input consists of multiple bitmaps,
they are each rendered on a separate page. This is a
multi-page, variable-sized, backend.
-b pdfpage, --backend pdfpage
The PDFPage backend is like the PDF backend, except that
it is fixed-size like the PostScript backend.
-s, --svg, -b svg, --backend svg
SVG backend. The output is a Scalable Vector Graphics
(SVG) file. This is a single-page, variable-sized back‐
end.
-b dxf, --backend dxf
DXF backend. The output is a file in the Drawing Inter‐
change Format (DXF). In this backend, all Bezier curves
are approximated by piecewise circular arcs; this is
suitable for processing in CAD software or for machining
applications using CNC tools. This is a single-page,
variable-sized, pixel-based backend. The -u option has
no effect for this backend.
-g, --pgm, -b pgm, --backend pgm
PGM backend. The output is a portable greymap (PGM)
file. It is a convenient backend for antialiasing a bit‐
map image. This is a multi-page, variable-sized, pixel-
based backend. If the input consists of more than one
image, the images are concatenated in the output.
-b gimppath, --backend gimppath
Gimppath backend. This backend produces output suitable
to be imported as a path by the GNU Image Manipulation
Program (Gimp) (in the Layers, Channels & Paths dialog,
select Paths, then right-click and select Import Path).
The output is actually an SVG file. The differences to
the SVG backend are: the --opaque option has no effect,
the --flat option is always on, and the dimensions are
pixel-based. This is a single-page, variable-sized,
pixel-based backend.
-b xfig, --backend xfig
XFig backend. The output is a file in the XFig format.
Note that XFig uses X-splines instead of Bezier curves,
thus it is not possible to translate the output of
potrace into the XFig format with absolute accuracy.
This backend does a reasonable approximation using two
control points for each Bezier curve segment. The -u
option has no effect for this backend, because control
points are always rounded to the nearest 1/1200 of an
inch in XFig. Curve optimization is disabled. Implies
--opaque.
Algorithm options:
For more detailed information on these options, see TECHNICAL DOCUMEN‐
TATION below.
-z policy, --turnpolicy policy
specify how to resolve ambiguities in path decomposi‐
tion. Must be one of black, white, right, left, minor‐
ity, majority, or random. Default is minority. Turn
policies can be abbreviated by an unambigous prefix,
e.g., one can specify min instead of minority.
-t n, --turdsize n
suppress speckles of up to this many pixels.
-a n, --alphamax n
set the corner threshold parameter. The default value is
1. The smaller this value, the more sharp corners will
be produced. If this parameter is negative, then no
smoothing will be performed and the output is a polygon.
The largest useful value is 4/3 or 1.334, which sup‐
presses all corners and leads to completely smooth out‐
put.
-n, --longcurve
turn off curve optimization. Normally potrace tries to
join adjacent Bezier curve segments when this is possi‐
ble. This option disables this behavior, resulting in a
larger file size.
-O n, --opttolerance n
set the curve optimization tolerance. The default value
is 0.2. Larger values allow more consecutive Bezier
curve segments to be joined together in a single seg‐
ment, at the expense of accuracy.
-u n, --unit n set output quantization. Coordinates in the output are
rounded to 1/unit pixels. The default of 10 usually
gives good results. For some of the debug modes, a value
of 100 gives more accurate output. This option has no
effect for the XFig backend, which always rasterizes to
1/1200 inch, or for the DXF backend.
-d n, --debug n
produce debugging output of type n. This has different
effects for different backends. For the PostScript/EPS
backends, the values n=1,2,3 illustrate the intermediate
stages of the potrace algorithm.
Scaling and placement options:
-P format, --pagesize format
for fixed-size backends, set page size. The following
formats can be specified: A4, A3, A5, B5, Letter, Legal,
Tabloid, Statement, Executive, Folio, Quarto, 10x14.
Format names are case insensitive. Also, an argument of
the form dimxdim is accepted to specify arbitrary dimen‐
sions. The default page size is Letter (or A4, if this
was configured at compile time, see COMPILE TIME CONFIG‐
URATION below). Page format names can be abbreviated by
a prefix as long as it is unambiguous. This option has
no effect for variable-sized backends.
-W dim, --width dim
set the width of output image (before any rotation and
margins). If only one of width and height is specified,
the other is adjusted accordingly so that the aspect
ratio is preserved.
-H dim, --height dim
set the height of output image. See -W for details.
-r n[xn], --resolution n[xn]
for dimension-based backends, set the resolution (in
dpi). One inch in the output image corresponds to this
many pixels in the input. Note that a larger value
results in a smaller output image. It is possible to
specify separate resolutions in the x and y directions
by giving an argument of the form nxn. For variable-
sized backends, the default resolution is 72dpi. For
fixed-size backends, there is no default resolution; the
image is by default scaled to fit on the page. This
option has no effect for pixel-based backends. If -W or
-H are specified, they take precedence.
-x n[xn], --scale n[xn]
for pixel-based backends, set the scaling factor. A
value greater than 1 enlarges the output, a value
between 0 and 1 makes the output smaller. The default is
1. It is possible to specify separate scaling factors
for the x and y directions by giving an argument of the
form nxn. This option has no effect for dimension-based
backends. If -W or -H are specified, they take prece‐
dence.
-S n, --stretch n
set the aspect ratio. A value greater than 1 means the
image will be stretched in the y direction. A value
between 0 and 1 means the image will be compressed in
the y direction.
-A angle, --rotate angle
set the rotation angle (in degrees). The output will be
rotated counterclockwise by this angle. This is useful
for compensating for images that were scanned not quite
upright.
-M dim, --margin dim
set all four margins. The effect and default value of
this option depend on the backend. For variable-sized
backends, the margins will simply be added around the
output image (or subtracted, in case of negative mar‐
gins). The default margin for these backends is 0. For
fixed-size backends, the margin settings can be used to
control the placement of the image on the page. If only
one of the left and right margin is given, the image
will be placed this distance from the respective edge of
the page, and similarly for top and bottom. If margins
are given on opposite sides, the image is scaled to fit
between these margins, unless the scaling is already
determined explicitly by one or more of the -W, -H, -r,
or -x options. By default, fixed-size backends use a
non-zero margin whose width depends on the page size.
-L dim, --leftmargin dim
set the left margin. See -M for details.
-R dim, --rightmargin dim
set the right margin. See -M for details.
-T dim, --topmargin dim
set the top margin. See -M for details.
-B dim, --bottommargin dim
set the bottom margin. See -M for details.
--tight remove whitespace around the image before scaling and
margins are applied. If this option is given, calcula‐
tions of the width, height, and margins are based on the
actual vector outline, rather than on the outer dimen‐
sions of the input pixmap, which is the default. In par‐
ticular, the --tight option can be used to remove any
existing margins from the input image. See the file
placement.pdf for a more detailed illustration.
Color options:
These options are only supported by certain backends. The DXF backend
does not support color.
-C #rrggbb, --color #rrggbb
set the foreground color of the output image. The
default is black.
--fillcolor #rrggbb
set the fill color of the output image, i.e., the color
of the "white" parts. The default is to leave these
parts transparent. Implies --opaque.
--opaque fill in the white parts of the image opaquely, instead
of leaving them transparent. This only applies to inte‐
rior white parts, i.e., those that are enclosed inside a
black outline. Opaqueness is always in effect for the
XFig backend.
SVG options:
--group for SVG output, try to group related paths together.
Each path is grouped together with all paths that are
contained inside it, so that they can be moved around as
a unit with an SVG editor. This makes coloring individ‐
ual components slightly more cumbersome, and thus it is
not the default.
--flat for SVG output, put the entire image into a single path.
This makes it impossible to color the components indi‐
vidually, and thus it is not the default. But the
resulting SVG file can be more easily imported by some
applications such as Gimp. In fact, the Gimppath backend
is a variation of the SVG backend with the --flat option
and pixel-based scaling. The --flat option has no effect
if --opaque has been selected.
PostScript/EPS/PDF options:
-c, --cleartext
do not compress the output. This option disables the use
of compression filters in the PostScript and PDF output.
In the PostScript backend, if -c and -q are used
together, the resulting output can be easily read by
other programs or even by humans.
-2, --level2 use PostScript level 2 compression (default). The
resulting file size is ca. 40% smaller than if the -c
option is used.
-3, --level3 use PostScript level 3 compression, if available. This
gives slightly smaller files than using -2, but the
resulting files may not print on older PostScript level
2 printers. If support for PostScript level 3 compres‐
sion has been disabled at compile time, a warning mes‐
sage is printed and level 2 compression is used instead.
-q, --longcoding
turn off optimized numerical coding in PostScript out‐
put. Normally, potrace uses a very compact numerical
format to represent Bezier curves in PostScript, taking
advantage of existing redundancy in the curve parame‐
ters. This option disables this behavior, resulting in
longer, but more readable output (particularly if the -c
option is also used).
PGM options:
-G n, --gamma n
set the gamma value for anti-aliasing (default is 2.2).
Most computer displays do not render shades of grey lin‐
early, i.e., a grey value of 0.5 is not displayed as
being exactly half-way between black and white. The
gamma parameter corrects for this, and therefore leads
to nicer looking output. The default value of 2.2 is
appropriate for most normal CRT displays.
Frontend options:
-k n, --blacklevel n
set the threshold level for converting input images to
bitmaps. The potrace algorithm expects a bitmap, thus
all pixels of the input images are converted to black or
white before processing begins. Pixels whose brightness
is less than n are converted to black, all other pixels
to white. Here n is a number between 0 and 1. One case
is treated specially: if the input is in an indexed
color format with exactly 2 colors, then the blacklevel
is ignored and the darker of the two colors is mapped to
black.
Note: the method used by potrace for converting greymaps
to bitmaps is very crude; much better results can be
obtained if a separate program, such as mkbitmap(1), is
used for this purpose. In particular, mkbitmap(1), which
is distributed with potrace, has the ability to scale
and interpolate the image before thresholding, which
results in much better preservation of detail.
-i, --invert invert the input bitmap before processing.
Progress bar options:
--progress display a progress bar for each bitmap that is pro‐
cessed. This is useful for interactive use. The default
behavior is not to show any progress information.
--tty mode set the terminal mode for progress bar rendering. Possi‐
ble values are "vt100", which requires a vt100-compati‐
ble terminal, and "dumb", which uses only ASCII charac‐
ters. The default is system dependent.
BACKEND TYPES
Backends can be classified in several ways, which affects the available
command line options and their behavior:
Fixed-size or variable-sized:
For fixed-size backends, the size of the page is always the same
(for example Letter or A4, as specified at compile time or by the
-P option). By default, the image will be centered and scaled to
fit the page size. For variable-size backends, the size of the
page follows the size of the image. Currently the PostScript (PS),
PDFPage, and XFig backends are fixed-size, and the remaining back‐
ends are variable-size.
Dimension-based or pixel-based:
In dimension-based backends, distances are measured in physical
units such as inches or centimeters. In pixel-based backends, dis‐
tances are measured in pixel units. The -r option only works for
dimension-based backends, and the -x option only works for pixel-
based backends. Currently, the DXF, PGM, and Gimppath backends are
pixel-based, and the remaining backends are dimension-based. Cur‐
rently, all pixel-based backends are variable-sized.
Single-page or multi-page:
Single-page backends can only accept a single image. Multi-page
backends can accept multiple images, typically one per page of
output. Currently, the PostScript (PS), PDF, PDFPage, and PGM
backends are multi-page, and the remaining backends are single-
page. Note that multiple input images can be read in two ways:
from multiple input files (with the -o option), or from a single
input file that holds several concatenated images.
COMPILE TIME CONFIGURATION
Certain aspects of the behavior of potrace can be configured at compile
time by passing the following options to the ./configure script.
--disable-zlib
compile potrace without the zlib compression library. This means
PostScript level 3 compression will not be available.
--enable-metric
compile potrace with centimeters as the default unit instead of
inches.
--enable-a4
compile potrace with A4 as the default page size.
EXIT STATUS
The exit status is 0 on successful completion, 1 if the command line
was invalid, and 2 on any other error.
VERSION
1.10
AUTHOR
Peter Selinger <selinger at users.sourceforge.net>
TECHNICAL DOCUMENTATION
For a detailed technical description of the potrace algorithm, see the
file potrace.pdf, which is available from the potrace web site. For
information on the Potrace library API, see potracelib.pdf.
WEB SITE AND SUPPORT
The latest version of potrace is available from http://potrace.source‐
forge.net/. This site also contains a list of frequently asked ques‐
tions, as well as information on how to obtain support.
SEE ALSOmkbitmap(1)COPYRIGHT
Copyright (C) 2001-2011 Peter Selinger
This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2 of the License, or (at your
option) any later version.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of MER‐
CHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
59 Temple Place, Suite 330, Boston, MA 02111-1307, USA. See also
http://www.gnu.org/.
Version 1.10 August 2011 potrace(1)
