![]() | ![]() | ![]() | ![]() |
VIPS from the command-lineUsing VIPS — How to use the VIPS library from the command-line |
Use thevips command to execute VIPS operations from the command-line. For example:
$ vips rot k2.jpg x.jpg d90
Will rotate the imagek2.jpg
by 90 degrees anticlockwise and write the result to the filex.jpg
. If you don't give any arguments to an operation,vips will give a short description, for example:
$ vips rot rotate an imageusage: rot in out anglewhere: in - Input image, input VipsImage out - Output image, output VipsImage angle - Angle to rotate image, input VipsAngle default: d90 allowed: d0, d90, d180, d270
There's a straightforward relationship with the C API: compare this to the API docs forvips_rot()
.
You can list all classes with:
$ vips -l... VipsOperation (operation), operations VipsSystem (system), run an external command VipsArithmetic (arithmetic), arithmetic operations VipsBinary (binary), binary operations VipsAdd (add), add two images ... etc.
Each line shows the canonical name of the class (for exampleVipsAdd
), the class nickname (add
in this case), and a short description. Some subclasses of operation will show more: for example, subclasses ofVipsForeign
will show some of the extra flags supported by the file load/save operations.
The API docs have ahandy table of all vips operations, if you want to find out how to do something, try searching that.
Many operations take optional arguments. You can supply these as command-line options. For example:
$ vips gammagamma an imageusage: gamma in outwhere: in - Input image, input VipsImage out - Output image, output VipsImageoptional arguments: exponent - Gamma factor, input gdouble default: 2.4 min: 1e-06, max: 1000operation flags: sequential-unbuffered
vips_gamma()
applies a gamma factor to an image. By default, it uses 2.4, the sRGB gamma factor, but you can specify any gamma with theexponent
option.
Use it from the command-line like this:
$ vips gamma k2.jpg x.jpg --exponent 0.42
This will read filek2.jpg
, un-gamma it, and write the result to filex.jpg
.
Some operations take arrays of values as arguments. For example,vips_affine()
needs an array of four numbers for the 2x2 transform matrix. You pass arrays as space-separated lists:
$ vips affine k2.jpg x.jpg "2 0 0 1"
You may need the quotes to stop your shell breaking the argument at the spaces.vips_bandjoin()
needs an array of input images to join, run it like this:
$ vips bandjoin "k2.jpg k4.jpg" x.tif
vips will automatically convert between image file formats for you. Input images are detected by sniffing their first few bytes; output formats are set from the filename suffix. You can see a list of all the supported file formats with something like:
$ vips -l foreign
Then get a list of the options a format supports with:
$ vips jpegsave
You can pass options to the implicit load and save operations enclosed in square brackets after the filename:
vips affine k2.jpg x.jpg[Q=90,strip] "2 0 0 1"
Will writex.jpg
at quality level 90 and will strip all metadata from the image.
Because each operation runs in a separate process, you can't use libvips's chaining system to join operations together, you have to use intermediate files. The command-line interface is therefore quite a bit slower than Python or C.
The best alternative is to use vips files for intermediates. Something like:
vips invert input.jpg t1.vvips affine t1.v output.jpg "2 0 0 1"rm t1.v
Finally,vips has a couple of useful extra options.
Use--vips-progress
to getvips to display a simple progress indicator.
Use--vips-leak
andvips will leak-test on exit, and also display an estimate of peak memory use.
SetG_MESSAGES_DEBUG=VIPS
and GLib will display informational and debug messages from libvips.
VIPS comes with a couple of other useful programs.vipsheader is a command which can print image header fields.vipsedit can change fields in vips format images.vipsthumbnail can make image thumbnails quickly.