optisorl
Copyright Peter Bengtsson, mail@peterbe.com, 2015-2016
License: BSD
About optisorl
sorl-thumbnail is a
great Django library that takes your stored images and automatically
convert them into desired sized thumbnails and store them with nice
names in the MEDIA_ROOT
. The problem is that the engines that do
the resizing often doesn't do an amazing job of optimizing them.
Usually optimizing an image means carefully deleting things the human
eye can't notice anyway. This becomes incredibly relevant when the
thumbnail you create is so small in resolution that the user really
stands very little chance to notice.
This package, is a pluggable backend to sorl-thumbnail
that
attempts to do a good job of optimizing the generated thumbnail just
right after it has been written to disk.
Installation
First, simply pip install optisorl.
Then add, in your Django settings:
THUMBNAIL_BACKEND = 'optisorl.backend.OptimizingThumbnailBackend'
Then review the sections below about being prepared for PNGs, GIFs and JPEGs.
Optimizing PNGs
optisorl
uses a binary called pngquant
which is a command line tool that do lossy compression of PNG images
and supports alpha transparency. pngquant
is
BSD licensed. It's easy to install on most systems. For example
brew install pngquant
or apt-get install pngquant
.
What happens is that when optisorl
notices that a thumbnail was
created it (and stored in MEDIA_ROOT
) it then takes that file and
executes pngquant
something like this:
pngquant -o /path/file.tmp.png --skip-if-larger -- /path/file.png
Note the --skip-if-larger
which means that if the thumbnail is really
really small already the resulting optimization might not be any better
and it thus omits doing an optimization.
If you want to override the location of the executable pngquant
you
can set this setting for example:
# in settings.py or equivalent
PNGQUANT_LOCATION = '/opt/special/bin/pngquant2.0'
Optimizing GIFs
optisorl
uses gifsicle with
level 3 optimization. gifsicle
is GPL licensed but use is not
restricted by a license. To install it use brew install gifsicle
or apt-get install gifsicle
.
To override where the gifsicle
executable is located you can set
in your settings:
# in settings.py or equivalent
GIFSICLE_LOCATION = '/opt/special/bin/gifsicle'
If you want to disable all optimization of GIFs just set
GIFSICLE_LOCATION
(in your settings.py
) to None
or False
.
Optimizing JPEGs
optisorl
uses mozjpeg to
optimize JPEGs. It's a great fit because it almost never reduces
the quality such that human eyes can notice it. Especially when the
thumbnails are relatively small. The command that we use to execute
mozjpeg
looks like this:
mozjpeg -outfile DESTINATION -optimise SOURCE
You can override where the executable is by setting:
# in settings.py or equivalent
MOZJPEG_LOCATION = '/my/bin/mozjpeg'
For an example of what kind of results you can get with mozjpeg
see this blog post:
Examples of mozjpeg savings.
Also see blog post mozjpeg installation and sample.
Limitations
Help is most welcome. At the moment...
- Does not support S3 storage
- Unable to NOT optimize images in run-time
- Not possible to override certain
pngquant
parameters - Not possible to override certain
gifsicle
parameters - Not possible to override certain
mozjpeg
parameters