OptiPNG − Advanced optimization program for Portable Network Graphics (PNG)
optipng
[−? | −h | −help]
optipng [options...] files...
The OptiPNG program shall attempt to optimize PNG files, i.e. reduce their size to a minimum, without losing any information. In addition, this program shall perform a suite of auxiliary functions like integrity checks, metadata recovery and pixmap-to-PNG conversion.
The optimization attempts are not guaranteed to succeed. Valid PNG files that cannot be optimized by this program are normally left intact; their size will not grow. The user may request to override this default behavior.
The input files are raster image files encoded either in PNG format (the native format), or in an external format. The currently supported external formats are GIF, BMP, PNM and TIFF.
OptiPNG processes each image file given in the command line as follows:
− If the image is in PNG format:
Attempts to optimize the given file in-place. If optimization is successful, or if the option −force is in effect, replaces the original file with its optimized version. The original file is backed up if the option −keep is in effect.
− If the image is in an external format:
Creates an optimized PNG version of the given file. The output file name is composed from the original file name and the .png extension.
General
options
-?, −h, −help
Show a complete summary of options.
−o level
Select the optimization level.
The optimization level 0 enables a set of optimization
operations that require minimal effort. There will be no
changes to image attributes like bit depth or color type,
and no recompression of existing IDAT datastreams.
The optimization level 1 enables a single IDAT compression
trial. The trial chosen is what OptiPNG thinks
it’s probably the most effective.
The optimization levels 2 and higher enable multiple IDAT
compression trials; the higher the level, the more trials.
The behavior and the default value of this option may change
across different program versions. Use the option
−h to see the details pertaining to your
specific version.
−dir directory
Write output file(s) to directory.
|
−fix |
Enable error recovery. This option has no effect on valid input files. |
The program will spend a
reasonable amount of effort to recover as much data as
possible, without increasing the output file size, but the
success cannot be generally guaranteed. The program may even
increase the file size, e.g., by reconstructing missing
critical data. Under this option, integrity shall take
precedence over file size.
When this option is not used, the invalid input files are
left unprocessed.
|
−force |
Enforce writing of a new output file. |
This option overrides the program’s decision not to write such file, e.g. when the PNG input is digitally signed (using dSIG), or when the PNG output becomes larger than the PNG input.
|
−keep |
Keep a backup of the modified file(s). |
The files that use the backup names prior to the program execution are not overwritten.
−log file
Log messages to file. For safety reasons, file must have the extension .log.
−out file
Write output file to file. The command line must contain exactly one input file.
−preserve
Preserve file attributes (time stamps, file access rights, etc.) where applicable.
|
−quiet |
Run in quiet mode. |
These messages are still written to the log file if the option −log is in effect.
−simulate
Run in simulation mode: perform the trials, but do not create output files.
|
−snip |
Cut one image out of multi-image, animation or video file(s). |
Depending on the input format, this may be either the first or the most relevant (e.g. the largest) image.
|
−v |
Enable the options −verbose and −version. |
−verbose
Run in verbose mode.
−version
Show copyright, version and build info.
|
−− |
Stop option switch parsing. |
PNG encoding
and optimization options
−f filters
Select the PNG delta filters.
The filters argument is specified as a rangeset (e.g.
−f0−5), and the default filters
value depends on the optimization level set by the option
−o.
The filter values 0, 1, 2, 3 and 4 indicate static
filtering, and correspond to the standard PNG filter codes
(None, Left, Up, Average and
Paeth, respectively). The filter value 5 indicates
adaptive filtering, whose effect is defined by the
libpng(3) library used by OptiPNG.
|
−full |
Produce a full report on IDAT. This option might slow down the trials. |
−i type
Select the interlace type
(0−1).
If the interlace type 0 is selected, the output image shall
be non-interlaced (i.e. progressive-scanned). If the
interlace type 1 is selected, the output image shall be
interlaced using the Adam7 method.
By default, the output shall have the same interlace type as
the input.
|
−nb |
Do not apply bit depth reduction. | ||
|
−nc |
Do not apply color type reduction. | ||
|
−np |
Do not apply palette reduction. | ||
|
−nx |
Do not apply any lossless image reduction: enable the options −nb, −nc and −np. | ||
|
−nz |
Do not recode IDAT datastreams. |
The IDAT optimization
operations that do not require recoding (e.g. IDAT chunk
concatenation) are still performed.
This option has effect on PNG input files only.
−zc levels
Select the zlib compression
levels used in IDAT compression.
The levels argument is specified as a rangeset (e.g.
−zc6−9), and the default levels
value depends on the optimization level set by the option
−o.
The effect of this option is defined by the zlib(3)
library used by OptiPNG.
−zm levels
Select the zlib memory levels
used in IDAT compression.
The levels argument is specified as a rangeset (e.g.
-zm8−9), and the default levels value
depends on the optimization level set by the option
−o.
The effect of this option is defined by the zlib(3)
library used by OptiPNG.
−zs strategies
Select the zlib compression
strategies used in IDAT compression.
The strategies argument is specified as a rangeset
(e.g. -zs0−3), and the default
strategies value depends on the optimization level
set by the option −o.
The effect of this option is defined by the zlib(3)
library used by OptiPNG.
−zw size
Select the zlib window size
(32k,16k,8k,4k,2k,1k,512,256) used in IDAT compression.
The size argument can be specified either in bytes
(e.g. 16384) or kilobytes (e.g. 16k). The default
size value is set to the lowest window size that
yields an IDAT output as big as if yielded by the value
32768.
The effect of this option is defined by the zlib(3)
library used by OptiPNG.
Notes
Options may come in any order (except for
−−), before, after, or alternating with
file names. Option names are case-insensitive and may be
abbreviated to their shortest unique prefix.
Some options may have arguments that follow the option name, separated by whitespace or the equal sign (’=’). If the option argument is a number or a rangeset, the separator may be omitted. For example:
−out
newfile.png <=>
−out=newfile.png
−o3 <=> −o 3 <=>
−o=3
−f0,3−5 <=>
−f 0,3−5 <=>
−f=0,3−5
Rangeset arguments are cumulative; e.g.
−f0
−f3−5 <=> −f0,3−5
−zs0 −zs1 −zs2−3 <=>
−zs0,1,2,3 <=>
−zs0−3
The PNG optimization algorithm consists of the following steps:
|
1. |
Reduce the bit depth, the color type and the color palette of the image. This step may reduce the size of the uncompressed image, which, indirectly, may reduce the size of the compressed image (i.e. the size of the output PNG file). | ||
|
2. |
Run a suite of compression methods and strategies and select the compression parameters that yield the smallest output file. | ||
|
3. |
Store all IDAT contents into a single chunk, eliminating the overhead incurred by repeated IDAT headers and CRCs. | ||
|
4. |
Set the zlib window size inside IDAT to a mininum that does not affect the compression ratio, reducing the memory requirements of PNG decoders. |
Not all of the above steps need to be executed. The behavior depends on the actual input files and user options.
Step 1 may be customized via the no-reduce options −nb, −nc, −np and −nx. Step 2 may be customized via the −o option, and may be fine-tuned via the options −zc, −zm, −zs and −zw. Step 3 is always executed. Step 4 is executed only if a new IDAT is being created, and may be fine-tuned via the option −zw.
Extremely exhaustive searches are not generally expected to yield significant improvements in compression ratio, and are recommended to advanced users only.
optipng file1.png file2.gif file3.tif
optipng −o5 file1.png file2.gif file3.tif
optipng −i1 −o7 −v −full −sim experiment.png
Lossless image reductions are not completely implemented. (This does NOT affect the integrity of the output files.) Here are the missing pieces:
− The
color palette reductions are implemented only partially.
− The bit depth reductions below 8, for grayscale
images, are not implemented yet.
TIFF support is limited to uncompressed, PNG-compatible (grayscale, RGB and RGBA) images.
Metadata is not imported from the external image formats.
There is no support for pipes or streams.
png(5), libpng(3), zlib(3), pngcrush(1), pngrewrite(1).
The files
produced by OptiPNG are compliant with
PNG−2003:
Glenn Randers-Pehrson et al. Portable Network Graphics
(PNG) Specification, Second Edition.
W3C Recommendation 10 November 2003; ISO/IEC IS 15948:2003
(E).
http://www.w3.org/TR/PNG/
OptiPNG is written and maintained by Cosmin Truta.
This manual page was originally written by Nelson A. de Oliveira for the Debian Project. It was later updated by Cosmin Truta, and is now part of the OptiPNG distribution.