The -cc option should usually not be used when writing to a lossy image
format (JPEG, WebP). Lossy formats do not store colors precisely enough,
so the resulting image will have more colors than requested.

If you use -cc expecting to get a PNG image in a particular format, be aware
that background color labels are not affected by -cc, and can prevent some
image format optimizations from occurring. For example, "-grayscale -cc 2"
will not always produce a 1-bpp grayscale image, unless you also use
-nobkgdlabel.

The -ccX options let you have a different setting for different channels.
If you specify overlapping options, the most specific option will have
priority.
-cccolor affects all channels except the alpha channel.
-ccalpha affects only the alpha channel. Use "-ccalpha 2" for binary
transparency.
-ccgray applies only if you force grayscale output, using "-grayscale".

-dither (-dithercolor -ditheralpha -ditherred -dithergreen
-ditherblue -dithergray)
Enable dithering.

Dither types available:
"none": No dithering.
"f" or "fs": Floyd-Steinberg.
"o": 8x8 Ordered dither.
"halftone": A sample 8x8 halftone dither.
"r": Random dither. See also the -randseed option.
"r2": "Random2" dither - Same as Random, except that all color channels
(but not alpha) use the same random pattern. The colors will be more
consistent than with Random, but the image will be grainier.
"sierra" or "sierra3"
"sierra2"
"sierralite"
"jjn": Jarvis, Judice, and Ninke.
"burkes"
"atkinson"

You can enable dithering on any image, no matter how many colors it uses.
However, if lots of colors are available, the dithering effect will be
invisible to the human eye (unless you use a paint program to turn the
image's contrast *way* up). Normally, if you enable dithering, you should
also use one or more of the "-cc" options to limit the number of colors
available.

If you are applying a background to an image with transparency, dithering
is only done after the background has been applied, so -ditheralpha has no
effect. There's no way to dither the alpha channel and then apply a
background, unless you invoke IW multiple times using an intermediate image
file.

-cs
The colorspace to use for the output image. Ideally, the image should will
*look* about the same regardless of what colorspace you choose. But it
might not if your viewer doesn't support color correction, or the image
format doesn't support colorspace labels, or IW doesn't know how to write
an appropriate label, or you use the -negate option. This is fairly safe to
use when writing PNG files, but should be used with caution with most other
formats.

= linear | gamma | srgb | srgb[prsa] | rec709

= The gamma value. This is an "image gamma", like 2.2 (not a "file
gamma" like 0.4545"). "g1.0" is the same as "linear".

"srgb": sRGB colorspace. Can optionally be followed by a letter indicating
the "rendering intent" (equivalent to using the -intent option):
p=perceptual, r=relative, s=saturation, a=absolute.

"rec709": The color response curve defined by ITU-R Recommendation BT.709
(Rec. 709).

By default IW will choose a colorspace that works with the output format;
almost always sRGB.

-intent
Specify the color profile "rendering intent" setting to be written to the
output file. This does not affect image processing.

By default, the rendering intent of the input image will be used. If that's
not possible, a default setting will be used that may depend on the output
image format.

Rendering intents available:
"p" or "perceptual"
"r" or "relative"
"s" or "saturation"
"a" or "absolute"
"default": Ignore the intent of the input image.
"none": Request that no rendering intent be written.

-inputcs
Assume the input image is in the given colorspace. This is intended to be
used in case IW cannot correctly figure out the colorspace automatically.
Although it can be used to apply a gamma-correction operation to the image,
that's not really what it's intended for.

-nogamma
Disable all color correction.
The main purpose of this option is to let you try to replicate what some
other applications might do.
Note that there are many *wrong* ways to handle color correction, and this
only recreates one of them. For example, some apps might do gamma correction
when the image is read in, but not when it is resized. That's not what
-nogamma will do.

If you use this with "-grayscale", you should probably also use "-gsf c".

This does not affect the colorspace label that will be written to the output
file. If a label is written, it may not be the label you want.

-nocslabel
Do not write a colorspace label to the output image file (if applicable).
This does not affect the image processing.
May be useful for dealing with defective web browsers (I'm looking at
*you*, Firefox) that display sRGB image colors differently than they
display sRGB CSS/HTML colors.

-density
-density [,]
Control how the density label (i.e. pixels per inch) of the output image is
calculated.

Density policies available:
"auto": The default. Currently, writes a density only if the image size is
not being changed.
"none": Do not write a density.
"keep": Use the same density as the source image.
"adjust": Adjust the density so that the target image is the same physical
size as the source image.

Instead of a policy, you can request a specific density. Use "i" for pixels
per inch, or "c" for pixels per centimeter. For example, use
"-density i300" for 300 dpi.

This is limited by the target image format's support for a density labels.
Some formats do not support them, while others require them.

-translate [s],
Move the entire image by the given amount, measured in target pixels. This
is mainly intended for use with -imagesize, or for fine-tuning the image's
position relative to the pixel grid.
Note that x and y do not have to be integers.
If you use the "s" prefix, the measurements are in source pixels instead of
target pixels.

-offset (-offsetred -offsetgreen -offsetblue -offsetrb
-offsetvred -offsetvgreen -offsetvblue -offsetvrb)
While scaling the image, shift the position of a color channel by n output
pixels. A fractional number of pixels is allowed.

The options containing "v" shift the channel vertically; the others shift
it horizontally.
The "rb" options shift the red channel by the amount you specify, and the
blue channel in the opposite direction by the same amount.

Transparency is incompatible with this feature. If there's any chance your
image has transparency, you should also use -bkgd. Otherwise, IW will
select a background color that you probably won't like.

Checkerboard backgrounds are currently not supported when -offsetX is used.

-randseed
n is either an integer or the letter "r".
The seed to use if IW needs to generate random numbers.
"r" means to use a different random seed every time.
Default is 0.

-compress
Suggest the data compression method to use when writing the image.
Recognized options:
"none"
"zip"
"lzw"
"jpeg"
"rle"
The MIFF format supports "zip" (the default) and "none".
The BMP format supports "rle" and "none" (the default). RLE compression
will only be used if the number of colors is 256 or fewer.
For all other formats, this option currently has no effect.

-colortype
Deprecated. Use "-opt jpeg:colortype=".

-interlace
Write an interlaced PNG image, or a progressive JPEG image.

-noopt
Disable a class of image storage optimizations. This option can be used
more than once, to disable multiple optimizations.
This is based on the properties of the unoptimized output image, not on
the properties of the input image.
Optimization names:
"grayscale": Do not reduce color images to grayscale, even if all pixels
are gray.
"palette": Never convert to a paletted image.
"reduceto8": Do not reduce an image with a bit depth of 16 to one with a
bit depth of 8, even if it can be done losslessly.
"stripalpha": Do not strip a superfluous alpha channel, or use binary or
palette-based transparency.
"binarytrns": Never use binary (color-keyed) transparency.
"all": All of the above.

-page
Select the page to read from a multi-page file. The first page is number 1.
Currently, this only works with GIF files. It does not play through the GIF
animation, so you might only get a partial image.

-opt :=
Set a format-specific option. The syntax may be slightly inconvenient, but
it allows for a large number of options to exist, without much trouble.
Note that you will not get any warnings if you use an unrecognized option.
Options:
"bmp:version=": The BMP file version to write. Currently supports "2"
(Windows BMP v2, also known as OS/2 BMP v1), and "3" (the default; the
standard version that's compatible with almost everything), "5" (supports
transparency), and "auto" (v5 if needed for transparency, otherwise v3).
"deflate:cmprlevel=": zlib-style compression level setting for
zip/deflate/zlib compression. This applies to all PNG files, and to MIFF
files that use zip compression.
Values range from 0 (no compression) to 9 (best, slowest).
"-1" can be used to mean "default", but the exact meaning of this is not
well-defined.
"jpeg:arith": When writing a JPEG file, use arithmetic coding instead of
Huffman coding. This reduces the file size by about 5 to 10% for free,
but many image viewers don't support JPEG files with arithmetic coding.
"jpeg:bgycc": Enable libjpeg's "big gamut YCC" mode. (For experimental use
only.)
"jpeg:block=": Set the DCT block size. Valid values are 1 to 16. (For
experimental use only.)
"jpeg:colortype=": Suggest the color type to use for the output
image. This option is recommended for experts only.
JPEG color types:
"rgb": If a color image is written to a JPEG file, leave it in RGB
format instead of converting it to YCbCr. The resulting file will
likely be larger and less portable.
"rgb1": Use libjpeg's "reversible color transform" feature. (For
experimental use only.)
"ycbcr": Convert color JPEG images to YCbCr (the default).
"jpeg:optcoding": Enable libjpeg's "optimize_coding" feature, which makes
the output JPEG file slightly smaller in most cases.
"jpeg:quality=": libjpeg-style quality setting to use if a JPEG file is
written. Default is (probably) 75.
"jpeg:rstm=": Include restart markers in the output JPEG file, every
"MCUs".
"jpeg:rstr=": Include restart markers in the output JPEG file, every
"MCU rows".
"jpeg:sampling=,": The sampling factors to use if a color JPEG file
is written. For example, 2 means the chroma channels will have 1/2 as
many samples as the luma channel. For highest quality, use "1,1". The
default depends on the "jpeg:quality" setting. Each factor must be
between 1 and 4. Not all combinations are allowed.
"webp:quality": WebP-style quality setting to use if a WebP file is
written. This is on a scale from 0 to 100. Default is 80.

-includescreen
-noincludescreen
By default, a GIF image will be painted onto the GIF "screen". Use
-noincludescreen to extract just the individual image.

-zipcmprlevel
Deprecated. Same as "-opt deflate:cmprlevel=".

-jpegquality
Deprecated. Same as "-opt jpeg:quality=".

-jpegsampling ,
Deprecated. Same as "-opt jpeg:sampling=,".

-jpegarith
Deprecated. Same as "-opt jpeg:arith".

-bmpversion
Deprecated. Same as "-opt bmp:version=".

-bmptrns
Attempt to write a BMP image with transparency. Refer to technical.txt for
more information.

-webpquality
Deprecated. Same as "-opt webp:quality=".

-encoding
Set the encoding used for text output (informational and error messages).
This is usually unnecessary, because IW can usually figure out what
encoding to use. However, it may be useful if you're capturing or
redirecting the output, or using a nonstandard terminal program.

Encoding names:
"auto": The default. On Windows, write Unicode text if writing to a
console; otherwise write UTF-8. On non-Windows systems, try to detect
whether to use UTF-8 or US-ASCII.
"ascii": US-ASCII
"utf8": UTF-8
"utf16": (Windows only.) Write Unicode text if writing to a console;
otherwise write little-endian UTF-16.

-noinfo
Suppress informational messages.

-nowarn
Suppress warnings.

-quiet
Suppress informational messages and warnings.

-version
Display the version number of IW, and of the libraries it uses.

-help
Display a brief help message.