ImageMagick v6 Examples --
Color Quantization and Dithering

Index
ImageMagick Examples Preface and Index
Color Reduction Introduction (what is involved)
The Colors in an Image (what colors are used by an image)
Color Quantization (reducing the number of colors in an image)
Error Correction Dithering (Or Pseudo-Randomized Dithering)
Threshold Dithering Methods
Ordered Pattern Dithers (Using a tiled threshold map)
DIY Dither Patterns and Threshold Maps (dither images in your own way)
Reducing the number of colors or replacing specific colors is a very complex and difficult step in in Image Processing, and that is the topic covered in this example pages. This includes determining the what colors to use (color quantization), and how to place those colors on the image (dithering, and patterning). It also includes the generation of bitmap or two color images, and even handling Boolean (on/off) transparency.

This is so important that color reduction or quantization often happens automatically and behind the scene, just so ImageMagick can perform its original primary task of converting images from one file format to another, less colorful format, such as GIF, XPixmap, and XBitmap formats. Knowing how this works can allow you greater control of the process, so as to improve the resulting image stored in a specific image file formats.


Color Reduction Introduction

Color reduction is a very important aspect of ImageMagick. For example to convert a JPEG or PNG image containing millions of colors, into a GIF image containing a maximum of 256 color, you really have to be able to reduce colors in a efficient and effective way. Often during an image format conversion, this happens automatically behind the scenes, but there are other times when you want to do this manually.

Reducing the number of colors in image is a typically a three step process,
  1. First you typically need to survey the colors an image uses. Not only to see how many colors are actually used, but how often a particular color is used. It is no good preserving one specific color, if only a single pixel is using that color, though sometimes you still need to do that.
  2. Next you need to somehow decide on the final color set to which you want to limit your image. You may what IM to try and dertermine the 'best' set of colors for a specific image. Other times you may want something more general and global that can be used on any image. You may even what to specifically add or remove a color from the set of colors that will be used.
  3. And finally you need to modify the image so as to only use the colors you have selected. Preferably you want it so the results will look good, or perhaps you want it so that it will compress, compare, or optimize well.
To further complicate matters, these steps are often interlinked, as one method of replacing colors, often can only be applied using specific sets of colors. And if you are using a specific set of colors, doing some sort of color survey is not needed, or perhaps you need to make exceptions for specific colors.

Basically while color reduction is often automatically handled behind the scenes, it is good to at least be aware just what it is happening, and what its effects will be.

Color Survey

This is probably the lest important, and while IM provides you with methods to perform a survey, it is rarely done by users for the purposes of color reduction. I will leave further discussion to the relevant section, Extracting Image Colors.

Color Selection (Quantization)

For a good initial overview see Wikipedia, Color Quantization.

There are four basic methods for the selection colors.

These four color control methods: Quantization, Predefined Color Map, Uniform Colors, and Threshold; all have their limitations, as you will see.

Here is an example of each of these four methods...

  convert colorwheel.png +dither    -colors 32          color_quantize.gif
  convert colorwheel.png +dither -remap colortable.gif  color_predefined.gif
  convert colorwheel.png +dither   -posterize 3         color_uniform.gif
  convert colorwheel.png \
                  -separate -threshold 50% -combine     color_threshold.gif
[IM Output] ==> [IM Output] [IM Output] [IM Output] [IM Output]

The number of colors in each of the final images is only a representative set, but approximately 32 colors in each case (except for threshold which only has 8). From this you can get an idea of what you can expect from each of them.

All the other methods have a fix set of colors (according to the operators argument) regardless of the image that is being color reduced.

Only the first method ("-colors") will actually pick colors based on the the current image contents. As the test image is predominately white, a lot of lighter colors is selected.

It surveys the colors in an image using a technique known as "Adaptive Spatial Subdivision" using octrees. Then attempts to choose a specific set of colors to best match a specific image, within the limits given. See the Color Quantization Operator below.

The "-remap" lets you give IM your own set of predefined colors (See User Defined Color Maps).

The color map "colortable.gif" used in the above, is a set of 32 colors specifically picked for use in an old X Window Icon Library and is designed with cartoon like icons in mind. (See AIcon Library, X Icon Color Selection for details).

Using "-posterize" can also mathematically divide up each color channel into a set of color levels or intensities producing a 'uniform color map'. That is a color map with each channel set to a constant set of values or intensities.

And finally can "-threshold" all or specific color channels of the image, essentially making each color channel purely Boolean or on/off. That is each color channel can be given a value of zero or MaxRGB (IM 'Q' level dependant). This however only produces a minimal set of about about 8 colors. A very limited color set.

Thresholding is also equivelent to a "-posterize" level '1' which picks 2 colors.

Applying a Color Set

Once you have a set of colors, the next problem is to apply the color to an image so that the existing colors are replaced by the selected set of colors. This is known as 'Dithering', and is named as such because of its, "should I pick this, or should I pick that?", either-or nature.

Basically the idea of dithering is to place pixels of different colors near each other in such a way as to fool the eye into seeing more colors in the image than is actually used. That is the color in that area of the image more closely matches the original color of the image, because of the way the human eye 'merges' neighbouring colors together.

One of the best introductions to Dithering is on Wikipedia though you will need to skip over the 'Audio Dithering' section at the start. This presents an excellent set of examples of the benefits of using a dithered pattern of pixels, when you have a limited set of colors.

The basic styles of color replacement include...

Direct mapping if the nearest color in a given set is what was shown above. Basically you get distinct areas of solid, unchanging colors. When this is applied to a image of slowing varying color, such as a real life photo of sky, you get bands of colors across the image, especially in what would otherwise be a smooth gradient of colors, such as in sky areas. The result is typically regarded as not being very good.

The only time direct color mapping is usually thought of as acceptable is for logos, symbols, icons, and cartoon-like images. It is actually rarely an option. This is why you generally have to turn off the normal dithering method, if you do not want directly map colors in your images.

Dither however has its own problem. Once a image is dithered, a pattern of colors becomes part of the image. Once such a pattern is present, it is extremely difficult to remove. Also it is generally a bad idea to re-apply dithering to an image multiple times, as that just degrades the image.

Because of this, most of the quantization examples below will generally show you how to create un-dithered versions for each technique. It is done that way so you can see what color selections are being made before dithering hides that information.

Random dithering is the simplest dithering method created. It is also regarded as the worst possible dithering method. However it has some special uses. Within IM only works with two colors, so it is usually restricted to special case bitmap dithering. For more see Random Dither with Threshold below.

Error Correction Dithering is generally regarded as being the best general method of dithering colors across images as it will produce the closest approximation to the original color of areas in the image. It is also currently the only method that can dither any set of colors, and as such can be used for all four color reduction techniques. See How E-Dithers work below for more detail.

However Error Correction Dithering has some serious problems, especially with regards to animations of images.

The last two dithering techniques Ordered Diffused Pixel and Digital Halftoning is also regarded as a good method, and one that works well for animations, but currently it cannot use any set of colors, only a fixed set of uniform colors.

It does provide a means of coloring an image using patterns, allowing you to produce interesting effects otherwise not easy to produce.


All these aspects of color reduction are important techniques, and with understanding you can improve the results of your image operations, beyond the generalised defaults that IM provides. It is well worth the studying.


The Colors in an Image

Information about images, such as the number of colors used and the overall spread can be very important to programs and scripts that are trying to make decisions about the best techniques to use. Here I look at some of the methods you can use to determine this type of information, and not just for color reduction.

Extracting Image Colors

Extracting the color table

You extract a color palatte from an image using a verbose "identify", using any of these methods which basically all does exactly the same thing.

  identify -verbose  image.png
  convert  image.png miff:- | identify -verbose -
  convert  image.png  -verbose -identify null:
  convert  image.png  -verbose info:

The output from any of the above verbose identification will not return the color tables or histogram if there are more than 1024 colors! As such for large colorful images this is a hit or miss affair, and not recommended, though it can still be useful.

The better way however is to generate a "histogram:" of the image and extract the comment that is included in the result.


  convert  tree.gif  -format %c  -depth 8  histogram:info:-
[IM Output]
[IM Text]

The "info:" output format was added to IM v6.2.4. For IM versions before this use..

  convert  tree.gif histogram:- | identify -depth 8 -format %c -

The problem with these methods is that you are given a plain text output of the colors, which you will need to parse for your own needs.

However as of IM v6.2.8-8, the "-unique-colors" operator will convert an image into a smaller one containing just one pixel per unique color found in the original image, all in a single row.

This means you can convert a image into a simpler color table image, listing each color present. The width of the image returns the number of colors, and if you need to actually list the colors you can output it to a "txt:" image format.

For example here is the color table for the tree image.

  convert tree.gif -unique-colors -scale 1000%  tree_colors.gif
  convert tree.gif -unique-colors -depth 16  txt:-
[IM Output]
[IM Text]

This reduced color table is also very important as a way of storing a colormap of generated colors in a very small file. Such maps are particularly important for the "-remap" color reduction operator. (See Pre-Defined Color Maps below)

If you like to get a image containing not just the colors in a image but the color counts, here is one color-histogram solution what was developed from a IM Forum Discussion.

  convert rose: -colors 256 -format %c histogram:info:- |
    sed 's/:.*#/ #/' |
      while read count color colorname; do
        convert -size 1x$count xc:$color miff:-
      done |
        convert - -alpha set -gravity south -background none +append \
                unique_color_histogram.png
[IM Output] ==> [IM Output]

Note that I had to Color Reduce the image, as the built-in "rose:" image contains 3020 unique colors, which would take a long time and generate a very long image. The GIF image of the rose shown above contains the same set of color reduction.

The resulting image still contains the same number of pixels, though padded with extra transparent pixels, and as you can see shows a predominance of greenish greys, strong reds, as well as a very strong peak of pure white.

This may not be the best general color histogram method, but it works well for this image.

The order of the colors for both "histogram:" and "-unique-colors" operator, is undefined, but appears to be sorted by red, then green, and finally blue channel value. This may not the best way for a specific image, but it is impossible to generally sort 3-dimensional colors into a 1-dimensional order.

Extracting the Average Color

The average color of an image can be found very quickly by using "-scale" to reduce an image to a single pixel. Here for example is the average color of the built-in "rose:" image. I output the color using the FX Escape Format which returns a color string that can be used directly IM without change.

  convert rose: -scale 1x1\! -format '%[pixel:s]' info:-
[IM Text]

The problem with using the "%[pixel:...]" FX Escape is that it may return a color name such as 'white' or 'silver' instead of a RGB value.

However you can simulate this by using three FX Escapes to return the actual RGB values at the bit depth wanted. For example...

  convert rose: -scale 1x1\! \
     -format '%[fx:int(255*r+.5)],%[fx:int(255*g+.5)],%[fx:int(255*b+.5)]' info:-
[IM Text]

As of IM v6.3.9 there are a number of new "-format" escapes that can be useful to extract more specific information about images without needing to parse a verbose "identify" or "info:" output.

For example you can get the average red channel colour by getting the '%[mean]' greyscale value from of the images, red channel image.

  convert rose: -channel R -separate -format '%[mean]' info:
[IM Text]

Extracting a Specific Color

From the command line there are two basic ways of extracting a specific pixel color from an image. Either use an FX Escape such as "%[pixel:...]" or "%[fx:...]" (see above) on a specific pixel location...

  convert rose: -format '%[pixel:p{40,30}]' info:-
[IM Text]

Alternativally you can simplify the image by using "-crop" to cut out a single pixel that you may be interested in, and just that one pixel, using any of the methods shown above. For example...

  convert rose: -crop 1x1+40+30 -depth 8 txt:-
[IM Text]

Counts of a Specific (or near) Color

This can be used to get the pixel count or percentage of a specific color.

What you do is make anything not that color black, and then make that color white. For example lets get the number of colors in the "yellow" sun in the 'tree' image.

  convert tree.gif -fill black +opaque yellow \
                   -fill white -opaque yellow \
                   -print "yellow sun pixels = %[fx:w*h*mean]\n"   null:
[IM Text]

There is one cavat, it will not work if the color under test is itself black. To handle black (or very dark colors) swap the fills to map non-black colors to white and then Negate the results to generate the white mask of all black pixels.

Remember the "-print" option is equivalent to using "-format ... -write info:" and can be used anywhere within your image processing. I then junked the unwanted image using the special "null:" file format. You can also save the image to use as a mask for later work too.

Note that while this will work fine for small images, with much larger images (like high resolution digital photos) the 'mean' will not be accurate enough to get an exact pixel count! Basically the above use of 'mean' is sutiable for generating a ratio, but not for exact pixel counts. To get an exact pixel count you are better of using a histogram 'comment' output that has exact pixel counts (see above).

The above can also used a Fuzz Factor option "-fuzz" before the "-opaque" operator to specify 'near' colors as well.

Comparing Two Colors

So you have two specific colors and you want to compare them. You can use "compare" get the RMSE (on standard error)...

  compare -metric RMSE xc:Navy xc:blue null:
[IM Text]

This is good as it will get you the distance between the two colors, both in terms of values, and as a normalized percentage of the distance from black to white.

However this method will not handle transparency properly. For example comparing 'fully-transparent black' vs 'fully-transparent white'.

  compare -metric RMSE xc:'#0000' xc:'#FFF0' null:
[IM Text]

Transparent colors should actually have a zero distance as fully transparent is the same regardless of the underlying color. Instead we got a 4-d hypercube distance). As such the above method of color distance is only suitable for comparing fully-opaque colors only.


Rather than getting an actual distance, you can also use a Fuzz Factor to check is two colors are close.


  compare -fuzz 20% -metric AE xc:Navy xc:Blue null:
  compare -fuzz 30% -metric AE xc:Navy xc:Blue null:
[IM Text]

Remember however that the result will be '1' if the pixels do not match (number of error pixels). To get the actual 'fuzz' factor distance that seperates the values you can use the 'FUZZ' metric.

  compare -metric FUZZ xc:Navy xc:Blue null:
[IM Text]

The 'normalized' value shows that the actual distance is 28.7%.

Using Fuzz Factor, is different to calculating the RMSE when transparency is involved. That is because the fuzz factor is designed so that any two fully-transparent colors are treated as being equal.

As such 'fully-transparent black' and 'fully-transparent white' are exactly equivelent (producing a value of 0 or no error pixels)...

  compare -metric FUZZ xc:'#0000' xc:'#FFF0' null:
[IM Text]

Another method of color comparing is to try and Replace Colors with an appropriate Fuzz Factor percentage.

For example...

  convert xc:Navy  -fuzz 20% -fill Blue -opaque Blue txt:
[IM Text]
As 'Navy' did not change to 'Blue' it is more than 20% different to 'Blue'. Where as

  convert xc:Navy  -fuzz 30% -fill Blue -opaque Blue txt:
[IM Text]
This did change the color to 'Blue', so we now no know that 'Navy' is somewhere between 20% and 30% distant from each other.

To do this in a script use something like...

  fuzz=%1
  color1="red"
  color2="#e00"

  color2=`convert xc:"$color2" -format '%[pixel:s]' info:`
  result=`convert  xc:"$color1" -alpha set -channel RGBA -fuzz $fuzz \
            -fill $color2 -opaque $color2 -format '%[pixel:s]' info:`
  if [ "$result" = "$color2" ]; then
    echo "Colors match according to Fuzz Factor"
  else
    echo "Colors DO NOT match"
  fi

The special options "-alpha set -channel RGBA" are important to allow us to for fuzzy matching of transparent and near transparent colors.


Color Quantization

Color Quantization Operator

The primary work horse of color quantization, and what is used internally for all automatic color reduction, is the "-colors" operator.

This implements a "Adaptive Spatial Subdivision" color reduction algorithm, and is an extremely good color reduction algorithm.

Here is a typical example, I have a image of a 'colorwheel' image containing a lot of colors, and we ask IM to reduce the number of colors down to only 64 colors, using various dither methods.

  convert colorwheel.png  -dither None       -colors 64  colors_64_no.gif
  convert colorwheel.png  -dither Riemersma  -colors 64  colors_64_rm.gif
  convert colorwheel.png  -dither FloydSteinberg \
                                             -colors 64  colors_64_fs.gif
[IM Output] ==> [IM Output] [IM Output] [IM Output]

You can also see IM will by default use a 'dither' to shade the colors over the image. This prevents the sudden changes in color over smoothly changing gradients. If you turn off dithering (using 'None' or a "+dither" setting) you can clearly see what colors were merged together to generate what IM regarded as the best set of colors for this specific image. You can also see the sudden color changes that gradients of color will produce if dithering was not done.

Of course this image uses a lot more colors than what most images use. As such while a 64 color limit is often acceptable for many images, it is completely unacceptable for this image. In other words color quantization tries to find the best set of colors for a particular image.

Here are example of color quantization for part of IM logo, using an extremely small number of colors.

  convert logo: -resize 40% -crop 100x100+105+50\! -normalize  logo.png
  convert logo.png  +dither             -colors 8  colors_8_no.gif
  convert logo.png  -dither Riemersma   -colors 8  colors_8_rm.gif
  convert logo.png  -dither FloydSteinberg \
                                        -colors 8  colors_8_fs.gif
[IM Output] ==> [IM Output] [IM Output] [IM Output]

Compare that with some results for the built-in "rose:" photo image.

  convert rose:  +dither             -colors 16 colors_16_no.gif
  convert rose:  -dither Riemersma   -colors 16 colors_16_rm.gif
  convert rose:  -dither FloydSteinberg \
                                     -colors 16 colors_16_fs.gif
[IM Output] ==> [IM Output] [IM Output] [IM Output]

As you can see cartoon-like images require far less colors than a real photograph to produce a reasonable result.

Only one Color Quantization algorithm, "Adaptive Spatial Subdivision", is currently implemented in IM, and as it works very well, there has been little need to add others. However with feedback this algorithm is being steadily improved.

ASIDE: As a reference the "Gifsicle" program lists a number of other color quantization methods (using it "--color-method" option). I have no idea as to how well these color quantization methods compare to IM. If you find a good reference to different methods of color quantization, please mail me.

Color Quantization Internals

The process of selecting the limited number of colors to use in an image is called Color Quantization, and is a very complex process involving a number of factors. A full technical description of it is given on the ImageMagick web site Color Reduction Algorithm. However I'll try to example some of the more important aspects of this here.

Probably the biggest factor is the actual colors used in an image. It is no good picking a particular color for an image if their are very few pixels that are 'close' to that color. As such the color choice depends not only on the colors used in an image, but the number of pixels 'close' to the color.

I can demonstrate this quite easily by trying to reduce two different two color images to a single common color.

  convert -size 4x1 xc:blue -draw 'fill red   point 0,0' \
                                                -scale 20 colors_rb.gif
  convert -size 4x1 xc:red   -draw 'fill blue point 3,0' \
                                                -scale 20 colors_br.gif
  convert colors_rb.gif  -colors 1  colors_rb2.gif
  convert colors_br.gif  -colors 1  colors_br2.gif
[IM Output] ==> [IM Output]
[IM Output] ==> [IM Output]

As you can see the single final color depends not only on the colors present, but the amount of each color in the image.

Another thing that should be noted is that color quantization is not uniform over a linear gradient. This is to compensate for the typical 'gamma' correction that is present in images.

  convert -size 20x640  gradient: -rotate 90  gradient.png
  convert gradient.png   +dither  -colors 5   colors_gradient.gif
[IM Output]

FUTURE: Just what are the effects of the "-treedepth"  setting?
Mail me if you know

Color Quantization and ColorSpace

The other big influence on what colors are selected is defining exactly what we mean by colors that are 'close' or 'nearby'. This is defined by the colorspace used for the quantization (color selection), and is (as of IM v6.2.8-6) controlled by the "-quantize" colorspace setting.

The "-quantize" setting becomes particularly important when a very small number of colors are chosen. To demonstrate, lets reduce a standard 'colorwheel' image using various different color spaces and defining different 'color distances'.

  for S in    RGB CMY sRGB GRAY   YIQ YUV OHTA XYZ   HSL HSB HWB ; do \
     convert colorwheel.png   -quantize $S   +dither -colors 16 \
             -fill black -gravity SouthWest -annotate +2+2 $S \
             colors_space_$S.gif; \
  done
[IM Output] [IM Output] [IM Output] [IM Output]
[IM Output] [IM Output] [IM Output] [IM Output]
[IM Output] [IM Output] [IM Output]

As you can see the colors chosen depend heavily on how the colorspace is organized. The RGB (Red, Green, Blue) color cube will generally result in at least the colors close to the primary color being picked.

The RGB color space is particularly good at picking colors for cartoon like images and icons, but is actually a bad color space for general picture-like photos.

The CMY color space is exactly the same as RGB color space as the color channels are simply negated to convert between RGB and CMY colorspaces. As such the quantization colors end up with roughly the same solution.

The CMYK colorspace (not shown) also produces the same but for different reasons. Because internal the 'K' channel and a images 'colormap' use the same data pointer (See Palette Channel, IM converts it back to CMY and then RGB before quantization.

The sRGB colorspace as expected produces a simular result as RGB, but is warped to remove the number of near-black colors in the colorspace. As such there is less colors for the center of the colorwheel to select from, producing a larger 'not quite so black' spot. It does however work a lot better for most photos which is why JPEG images often use a sRGB colorspace.

On the other hand YIQ, YUV are designed to produce more natural 'pastel' and 'mid-tone' shades of colors that are much better suited for photographs and images of the real world involving subtle shades of colors.

Cristy, the ImageMagick Coordinator, notes that the best colorspace to use should be YIQ as it gives the best result when reducing a photograph of say a persons face to a limited color image such as GIF. I also agree with that recommendation.

Helmut Dersch notes on his web site that you should consider using a LAB colorspace for distortions.

The special color spaces involving a 'Hue' channel, such as HSL (Hue Saturation, Lightness), HSL (Hue, Saturation, Brightness), and HWB (Hue, White, Black), all have a cyclic color wheel representation of color as part of its color space. Actually it was using a HSL color space what was used to generate this color wheel. See Generating a Colorwheel.

At the time of writing, the color distance algorithm IM uses does not take into account the cyclic nature of the 'Hue' of the colorspace. The algorithm for this is very different. Because of this a strong discontinuity occurs along the 'red' path, where the 'Hue' wraps around, and results in very few red colors being selected in the color quantization process.

Finally there are also a large number of some very weird color spaces that are used for more specialized purposes. I have shown three of these above. If anyone has some info on what these (and other colorspaces) are used for, or can refer me to more info, please let me know.

In older versions of IM (specifically IM version 5) the color space that was used for quantization was set with the "-colorspace" option. However in IM version 6 this operator is used for modifying how images are stored in memory, and as such is not a setting for color quantization.

As such in IM v6.2.8-6, the "-quantize" setting was provided to do this job. However it is only as setting for the "-colors", Color Quantization process. It will not do anything for the replacement and dithering of colors using operators such as "-remap" and "-posterize", or the various dithering techniques.

For a complete list of the colorspaces available see the "-colorspace" operator.

You can see more effects of the colorspace on color selection by looking at the examples on Random Spots of Solid Color. There color quantization is used to reduce the number of colors in a randomized image using various colorspaces.

Quantization does NOT Preserve Colors

Note that in all the above images a pure-black color is never actually picked by Color Quantization. Mind you their is only one pure black pixel, and not many near-black colors in the image in any case. As a result the only black that appears in the final image was added later as part of the labeling of the image.

Even the 'GRAY' color space image did not produce a pure-black color. In fact none of the images contains any of the primary or secondary colors, such as: red, blue, green, cyan, magenta! The only exception to this is white, as the images did contain quite an amount of pure white color, making it a 'prefered color' (see below).

This situation however is not a bug!

First, a 'black' color was generally not selected in the above examples, usually because as their is very little black in the original image, so the Color Quantization generally did not worry too much about dark colors. In fact it generated more of the lighter colors as these are more common in the image. See the previous section for a specific example.

Secondly, as quantization is trying to pick colors that are close to the maximum number of existing color pixels in an image, this is best achieved by NOT matching a 'pure' primary or secondary color as these are the always at the very extremes of the color space being used. An 'off-color' will tend to match more colors than a 'primary' color, so these are more often selected.

So let me be clear...
Color Quantization ("-colors") will generally avoid picking primary colors!

As of IM version 6.3 the Color Quantization function was modified to try to include colors that are very common in the original image. As such if an image contains a area of a single color (such as 'white' in the above), that color will generally be included in the final color map.

This improves the situation somewhat, especially for 'cartoon' like images or images on a solid color background. The 'solid' color will generally be picked so as to help avoid Dither Speckling whcih we will look at below.

Specific Color in Colormap Solutions

At the moment there is only a few ways to gurantee a 'specific color' gets included into the selected colors for later dithering.

One way is to quantize the image as normal, but then output the generated color map (using "-unique-colors"). Now you can adjust that color map so your specific color is really that color. Finally you can use the Remap Colors operator to dither the image using the provided color map.

The colormap may no longer be the BEST colors for the image, and some other colors probably should also be adjusted, but it will be close to the colormap that you wanted.

Alternatively, append large patches of the specific colors wanted to the image, before using "-colors". The addition of the large 'swatch' of a specific color, will make it more likely to be picks in the final color map. Also all the other colors adjusted to fit that color map better).

If this works, you can then just Crop the image to remove the extra color patches added. If it doesn't, then IM should have at least added a color close to the wanted 'specific color' so only very slight adjustments of just those colors should be needed before you again Remap Colors to the original image yourself.

If you try this, with success or failure, please mail me

Ideally, what I would like to see added to IM is a way to specify a small number of specific colors, that must be part of the final color map, and then somehow ask IM to pick the best colors for rest of the colors in the color map, for a specific image.

Color Quantization and Transparency

ImageMagick by default not only generates fully opaque colors, but also attempts to generate semi-transparent colors. In this way, images containing transparent shadows or other overlay effects will not loose those effects.

However as IM v6.2.6, color quantization that involves transparency was modified so as to treat all fully-transparent colors as being the same color. This is a linear modification, so colors which are only half-transparent are also thought to be closer together than if they were fully opaque.

Because of this modification IM Color Quantization will still generate semi-transparent colors, but will concentrate more on the opaque colors and less on the fully transparent colors in the image.

For example here I generate a Rainbow Gradient of colors, with the image fully-opaque at the top, and fully-transparent at the top. I have displayed the images on a background pattern so that you can see just how transparent the image is.

  convert xc:red xc:yellow xc:green1 xc:cyan xc:blue \
          +append -filter Cubic -resize 100x100\!  -size 100x100 \
          gradient: +matte -compose CopyOpacity -composite alpha_gradient.png
  convert alpha_gradient.png  +dither  -colors 256  alpha_colors_256.png
  convert alpha_gradient.png  +dither  -colors 64   alpha_colors_64.png
  convert alpha_gradient.png  +dither  -colors 15   alpha_colors_15.png
[IM Output] ==> [IM Output] [IM Output] [IM Output]

As you can see, when we ask IM to reduce the number of colors needed by this image, it created a lot more opaque colors and used fewer highly transparent colors for the more translucent parts. The result is a very good spread of colors selected, especially when the number of colors is very small.

However just as I pointed out above, not only do primary colors not get picked, but the fully-transparent color will also not get picked for exactly the same reasons. In actual fact even fully-opaque colors will not get picked! In other words every color in the color quantized images in the previous example is semi-transparent.

Let me just make that clear.
When transparency is involved, IM Color Quantization
may not select any fully-opaque or even a fully-transparent color!

Of course as of IM v6.3, and the 'common color' bug fix (see Quantization does NOT Preserve Colors above), that is less likely to happen if the image contains a lot of opaque and fully-transparent colors, which is commonly the case.

As some images can contain a lot of semi-transparent colors, such as images involving smoke or shadows effects, you may like to do a trial run, to make sure a fully-transparent color is selected for inclusion in the resulting image. You can then map the most-transparent color to fully-transparent, and do the Remap Colors yourself.

If you really want to be sure you get both fully-opaque and fully-transparent colors in the resulting image you can Normalize OR Contrast-Stretch the alpha channel.

For example here I ensure the main color selects are made opaque by using "-contrast-stretch". Though this is probably a little heavy handed for a more normal situation.

  convert alpha_gradient.png  +dither  -colors 15 \
          -channel A -contrast-stretch 10%  alpha_colors_15n.png
[IM Output] ==> [IM Output] ==> [IM Output]

This is NOT a problem for GIF images which do not allow semi-transparent colors, or JPG which does not allow transparency, or even PNG which do not need quantization to save properly. It only becomes a problem in special cases where you may force a color reduction in an image when lots of semi-transparent colors are involved.

Remember for the GIF format saving semi-transparent colors is a useless endeavor. As such if you plan to do color quantization yourself for such an image format, you need to tell IM to ignore image transparency when generating its reduced color set. you can do that by using the special "-quantize" color space setting of 'transparent'.

  convert alpha_gradient.png -quantize transparent \
                            +dither  -colors 15   alpha_colors_15qt.png
[IM Output] ==> [IM Output]

Notice how the color quantization completely ignored the transparency of the colors, and did not touch the images alpha channel at all. This means you can process the alpha channel in a more appropriate way for your image, completely separately to the other colors. In fact you can do so either before or after using "-colors" without problems. It will make no difference in the result.

This quantization color space is thus recommended when reducing the number of colors for an image you plan to save to a format with Boolean or no transparency, such as GIF or XPM image formats.

If you count up the number of colors generated you will also see that it generated exactly the number of colors requested. As such if you you also need a fully transparent color (likely) then you need to reduce the argument of "-colors" by at least one, to leave space for it in the images final color table.

Thus to handle the GIF file format 256 color color table limit, you will need to reduce colors to 255, and not 256, leaving the extra space for the fully-transparent color index, as defined by the "-transparent-color" setting. Adjust this for a smaller color table sizes.

This quantization behaviour is automatic when IM saves to the GIF file format, but is important when you need to DIY the quantization yourself while generating global or shared color tables.

Of course you do still need to handle the semi-transparent pixels, so they are correct for what you want your image to look like.

FUTURE: This last part will probably move to a new section on 'Dithering
Alpha Channel' to be created in the near future. And a reference to this
section added here. 
Here are some examples of dithering just the alpha channel to just a Boolean or on/off setting, without effecting the rest of the color channels in the image.

  convert alpha_gradient.png \
          -channel A   -threshold 50%       alpha_dither_threshold.gif
  convert alpha_gradient.png \
          -channel A -ordered-dither checks alpha_dither_checks.gif
  convert alpha_gradient.png \
          -channel A -ordered-dither o8x8   alpha_dither_ordered.gif
  convert alpha_gradient.png \
          -channel A -ordered-dither h8x8a  alpha_dither_halftone.gif

  convert alpha_gradient.png -channel RGBA -separate \
          \( +clone -monochrome \) \
          +swap +delete -combine alpha_dither_monochrome.gif
  convert alpha_gradient.png -channel RGBA -separate \
          \( +clone -dither FloydSteinberg -monochrome \) \
          +swap +delete -combine alpha_dither_monochrome_fs.gif
  convert alpha_gradient.png -channel RGBA -separate \
          \( +clone -remap pattern:gray50 \) \
          +swap +delete -combine  alpha_dither_map.gif
  convert alpha_gradient.png -channel RGBA -separate \
          \( +clone -dither FloydSteinberg -remap pattern:gray50 \) \
          +swap +delete -combine  alpha_dither_map_fs.gif
[IM Output] [IM Output] [IM Output] [IM Output]
[IM Output] [IM Output] [IM Output] [IM Output]

When dithering a copy of the Alpha Channel, so you can dither it using either "-monochrome", or "-remap", make sure the image is a pure grayscale image, and not a shape mask containing transparency. If you don't you will probably end up with non-linear effects from the alpha channel still being present.

There are a number of ways of extracting and restoring the alpha channel from an image, as a gray-scale mask so you can dither it. The above uses the Channel Separation and Combine to do this. Other methods use Alpha Extraction with CopyOpacity Composition.


Error Correction Dithering

As discussed in the introduction an error correction dither is generally regarded the best choice for producing the truest representation of the original image with a reduced color set. It also limits itself to any pre-defined palette of colors, whether it was user supplied, or as determined by the IM color quantization routines.

Because of this it is the logical default choice for general color reduction as provided by the IM operators, "-colors", "-remap", "-posterize" and "-monochrome".

E-Dither Methods

As of version 6.4.2-9, IM now provides more than one type of dithering style or method, which can be selected using the "-dither" setting. Before this IM was limited to a variation of the Riemersma Dither, or Hilbert Curve Dither. which you can set using "-dither Riemersma".

Now you can also select a Floyd-Steiberg Dither using "-dither FloydSteinberg". You can see what types of dither methods has been implemented in your version of IM using...

  convert -list dither
[IM Text]

For example here is the color wheel dithered using different dithering methods.

  convert colorwheel.png -dither Riemersma      -colors 16 dither_riemersma.gif
  convert colorwheel.png -dither FloydSteinberg -colors 16 dither_floyd.gif
[IM Output] [IM Output]

As you can see the Floyd-Steinberg dither produces a much more uniform dither pattern than the default Riemersma dither. The biggest difference between them is how each of them distributes the 'color error' between neighbouring pixels. So lets have a look at just how a E-Dither works.

How an E-Dither Works

Re-Write In Progress

The specific method used by IM for general dithering is a variation of the "Hilbert Curve Error Correction Dither". This is actually a very good dithering technique, very well defined and reasonably quick. For a full description (and a very similar variation) see... Riemersma Dither.

Basically each pixel in the image is looked at in a very complex path known as a 'Hilbert Curve'. The pixel is assigned the color closest to that pixels value, and any difference between the pixels original color and the selected color, is saved and added to the next pixels color values (which is always a neighbouring pixel) before a new color is again selected. In this way any color variations between selected colors and the images original color is distributed to the other pixels in the same area. The result is that while only specific colors will be assigned to the final image, the same basic overall color for that area will closely match the original image.

For example, here is a small grey image that I asked IM to dither using a set of colors that does not include the original color. The resulting image is magnified so you can see the individual colored pixel assigned.

  convert -size 10x10 xc:'#999999' -scale 80x80  dither_not.gif
  convert -size 10x10 xc:'#999999' \
          -remap colortable.gif   -scale 80x80  dither.gif
[IM Output] ==> [IM Output]

As you can see the as the original images color was not in the specified color map, the original color is approximated using a pattern of the three nearest colors that were in the given color table.

If we were to average the color generated by the above dither pattern, we would get the color [IM Text], which is very close to the images original average color of [IM Text] and that is the whole point of the dither pattern that was produced.

However as the 'path' used to assign colors is complex (though generally remains in the local area), the color assignments produce an essentially random pattern. It isn't technically random however as the same image will produce the same pattern, but the results may as well be random, or at least pseudo-random.

The "F-S" dither is actually only one (the first one) of several 'Rasterized E-Dithers' that has been developed since its inception in the early 1970's. It is also probably the most widely implemented, even though it is not regarded as the best one. See the paper, Dithering Algorithms, for a more complete summary of such algorithms.

As of IM v6.4.3 it is also directly available in IM, and is implemented so as to follow a 'serpentine' path row-by-row from the top of the image to the bottom.

  convert -size 10x10 xc:'#999999'  -dither FloydSteinberg \
          -remap colortable.gif   -scale 80x80  dither_fs.gif
[IM Output] ==> [IM Output]

The "Floyd-Stienberg Dither" in particular I find produces a more 'hash' like pattern of pixels than the "Hilbert Curve Dither", and was in fact designed to do so.

Such a regular pattern can make a low level manual clean up of small color icon images, a lot easier. This was something I did a lot of in my past for Anthony's Icon Library, but that type of thing is not often needed anymore, except posibly for small monochrome images.

E-Dither Problem - Change Sensitive

One of the biggest problems you face when you use an error correction dither is that you get an essentially random pattern of pixels, that is also highly sensitive to changes.

Here for example take the original grey image and replace one pixel with a different color before dithering it again. The result is a complete shift of the dithering pattern in every pixel that is further along the path followed by the Hilbert Curve Dither.

  convert -size 10x10 xc:'#999999' -draw 'fill #C28 point 2,2' \
          -remap colortable.gif   -scale 80x80   dither_modified.gif
  compare dither.gif dither_modified.gif dither_difference.gif
[IM Output]
original dither
[IM Output]
one pixel change
==>
 
[IM Output]
comparison of changes

As you can see just adding a single pixel to the image resulted in the dither pattern changing dramatically! It only takes a single bit change for the resulting image to become different, even though overall look of an image (when not enlarged) is still basically the same (which is the purpose of a good dither algorithm after all).

The "compare" image also shows the extent of the change in the dither pattern. In this case, approximately 80% of the pixels were assigned a completely different color. In a Hilbert Curve Dither, a single pixel change actually will result in every pixel that comes later being possibly different, which means from 0 to 100% percent of the dither pattern could be different. It just depends on where in the complex Hilbert curve the change occurred.

The Floyd-Steinberg dither however only progresses though the image in one direction, and as such a single pixel change will modify the pattern only to one side of the change.

  convert -size 10x10 xc:'#999999' -draw 'fill #C28 point 2,2' \
          -dither FloydSteinberg -remap colortable.gif \
          -scale 80x80   dither_fs_modified.gif
  compare dither_fs.gif dither_fs_modified.gif dither_fs_difference.gif
[IM Output]
FS Dither
[IM Output]
one pixel change
==>
 
[IM Output]
comparison of changes

As you can see it has exactly the same problem. A single pixel change causing an almost complete change in the dither pattern for the areas of the image processed after that pixel. That is from that row downward.

For a single image the resulting pattern of the dithered colors is unimportant. The average color of the pattern should give the image the appropriate color for that area of the image. But when you have an animation in which one image is followed by other very simular images, with large areas of constant color, the changing dither pattern becomes highly noticeable and irritating as a low level background 'noise'.

For example, here I generate a 3 image animation of the same dithered color but with a single pixel change in each frame. I also magnify a central region so you can see this changing pattern more clearly.

  convert -size 80x80 xc:'#999999' \
          \( +clone -draw 'fill #C28 point 2,2' \) \
          \( +clone -draw 'fill #28C point 2,2' \) \
          -remap colortable.gif  -set delay 50 -loop 0   dither_anim.gif
  convert dither_anim.gif -crop 10x10+40+40 +repage \
                              -scale 80x80     dither_anim_magnify.gif
[IM Output] ==> [IM Output]

As you can see you get a sort of churning background to the image, caused by the pseudo-randomness generate by the E-Dither. In most cases the colors used are close enough together so as not to make this 'dither noise' visible. But when the dithering colors are visibly different (in this case forced by the use of a colormap) it definitely becomes an issue.

See Video Color Optimization for a more practical example of an animation showing this 'dither noise'.

The change in pattern also cause problems in optimizing animations. That is a different pattern means that simple frame optimization fails to reduce the size of frame overlays. For one solution see fuzzy color optimization, though that only works when the churn is using very simular colors.

Unlike other dithering methods (such as threshold, and ordered-dither) the "-channel" setting does not effect color quantization, or error correction dithers. Basically has it has no place in how these image operation work.

Ordered dithers do not have any of these problems, containing changes to the immediate local area of the change. Unfortunatally are also generally limited to using a mathematically derived color set. (See Ordered Dither using a Uniform Color Map).

E-Dither Pixel Speckling

Another problem with E-Dithers is that they can produce the occasional odd-colored pixels in areas which would otherwise be fairly uniform in color. For example the occasional green, pixel in a greyscale image. Or as in the examples below, a white pixel in a areas of otherwise plain flat blue color.

This is especially the case in large images which containing objects with large numbers of colors, and other areas of plain solid unchanging colors. This is especially typical of colored objects overlaid onto flat colored backgrounds, as you often get in diagrams and drawings.

You can see such a odd-colored pixel in the enlargement of the test examples above, where an extra light purple pixel was added quite a distance from the small single pixel change. The odd-colored pixels added to the above is not however readily visible and the color map does cover the image rather well, so the odd pixels are reasonably close to the normal three colors used for dithering the image.

For a more extreme example, here I have a blurred gradient background, which I heavily color reduced to 64 colors to really stress the error correction dither.

  convert -size 100x60 xc:SkyBlue \
          -fill DodgerBlue -draw 'circle 50,70 15,35' \
          -fill RoyalBlue  -draw 'circle 50,70 30,45' \
          -blur 0x5  -colors 64     speckle_gradient.gif
[IM Output]

As you can see with this highly reduced color map, the error correction dither did a reasonably good job of representing the original gradient.

But if we add a patch of pure white to the above...

  convert -size 100x60 xc:SkyBlue \
          -fill DodgerBlue -draw 'circle 50,70 15,35' \
          -fill RoyalBlue  -draw 'circle 50,70 30,45' -blur 0x5 \
          -fill white -draw 'rectangle 40,40 60,55' \
          -colors 64   speckle_problem.gif
[IM Output]

You can see that the E-dither suddenly started to produce a sprinkling of white pixels in the upper area of the image where we didn't have any before.

Here is a enlargement a small section so you can see these pixel more clearly...

  convert speckle_problem.gif -crop 15x15+75+0 +repage \
          -scale 90x90    speckle_prob_mag.gif
[IM Output]

The odd colored pixel is caused by two factors.

First, the Color Quantization was forced to include a single pure white color (but no other white-blue anti-aliasing colors) into the final colormap for the image, thus allowing the dithering process to use this extra color.

But as E-Dithers slowly accumulate errors, especially in areas of extreme colors, such as in the top section of the above image. Eventually the errors will add up to a value that is large enough make the one additional color the cloest match. As such, every so often a highly contrasting white pixel is output to 'correct the error', at a pseudo-random location.

The result is a very light speckling of white pixels. The slower the accumulation of error, the more spread out those white pixels are and the more, out-of-place, they appear.

The best solution is to switch to some other image format that does not have a limited color table. For example convert your GIF format image to PNG. This will avoid the need for color quantization (reduction) and hence the need to dither the reduced colors.

The next solution is to replace the use of E-dither with some other dithering method, that 'localizes' any errors, such as Ordered Dithering. However that is currently not easy thing to apply in IM at this time. See Better Ordered Dither Results, for one such method until a more general one is found.

If switching to another image format, or using a different dithering method is not practical (and often isn't), then you are left with attempting to fix the situation for that specific image.

The best fix, to this is to somehow insure you have other colors just outside the large group of colors that is causing the E-Dither error accumulation. However normal Color Quantization does not do this. It tends to pick a set of average colors representing color groups. What is needed is extra colors that 'picket fence' the edges of the a large color group, rather than a simple average color.

Here for example, I used a circle rather than a square, so that not only is a pure white color added, but a number of white-blue colors also. These were added automatically due to the anti-aliasing of the circle edges, to smooth its look.

  convert -size 100x60 xc:SkyBlue \
          -fill DodgerBlue -draw 'circle 50,70 15,35' \
          -fill RoyalBlue  -draw 'circle 50,70 30,45' -blur 0x5 \
          -fill white -draw 'circle 50,45 40,40' \
          -colors 64  speckle_fixed.gif
[IM Output]

And a magnification of the same area as before.

  convert speckle_fixed.gif -crop 15x15+85+0 +repage \
          -scale 90x90    speckle_fix_mag.gif
[IM Output]

As you can see the additional colors provide the extra colors, just outside the blue-cyan gradient. Now while these extra colors will mean there are fewer colors available for the actual gradient, they do provide other blue-white colors that will allow the E-dither to correct itself sooner and more often, before the accumulated error can grow too large.

That is not to say we have preventing E-dither speckling, just provided better colors for the dither algorithm to work with. If you study the magnified portion of the image you will still see a speckle pattern, but the colors are closer to the background color, and there are a lot more of them producing a more even, spread of speckles.

Another way is to generate our own color table, perhaps based on the one IM generated, and add the appropriate colors to prevent the error accumulation. This is however not an easy thing to do, especially with a 3-dimentional color space.

For this specific image example, one way to prevent 'speckling' is to generate and dither the background seperatally, with slightly less colors that is needed, then overlay the white box and its additional color.

  convert -size 100x60 xc:SkyBlue \
          -fill DodgerBlue -draw 'circle 50,70 15,35' \
          -fill RoyalBlue  -draw 'circle 50,70 30,45' -blur 0x5 \
          -colors 63 \
          -fill white -draw 'rectangle 40,40 60,55'  speckle_perfect.gif
[IM Output]

This will add a 'white' color to the image, but the background will not have any speckling effect, as white was not available when the error correction dither was used.

The result is an image with exactly 64 colors, and no speckling at all. However this is very dependant on the image and what you are trying to achieve, so is not a general solution to the specking problem.


A more general alternative to adding extra colors is try to remove the speckles from the final dithered image. That is clean up the image in some way.

However this is itself a tricky problem, as you do not what to remove pixels which are part of the normal dithering pattern.

What we need is to find color pixels which are somehow very different to all the the colors surrounding it, but which are also well isolated from all other similar colors, by some distance.

The solution is to pass the dithered image though some time of 'speckle filter', such as "-median" image filter.

  convert speckle_problem.gif -median 1 speckle_median.gif
[IM Output]

Note however that the dithering between the various color 'zones' was adversly effected, and the corner of the square was also rounded by this method.

Do you have a better image filter solution?

Summary

To me speckling is a very annoying problem, especially for desktop icon images using a very limited color table. I myself often edit smaller 'icon' images to remove speckles or fix some of the other dithering effects, such a vertical banding.

If you know of another better solution, please let me know.

Monochrome Dithered Bitmap Images

The "-monochrome" operator is a specialized form of both the "-colors" operator to generate a bitmap image. It is as such an ideal operator to demonstrate not only 'Hilbert Curve Dithering', but also have a closer look at color selection.

Here is a typical example.

  convert  logo.png  -monochrome     monochrome.gif
[IM Output]

The operator dithered the image solely based on their grey-scale brightness 'intensity' or 'level', however it doesn't dither the whole grey-scale range directly but thresholds the most extreme values to their maximum values.

We can see this by asking IM to dither a gradient image.

  convert -size 15x640 gradient: -rotate 90 \
                   -monochrome     monochrome_gradient.gif
[IM Output]

As you can see the gradient only has about the middle 50% of its colors dithered by the "-monochrome" operator. Special thanks goes to Ivanova <flamingivanova@punkass.com> for pointing out this interesting fact of the was IM works.

If you like to dither using the whole grey-scale range, you can use the "-remap" operator using a pure black and white colormap (supplied by a built-in pattern image).

  convert  logo.png  -remap pattern:gray50  mono_remap.gif
  convert -size 15x640 gradient: -rotate 90 \
                   -remap pattern:gray50     mono_remap_gradient.gif
[IM Output]
[IM Output]

By more careful selection of colors using "-remap" you can effectively produce the same 'threshold' range as used by the "-monochrome" operator, or any other threshold range you like.


  convert xc:gray20  xc:white  +append   ctrl_colors.gif
  convert logo.png -colorspace Gray \
          -remap ctrl_colors.gif  -normalize  mono_remap_ctrl.gif
  convert -size 15x640 gradient: -rotate 90 \
          -remap ctrl_colors.gif  -normalize  mono_remap_grad_ctrl.gif
[IM Output]
[IM Output]

What "-monochrome" actually does is to first convert the given image first into a grey scale image, after that it performs a two color 'Color Quantization', to decide the threshold colors to dither the image with. This is what the next section of examples will explore.

The "+dither" setting currently has no effect on the result of "-monochrome". This however may change in the future, so make sure it is not turned off in your scripts when using this operator.

Two Color Quantization

Rather than picking the two control colors yourself, you can use color quantization to pick the best two colors in the image by using the "-colors" operator.

  convert  logo.png   -colors 2 -colorspace gray  -normalize \
                                                 colors_monochrome.gif
[IM Output]

However the result will not be the same as using "-monochrome", as we didn't convert the image to grey-scale first.

Instead the image was dithered directly between the two non-grey color values chosen. That is the best two colors is selected to dither the image with, rather than two gray-scale brightness levels. Consequently it will produce a better result for say image that only uses colors of about the same gray-scale 'level'.

Here for example we use "-colors", as well as a "-monochrome" bitmap dithering operator, on a red-blue gradient. As you can so you can see that the results are not the same.

  convert -size 20x640 gradient:red-blue -rotate 90    gradient_rb.png
  convert gradient_rb.png   -colors 2 -colorspace gray \
                                -normalize        colors_threshold.gif
  convert gradient_rb.png       -monochrome       mono_threshold.gif
[IM Output] [IM Output] [IM Output]

The "-monochrome" operator in the above failed to find any differences to bitmap dither as both blue and red are very nearly the same intensity. Using a "-colors" quantization method however had no problem in finding acceptable colors to dither between.

You can also see that only the middle portion of the colors were dithered. This is due to the color quantization picking colors in the middle of two color 'clusters' it selected. Colors on the 'outside' of the selected colors in is thus effectively threshold directly to that color without dithering.

This demonstrates that colors on the outside of the quantization color space does not get dithered, though this fact is difficult to make use of in a practical way.

By setting the "-colorspace" to gray-scale before quantization, you will reproduce the internal operation of the "-monochrome" operator.

  convert  logo.png -colorspace gray   -colors 2  -normalize \
                                                 monochrome_equivelent.gif
[IM Output]

And finally, by turning off the dithering, you can produce a more automatic separation of the colors in the image than produced by using a fixed "-threshold" setting.

  convert  logo.png  -colorspace gray  +dither  -colors 2  -normalize \
           threshold_two_grays.gif
[IM Output]

Remember "-monochrome" currently ignores the "+dither" setting, so you can't just use that operator to do a 'smart threshold'.

If you remove the "-colorspace" for the color quantization stage of the image processing, you can threshold an image based on the best color separation (rather than greyscale color separation) possible for that image.

  convert  logo.png  +dither  -colors 2  -colorspace gray -normalize \
           threshold_two_color.gif
[IM Output]

Dither using Pre-Defined Color Maps

As shown above "-colors" attempts to choose an optimal limited set of colors with which to represent an image. With "-remap" you provide IM with the final set of colors you want to use for the image, weather you plan to dither those colors, or just replace the ones with their nearest neighbours.

The argument is given as a image containing all the colors you would like to use. If you what to reduce a large image of colors to just its list of colors, you can use "-unique-colors", before saving, it for later use by "-remap".

Note that while the "-remap" operator will accept any image to use, do not use a JPEG image for this image or you will get a lot of extra colors due to its 'lossy compression' generating extra colors.

For example here I limit the colors used in the IM logo to a predefined map of named X window colors. The default is the 'Riemersma' dither, but as of IM v6.4.4 "-dither" was expanded to allow the selection of other dither methods such as 'FloydSteinberg'. You can of course still turn off dithering using the "+dither" option.

  convert logo.png  -dither None       -remap colortable.gif  remap_logo_no.gif
  convert logo.png  -dither Riemersma  -remap colortable.gif  remap_logo_rm.gif
  convert logo.png  -dither FloydSteinberg \
                                       -remap colortable.gif  remap_logo_fs.gif
[IM Output]  + [IM Output] [IM Output] [IM Output] [IM Output]

As you can see IM attempted to do a reasonable job of representing the image using just the given colors, though the results is nowhere near as good as the image you get if you allowed IM to select the color set to use.

Mind you this "colortable.gif" image was never designed for dithering images, but as a color set for designing cartoon-like color icons for older more primitive X window color displays (See Anthony's X Window Icon Library and AIcons Color Selection for details).

Mind you IM's selection of 'LightSeaGreen' instead of say 'DodgerBlue' for the wizards jacket is rather surprising and is probably due to the use of RGB colorspace for color distancing. Unfortunately you do not have a choice about the color space for actual color internal map/dither operation.

Also note that the final image did not use all 32 colors provided by this map, though more of the colors in the map will be used when some form of dithering was enabled ( and respectivally), than when it was turned off ().

This last example shows just how important selecting a good colormap is. Because of this I recommend you let IM optimize the color selection used in an image using the "-colors" operator, and modify that to suit your needs, unless you have more pressing reasons not to do so.

One final point, while you can specify a color space in which "-colors" will find the best set of colors, you currently can NOT define a color space for the color mapping or dithering phase. All my experiments seem to show that the color set is applied (both error correction dither and nearest color replacement) based on RGB space. The "-quantize" colorspace setting is only used for the selection of colors, not its mapping.

So if using a color map is such a bad idea, why would you want use it?

There are a number common reasons, usually because you need more control of the specific palette of colors used in an image. If you know of another reason to use the "-remap" operator which I have not presented below - Mail me.

Common or 'Best' Colormap

The other technique, when handling multiple images, is to generate a common color table for all the images involved.

Basically you append all the images together into one large image, then use the "-colors" operator to figure out a good color map to use that is common to all the images. Once you have that color map image you can use it to re-color each of the original images using this juts generated Pre-Defined Color Map.

Alternatively, you can use the special "+remap" operator, which does the same thing to a 255 color colormap. It counts up the colors, perform the color quantization to form a good, common colormap, then dithers the images to use that map, if needed.

Both "-remap" and "+remap" forms however has one very important feature for GIF animations. It converts all the images to a image "-type" of 'Palette' with all the images using the same palette of colors.

The reason is that when writing a GIF image, the first images color palette, will be used for the file formats 'global colormap'. Then as each image is written it notes that those images use the same set of colors, so it does NOT create a 'local colormap'. This can save up to 256 × 3, or 768 bytes of colormap space for each and every image the final GIF file.

Only the "-remap" operators can do this. So when handling GIF's and in particular GIF animations, it can be an important point to remember.

For more details, and an example see Gif Animations, Global Color Table.

Web Safe Coloring

When the WWW was first created, computer displays had a limited range of colors available, and web browsers usually used a simpler set of colors for images. As such it was common to recolor images to this color set, to make them both smaller, and to ensure they will look okay on users browsers. For more detail see Web Style Guide, Dithering.

To help with this IM provided a built-in colormap image of this special table of 216 colors, called "netscape:". So lets look at how our test image would look on of a old web browser display using these colors.

  convert logo.png         -remap netscape:  remap_netscape.gif
  convert logo.png +dither -remap netscape:  remap_netscape_nd.gif
[IM Output]  + [IM Output] [IM Output] [IM Output]

This color set was a mathematically determined pallet, designed by engineers of displays and computers, not graphic artists, and while it is is general enough that it works reasonably well for real images such as photos, it is very bad for images containing large flat areas of color, such as logos, backgrounds, computer generated images such as graphs, and cartoon-like images.

Basically this works in areas of highly variable colors, but for the larger flat areas of constant colors a dithering of three colors (in general) is applied, such as the off-blue shirt of the IM logo test images (above).

In other words, if you were designing an image or logo for use on the web, you work generally try to use the colors in this pallet for the large flat areas, and only have dithered colors in areas where your have varying shades of color.

The above commands lets you test your images to see how they would look on more primitive computer displays, and to edit image to use these colors, so they will work well. This is particularly important for symbols and navigation images.

Of course today, thanks to the demands by game and web users, you can be pretty well assured that most users have a modern computer display that does not have those old color limitations, however the use of this "web safe palette" is still around, as it does have other benefits, such a image compression.

For a discussion about the use of Web-safe colors in the modern world see Death of the Web-safe Color Palette?, and probably a more important view from a graphic designer that first identified this color map, Lynda Weinman.

Generating Color Maps

Determining a good color map for any image, or a a specific set of images, can be very important. This becomes especially important when you are dealing with a sequence of images that will be used for GIF animation. Basically you want to make it so they only need one color table, to use for all the frames of the animation, rather than a separate color table for each frame. In other words you want one single color map for all the image.

You have really only two choices in this case. You can attempt to create a colormap that will work well for any image, or you try to optimize a color map for the specific set of images you are applying it to.

Web-Safe Colormap

[IM Output] The first method is typically a mathematically generated color map, such as the IM built-in "netscape:" color map. This provides a set of 216 colors, which will nicely fit into the GIF formats 256 color limit, and still have space for handling image transparency, or even adding some extra colors for special purposes, like shadows, or text overlays.

This color map was generated by creating 6 levels of colors for each of the three color channels, producing 6×6×6 colors or 216 colors. The number of the beast.

As only 219 color are used it still has space (for GIF images) to add more colors to the color map for specific purposes. For example a transparent color, as well as more grey scale shades. An old Macintosh version of the web-safe map actually did exactly this to try to improve its overall result, but it was only used on Macintosh web clients.

This is probably the most common 'uniform' (or mathematically derived) colormap in general use, thanks to its simplicity, and its general use on the the World Wide Web.

Uniform 332 Colormap

Another uniform color mapping that is commonly used is the "332 RGB color map". The number refer to the number of bits used to represent each color within a 8 bit color index. That is 3 bits (or 8 levels) of red, 3 for green, and 2 bits (or 4 color levels) for blue, seeing as our eyes do not respond well to blue. This gives 3+3+2 bits or an 8 bit color index, or 256 colors. Perfect for the limited GIF color table. However it will not leave any space for a GIF transparency color, or other special use colors.

Here is one way to get IM to generate this color map...

  convert -size 16x16 xc: -channel R -fx '(i%8)/7' \
                          -channel G -fx '(j%8)/7' \
                          -channel B -fx '((i>>3&1)|(j>>2&2))/3' \
          -scale 600% colormap_332.png
[IM Output]

The bit-shifting operators '>>' and '<<' was missing from the "-fx" operator until IM version 6.2.9-2.

A simpler way of doing the same thing is to use Ordered Dither using Uniform Color Levels using the operation, "-ordered-dither threshold,8,8,4" (see that examples area). A far easier and faster technique than the above DIY FX method, and even allows you to use other built-in dithering maps for better gradient handling.

The only drawback with this map is that it does not actually provide any 'gray' colors at all. However this drawback can be a plus when dithering is used as the slight color differences reduce the effect of color boundary changes in a grey-scale gradient, making it just that little bit smother looking.

TrueColor 16bit Colormap

A similar uniform colormap to the '332 colormap' above is used by X windows in a rarely used 16 bit visual class. In this case, 16 bits are used for the color index which is divided into 5 bits for red, 5 for green, and 6 for blue.

In other words, this color map is more like a "556 colormap", and is best achieved using a Ordered Dither using Uniform Color Levels using a 'threshold' dithermap. Specifically the operation "-ordered-dither threshold,32,32,64". 16 bit colormaps are however rarely seen as images using colormaps typically need a 8 bit color table. As such I won't mention it further.

Gamma Corrected Uniform Colormaps

AT this time IM does not do gamma correctect color maps directly.

Instead what you should do is convert your image (assuming you have Q16 or better compile time quality version of IM), from the sRGB or whatever gamma level the image has to a linear RGB model, before doing your dithering.

This also goes for many other image processing operations, such as resize, bluring, etc.

See Resizing with Gamma Correction for an example.

Posterize, Recolor using a Uniform Color Map

The operators original purpose (using an argument of '2') is to re-color images using just 8 basic colors, as if the image was generated using a simple and cheap poster printing method using just the basic colors. Thus the operator gets its name.

The "-posterize" operator is actual fact is a special color reduction operator that generates a color map based on the the number of color 'levels' given, for each color channels in the image, dithering the image using an error correction dither.


  convert netscape: -scale 50%  +dither  -posterize 2   posterize_2_ns.gif
  convert netscape: -scale 50%  +dither  -posterize 3   posterize_3_ns.gif
  convert netscape: -scale 50%  +dither  -posterize 6   posterize_6_ns.gif
[IM Output] [IM Output] [IM Output] [IM Output]

As you can see a "-posterize" argument on '2' means to only provide 2 colors per color channel, producing a map of just 8 colors for a 3 channel RGB image, such as the above. Basically it will recolor images using the threshold set of 8 colors.

An argument of '3' will map image colors based on a colormap of 27 colors, including mid-tone colors. While an argument of '4' will generate a 64 color colortable, and '5' generates a 125 color colormap. Of course as mentioned above, an argument of '6' will reproduce the same set of 216 colors, as provided by the in the built-in "netscape:" image.

Please note that a "-posterize" argument of '0' or '1' is non-sensible, and with the latest IM releases just converts images to pure black (which though logical, is quite useless).

The result is that the image has been recolors using a mathematically derived or 'uniform' color map.

You can see this more clearly on a gradient image, producing an even distribution of posterized gray levels.

  #convert -size 20x640  gradient: -rotate 90  gradient.png
  convert gradient.png  +dither  -posterize 5   posterize_gradient.gif
[IM Output]

For example lets posterize the IM logo image at various levels...

  convert logo.png  +dither -posterize 2  posterize_logo.gif
  convert logo.png          -posterize 2  posterize_logo_dither.gif
  convert logo.png          -posterize 6  posterize_6_logo.gif
[IM Output] ==> [IM Output] [IM Output] [IM Output]

As a better test lets poserize the shaded "colorwheel" image.

  convert colorwheel.png  +dither  -posterize 2   posterize_2_cw.gif
  convert colorwheel.png  +dither  -posterize 3   posterize_3_cw.gif
  convert colorwheel.png  +dither  -posterize 6   posterize_6_cw.gif
[IM Output] [IM Output] [IM Output] [IM Output]

And here is the same thing with dithering enabled...
[IM Output] [IM Output] [IM Output] [IM Output]

Of course many of the bitmap dithers we look at in the next section can also generate level 2 ordered dithers, using various sorts of dither styles. However few can use a larger number of grey levels.

Ordered Dither as of IM v6.2.9, is also a posterization method, due to its current limitation in dithering using a uniform colormaps. However the dither pattern is more uniform, with a larger selection of styles, than the pseudo-randomized dither produced by "-posterize". Compare these with the dithered "-posterize" versions above.

  convert colorwheel.png  -ordered-dither o8x8,2   posterize_2_od.gif
  convert colorwheel.png  -ordered-dither o8x8,3   posterize_3_od.gif
  convert colorwheel.png  -ordered-dither o8x8,6   posterize_6_od.gif
[IM Output] [IM Output] [IM Output] [IM Output]

The 'threshold' dither map (instead of 'o8x8' used above) effectively converts "-ordered-dither" into a un-dithered posterization method.

Finally the Ordered Dither does allow you to specify a different number of color levels, for each individual color channel. Something that the "-posterize" operator does not currently allow.


Threshold Dithering Methods

Threshold Images

The simplest method of converting an image into black and white bitmap (to color) image is to use "-threshold". This is actually a simple mathematical operator that just provides a cut-off value. Anything equal to or below that value becomes black, while anything larger becomes white.


  convert logo.png     -threshold   -1   threshold_0.gif
  convert logo.png     -threshold  25%   threshold_25.gif
  convert logo.png     -threshold  50%   threshold_50.gif
  convert logo.png     -threshold  75%   threshold_75.gif
  convert logo.png     -threshold 100%   threshold_100.gif
[IM Output] [IM Output] [IM Output] [IM Output] [IM Output]

As you can see a value of '-1' will convert all colors white, while '100%' converts all colors to black. A '50%' is of course the most common value used.

A value of '0' is a special case which turns all non-pure black colors, white. Of course if the image does not have pure-black colors, then you will only get a solid white image!

  convert logo.png  -threshold   0    threshold_black.gif
[IM Output]

If you actually want to convert all non-pure white colors to black, then I recommend that you threshold the negated image instead of trying to work out the right threshold value to use (one less than IM's current 'MaxRGB'), a value that is dependant on your specific IM's compile time in Quality, or 'Q' setting.

  convert  logo.png  -negate -threshold 0 -negate threshold_white.gif
[IM Output]

The "-threshold" operator can be classed as a ultimate 'contrast' operator, maximizing the differences in colors by the threshold level. It is however a gray-scale operator, meaning that the "-channel" setting, can be used to adjust which color channel the operator will be applied to.

For example you can threshold each of the individual channels of the image to produce the same effect as a un-dithered level 2 "-posterize" operation.

  convert  logo.png  -channel R -threshold 50% \
                     -channel G -threshold 50% \
                     -channel B -threshold 50%   threshold_posterize.gif
[IM Output]

Note that "-threshold" treats any transparency in an image as a matte channel, not an alpha channel (just as it is stored internally within IM). As such caution is needed if you plan to apply this operator to the alpha channel. See Matte Channel for more details.

For a more automatic thresholding technique, you can use a Two Color Quantization technique that we showed previously.

For example, this will threshold the image based on the best two colors found in the image. These colors may not necessarily be greyscale or even opposites, just the two colors that best represent the whole image. The two colors are then mapped (using "-normalize") to pure black and white.

  convert  logo.png  +dither  -colors 2  -colorspace gray -normalize \
             threshold_two_color.gif
[IM Output]

Random Dither and Threshold

The "-random-threshold" operator is a special form of bitmap image converter. In this case it uses a very simple "random dither" to determine if a particular pixel is to become a white pixel or a black pixel.

Unlike the "-threshold" or the "-monochrome" operators, or even the variations in the previous section, the selected channels (set with "-channels") are not merged together into single grey-scale channel and dithered as a single unit. Instead "-random-threshold" works on each selected channel completely independently of each other.

Of course using the operator directly will result in a 2 level posterization of the image using a random dither.

  convert  logo.png  -random-threshold  0x100%  random_posterize.gif
[IM Output]

Converting to gray-scale will equalize all the channels in the image, before they are dithered. But as each channel is dithered independent of each other and in a random way, the result is not a bitmap image as you would expect. Instead you will get a splatter of color pixels, especially for mid-tone colors.

  convert  logo.png  -colorspace Gray -random-threshold  0x100% \
                                                   random_greyscale.gif
[IM Output]

Here is the correct way to generate a proper random dithered bitmap image.

  convert logo.png  -colorspace Gray -channel B \
          -random-threshold 0x100%    -separate   random_monochome.gif
[IM Output]

Basically what it did was to only dither one channel of the grey-scaled image, and then use the "-separate" channel operator to extract that channel as the final bitmap image. Tricky but effective.

As a special feature for this operator, IM will ensure a bitmap image is generated in the special "-channels" option of 'All' is used.

  convert  logo.png  -channel All -random-threshold 0x100% random_all.gif
[IM Output]

However please note that any alpha channel will be ignored and lost using this method, as such it is not normally recommended. I myself only discovered this ancient feature by accident from the source code.

Now that you know how to use the operator to correctly generate bitmaps from a color image lets look how the argument effects the range of the dithering. This also shows clearly the 'clumping' of pixels that this dither produces.

  #convert -size 20x640  gradient: -rotate 90  gradient.png
  convert gradient.png  -channel All \
                        -random-threshold 0x100%  random_grad_0x100.gif
  convert gradient.png  -channel All \
                        -random-threshold 10x90%  random_grad_10x90.gif
  convert gradient.png  -channel All \
                        -random-threshold 25x75%  random_grad_25x75.gif
  convert gradient.png  -channel All \
                        -random-threshold 50x50%  random_grad_50x50.gif
[IM Output] [IM Output] [IM Output] [IM Output]

A "-random-threshold" setting of '0x100%' will produce a purely 'Random Dither' of the image. If the two bounds are set to the same value (or even past each other) it will just produce a pure "-threshold" image.

Using any other sets of bounds (usually specified using a percentage) will threshold the bitmap outside the given range, while producing a random dither pattern for the values within the given range.

The best results can be obtained by using a slightly smaller range, just as you get using the "-monochrome" operator. A value of about '30x80%' probably the best result for most cases.

  convert  logo.png  -channel All -random-threshold 30x80%  random_30x80.gif
[IM Output]

Of course the result is still not very good. But then this is the simplest and worst form of dithering you can get.

What actually happens is that the randomized dither pattern tends to produce 'clumps' of pixels rather than a smooth dither pattern. this is due to the high frequency 'noise' in the random number generator. However at very high resolutions a random dither has been shown to produce a extremely good result, if random enough. IM uses a cryptographic level of randomness so it will likely be very random, though images are rarely used at a high enough a resolution to make it useful in this way.

One 'fix' for this dither, that has been suggested is to use a random 'blue noise' generator (a high frequency filter, as opposed to a low frequency 'pink noise' filter used in sound production). This should remove the clumping of the pixels, but is very difficult to implement digitally. No known implementation of 'blue noise randomized dither' has been found, and unlikely to be ever created.


Ordered Dithering

While a random dither produces random clumps of pixels, and the various error correction dithers produce a essentially random pattern of dots, ordered dithering is basically the opposite. It is designed to be as mathematically deterministic as possible. So deterministic, that you actually need to specify the pattern for it to should use in dithering the image.

The "-ordered-dither" operator will dither each of the selected "-channels" in the image a given predefined pattern. The argument defines the pattern (known as a threshold map) to use.

These threshold maps fall into three basic styles. Diffused Pixel Dithers where the pixels are placed as far from each other as possible, so as to avoid 'clumping' and tiling artifacts. Or to clump them together into tight dots that makes them easier to mechanically print, in a technique known as Digital Halftoning. There is also some specialized artistic threshold maps that we will also be looking at, and even designing our own dithering pattern or threshold maps.

In each case the number of pixels that is on or off in the threshold map depends on the grey-level intensity of the image (or individual color channel) that is being dithered into a bitmap.

The map adds pixel threshold levels in a consistent manner, so that once a pixel turns on at a particular 'threshold' it remains on for any lighter grey color. This consistency is very important, otherwise artifacts are produced along the boundaries of changes the dither pattern.

The important point about this is that the result for each pixel in an image is purely mathematically determined independently of any other pixel in the image. As such any small change to the original image will have absolutely no effect on the image in any other area a problem that Error Correction Dithers have, as we saw above.

This point is vital to consistent dithering of video images and optimized animations.

Diffused Pixel Dithering

The original purpose of ordered dithering, and what most graphic programmers expect to get when you use an ordered dither is sometime more correctly termed, "Diffused Pixel Ordered Dither".

What this means is that pixels are added to the tiled map as the threshold intensity increases, so that they are as far away from each other and evenly distributed as possible. This produces a highly consistent pattern that look quite smooth and near invisible on most modern displays.

Such patterns have been worked out for tiling sizes that are a power of 2, namely, tile sizes of 2, 4, and 8. Though IM also provides a reasonable threshold pattern for a 3 by 3 threshold map tile.

Here is the current set of built-in ordered dithers that IM currently provides. Remember the argument reflects the tile size of the ordered dither.

  convert logo.png    -ordered-dither o2x2    logo_o2x2.gif
  convert logo.png    -ordered-dither o3x3    logo_o3x3.gif
  convert logo.png    -ordered-dither o4x4    logo_o4x4.gif
  convert logo.png    -ordered-dither o8x8    logo_o8x8.gif
[IM Output] [IM Output] [IM Output] [IM Output]

Note how a larger tile size allows you to simulate more 'color levels', but also generates more noticeable defects or rectangular arrays of pixels at certain levels.

The 'o8x8' ordered dither was part of the IM core code for a long time, but was not used. It was only added as an option to the "-ordered-dither" operator IM v6.2.9, when IM Examples started to detail the use of this operator.

At this time the maps were given more definitive names to allow further expansion of the "-ordered-dither" operator, though the older backward compatible 'tile size' names, were retained as aliases to the new names.

Also the 'maps' that produced the 'o3x3' and 'o4x4' were completely revised to produce a better 'diffused pixel' dither pattern. Before this the maps produced distinct 'clumps' of pixels.

See Ordered Dither Upgrade notes page for examples of the old patterns before they were fixed, as well as other changes made during the development for the official release of the upgrades in IM v6.3.0.

Of course you need to convert the image to grey-scale first to produce a proper bitmap of all the channels in the image, however as the process is not random you don't need to post-process the image as you need to for the -random-threshold" operator simplifying things enormously.

  convert logo.png -colorspace Gray  -ordered-dither o2x2  logo_bw_o2x2.gif
  convert logo.png -colorspace Gray  -ordered-dither o3x3  logo_bw_o3x3.gif
  convert logo.png -colorspace Gray  -ordered-dither o4x4  logo_bw_o4x4.gif
  convert logo.png -colorspace Gray  -ordered-dither o8x8  logo_bw_o8x8.gif
[IM Output] [IM Output] [IM Output] [IM Output]

As a reference here are each of the "-ordered-dither" 'diffused pixel' patterns applied to a grey-scale gradient, so you can see clearly what they look like.

  # Threshold Non-Dither / Minimal Checkerboard Dither
  convert gradient.png   -ordered-dither threshold  od_threshold.gif
  convert gradient.png   -ordered-dither checks     od_checks.gif
  # Diffused Pixel Dither
  convert gradient.png   -ordered-dither o2x2       od_o2x2.gif
  convert gradient.png   -ordered-dither o3x3       od_o3x3.gif
  convert gradient.png   -ordered-dither o4x4       od_o4x4.gif
  convert gradient.png   -ordered-dither o8x8       od_o8x8.gif
[IM Output] [IM Output]
[IM Output] [IM Output] [IM Output] [IM Output]

The number of effective or pseudo-level patterns produced by a specific ordered dither, is typically (by not always) equal to the number of pixels in the pattern plus one. As such a 'o3x3' ordered dither will produce 3×3+1 or 10 effective grey levels per channel (black, white, and 8 artificial gray patterns) in the resulting image.

Also shown above are two special minimal dither threshold maps:
  1. a straight '50% threshold' non-dither, which does not produce any extra grey levels and
  2. a 'checks' or checkerboard dither pattern, that only inserts a single pattern to add a extra 'pseudo-level' to the resulting gradient.

Digital Halftone Dithers

The "-ordered-dither" was extended in IM v6.2.8-6 with a set of digital-halftone dither patterns (thanks Glenn Randers-Pehrson). All of which were set to produce a simple 45 degree dot pattern. With IM v6.3.0 this was further expanded with a similar set of larger un-angled halftones.

Before the release of IM v6.3.0 halftone screens were selected by using an argument of the form '{number}x1'. With the Re-Development of Ordered Dither this limitation was lifted, better naming selected, and extra halftone screens (orthogonal forms) was added (see example arguments below).

Note that digital halftoning is not strictly a true halftone screen, which is designed to handle round dots of ink mechanically deposited onto a medium such as paper, cardboard or even metal. Such dots can overlap, and smear during the printing process, thus requiring some non-linear level adjustment. This is not needed for producing purely digital halftone effects. For more details of the process see the document Dithering and Halftoning (PDF).

That said the Ordered Dither digital halftone patterns do provide the same basic effect as seen in newspapers and cheaply printed magazines.


  # Halftone Screen (45 degree angle)
  convert logo.png   -ordered-dither h4x4a    logo_h4x4a.gif
  convert logo.png   -ordered-dither h6x6a    logo_h6x6a.gif
  convert logo.png   -ordered-dither h8x8a    logo_h8x8a.gif
  # Halftone Screen (orthogonal)
  convert logo.png   -ordered-dither h4x4o    logo_h4x4o.gif
  convert logo.png   -ordered-dither h6x6o    logo_h6x6o.gif
  convert logo.png   -ordered-dither h8x8o    logo_h8x8o.gif
[IM Output] [IM Output] [IM Output]       [IM Output] [IM Output] [IM Output]

Again to use the "-colorspace" operator to generate a true bitmap dither of an image.

  # Halftone Screen (45 degree angle)
  convert logo.png -colorspace Gray  -ordered-dither h4x4a logo_bw_h4x4a.gif
  convert logo.png -colorspace Gray  -ordered-dither h6x6a logo_bw_h6x6a.gif
  convert logo.png -colorspace Gray  -ordered-dither h8x8a logo_bw_h8x8a.gif
  # Halftone Screen (orthogonal)
  convert logo.png -colorspace Gray  -ordered-dither h4x4o logo_bw_h4x4o.gif
  convert logo.png -colorspace Gray  -ordered-dither h6x6o logo_bw_h6x6o.gif
  convert logo.png -colorspace Gray  -ordered-dither h8x8o logo_bw_h8x8o.gif
[IM Output] [IM Output] [IM Output]       [IM Output] [IM Output] [IM Output]

And finally another gradient reference image to clearly show the halftone dither pattern, and how the pixel clumps within the dither pattern grow into each other as the grey-level changes.

  # Halftone Screen (45 degree angle)
  convert gradient.png   -ordered-dither h4x4a      od_h4x4a.gif
  convert gradient.png   -ordered-dither h6x6a      od_h6x6a.gif
  convert gradient.png   -ordered-dither h8x8a      od_h8x8a.gif
  # Halftone Screen (orthogonal)
  convert gradient.png   -ordered-dither h4x4o      od_h4x4o.gif
  convert gradient.png   -ordered-dither h6x6o      od_h6x6o.gif
  convert gradient.png   -ordered-dither h8x8o      od_h8x8o.gif
  convert gradient.png   -ordered-dither h16x16o    od_h16x16o.gif
  # Circle Halftones (black and white)
  convert gradient.png   -ordered-dither c7x7b      od_c7x7b.gif
  convert gradient.png   -ordered-dither c7x7w      od_c7x7w.gif
[IM Output] [IM Output] [IM Output]
[IM Output] [IM Output] [IM Output] [IM Output]
[IM Output] [IM Output]

Up until ImageMagick version 6.2.9 all the above threshold ordered dither map were all that was possible with IM. This has now changed, allowing users to add their own patterns, add even contribute them to the IM community.

The 'Circle' halftone thresholds were added by Glenn Randers-Pehrson, IM v6.6.5-6.

Offset HalfTone Dither

The only problem with the above halftone dither is that it is that the exact same threshold map (tile) is applied to all the color channels in the same way. That means the same set of primary color is arranged in dots with the same 'center'.

To get what is known as 'Offset Printing' the threshold pattern is rotated in a particular pattern such that the colors form small scale 'rosette patterns' that destroies any posible and more horrible looking interference patterns that you could otherwise develop.

This diagram basically explains the process.
[IM Output]

Note however that the rotated screens do not tile very well, as such the best idea is to actually generate the rotated pattern directly, rather than use a tiles threshold pattern.

Here is one way to give an image a offset-halftone printing look, using small rotated 2x2 pixel checkerboard pattern, which is about the smallest 'screen' that can be used.

  convert colorwheel.png  -set option:distort:viewport '%wx%h+0+0' \
          -colorspace CMYK -separate null: \
          \( -size 2x2 xc: \( +clone -negate \) \
                +append \( +clone -negate \) -append \) \
          -virtual-pixel tile -filter gaussian \
          \( +clone -distort SRT 60 \) +swap \
          \( +clone -distort SRT 30 \) +swap \
          \( +clone -distort SRT 45 \) +swap \
          \( +clone -distort SRT 0 \)  +swap +delete \
          -compose Overlay -layers composite \
          -set colorspace CMYK -combine -colorspace RGB \
          offset_colorwheel.png
[IM Output]

Note that the four rotated 'screens' are applied to the image as a whole, it is only the "-combine" step in CMYK colorspace, that actually extracts the 4 different color channels from the screened images.

Also the 'no-op' distort for the last 'black' channel is important as it will blur the input checker pattern according to the Gaussian filter that was used on the other channels during their rotations, even though that screen itself is not being rotated.

And here I use the scaling function of the SRT Distort used to generate the rotated tiles to create a slightly larger and more blurry 'screen pattern'.

  convert parrots_med.png  -set option:distort:viewport '%wx%h+0+0' \
          -colorspace CMYK -separate null: \
          \( -size 2x2 xc: \( +clone -negate \) \
                +append \( +clone -negate \) -append \) \
          -virtual-pixel tile -filter gaussian \
          \( +clone -distort SRT 2,60 \) +swap \
          \( +clone -distort SRT 2,30 \) +swap \
          \( +clone -distort SRT 2,45 \) +swap \
          \( +clone -distort SRT 2,0  -blur 0x0.7 \) +swap +delete \
          -compose Overlay -layers composite \
          -set colorspace CMYK -combine -colorspace RGB \
          offset_parrots.png
[IM Output]

Note that the pattern remains very 'square-like' especially the black screen which which all the others are derived.

FUTURE POSSIBILITY: Replace the 2 pixel checkboard in the above with "pattern:gray50" pixel level checkerboard pattern. Gaussian filter options can be used to adjust the bluriness of the scaled pattern. Alternatively, you can blurry scale the pattern, and thresholded it, to make rounder dots. This can then be rotated as previously to create the 4 color screens.

It also would be better is a larger screen using a pattern of hexagonal dots could be used, rather than the checkerboard pattern I have used above.


It is important to note that this is not really generating color dots like true offset printing, but faking it by simply multiplying the color screens against the original image. You can see that this is the case due to the sharp color change along the edges of the red parrot against the green background. True offset printing using only dots of pure color will not have any middle of dot color changes. It is the size of the pure colored dot that should change, depending on the average of the colors in that area reprended by the dot in the source image.

To actually generate an true offset printing image only containing round dots in each color channel of appropriate size would require a lot more work. The average color in each dot in each color channel will been to be determined, and from that the appropriate size of colored dot (anti-aliased circle) generated. Anyone like to give it a go?

The above was from a discussion in the IM forum discussion CMYK Halftone Effect which looks at how photoshop 'fakes it', and how ImageMagick could achieve the same effect.

This discussion is also related to B/W Halftone Dither which takes a closer look at generating true halftone screens using actual dots of appropriate sized. That discussion however did not take it to the next step of using offset (rotated) screens. Such screens would probably require rotating the image to generate the dots, then rotating the dot pattern back again for that specific color channel.

XML Threshold Maps

From IM version 6.3.0, rather than using a fixed set of map that were built into the source code of IM (shown previously), the maps are now read in from a set of XML data files external to the program itself.

As part of this change you can now list the available 'threshold maps' that "-ordered-dither" operator can use.

  identify -list threshold
[IM Output]

The above list shows not only the threshold maps available, but also the aliases provided for backwards compatibility or alternate naming, and those defined in my own personal "thresholds.xml" XML data file (saved into the ".magick" sub-directory of my home).

When "-ordered-dither" is looking for a map, the first map that it finds in the above list, will be used. As such you can not override a 'system defined threshold pattern.

The system file "thresholds.xml" (the path of which is given by the above "-list" option), contains a complete summary of the format of the XML file. A format that is simple enough (with error checks by IM) to allow users to define and create their own ordered dither threshold maps.

For example here is a copy of the 'diag5x5' threshold map I defined in my personal "threshold.xml" file.
[IM Output]

If you look it creates a simple 5x5 map of a single diagonal line that becomes thicker as the threshold level increases. The level numbers in the map goes from 0 to 5, one less than the divisor, which declares how many 'grays' it needs to divide the color gradient into.

Here is a gradient dithered using this personal threshold map.

  convert gradient.png   -ordered-dither diag      od_diag.gif
[IM Output]

And here is an example of using that threshold to dither the alpha channel of a simple shadowed image, a purpose for which I designed it.

  convert -size 70x60 xc:none -font Candice -pointsize 50 \
          -fill black -annotate +10+45 'A' -channel RGBA  -blur 0x5 \
          -fill white -stroke black -draw "text 5,40 'A'"   shadow.png

  convert shadow.png  -channel A  -ordered-dither diag   shadow_diag.gif
[IM Output] ==> [IM Output]

Pretty cool hey! More on dithering alpha channels later. First I need to also show how to use the coloring abilities of expanded "-ordered-dither" operator.

Ordered Dither using Uniform Color Levels

With the release of IM v6.3.0, not only was the threshold maps used by "-ordered-dither" changed so as to be read from external files, but the internal operations was enhanced to allow it to use mathematically defined 'posterized' color maps.

This means you can generate a more deterministic dithering of images than you can achieve with 'error correction dithering'. This is especially important for color reductions involving animations as you will not get problems from color differences between frames.

The posterization levels is passed to the "-ordered-dither" argument using a extra comma separated list of numbers appended to the name of the threshold map to use. If no numbers are provided then the operator falls back to the normal 2 color (or posterize level 1) color map.

For example and argument of 'checks,6' will use a classic Web Safe Color Map (posterized level 6) color map (also defined by the "netscape: built-in colormap image). However as the minimal dither map of 'checks' is used a single extra level of dithering is added between each of the 6 color levels creating 11 pseudo-levels of colors in each channel of the image.

In other words even though only 6 levels of color per channel is being used (producing 6^3 or 216 colors) the single dither pattern between levels increases the dither to and effective 11 levels (producing and effective 11^3 or 1331 colors).

For example, here is a gray scale gradient dithered using 6 grey levels and various threshold maps. The first map 'threshold' is a special non-dithering ordered dither threshold map, showing just the colors used.


  convert gradient.png   -ordered-dither threshold,6  od_threshold_6.gif
  convert gradient.png   -ordered-dither checks,6     od_checks_6.gif
  convert gradient.png   -ordered-dither o2x2,6       od_o2x2_6.gif
  convert gradient.png   -ordered-dither o4x4,6       od_o4x4_6.gif
  convert gradient.png   -ordered-dither o8x8,6       od_o8x8_6.gif
[IM Output] [IM Output] [IM Output] [IM Output] [IM Output]
As you can see even though only 6 colors are used, with ordered dithering you increase the effective number of colors used to define the gradient, to a point where you can be hard pressed to notice just how few colors were actually used!

Not only can you define the number of posterization levels for all channels, but unlike the "-posterize" error correction dither option, you can specify the levels for each channel. The numbers are assigned to the channels as per the "-channels" setting.

For example here we dithered the gradient using a special 332 color map (8 levels or red and green, 4 of blue) which defines a total of 256 colors.

  convert gradient.png   -ordered-dither o8x8,8,8,4   od_o8x8_884.gif
[IM Output]

Because of the different number of color levels per channel, the above image does not contain just pure grey colors, but includes some bluish and yellowish pixels which cancels each other out to produce extra levels of greys.

Now compare the O-dithered version against error correction dithered version using a posterization levels of 2, and 6, and a "332 colormap" (8 levels of red and green, 4 of blue).

  convert logo.png  -ordered-dither o8x8        logo_o8x8_2.gif
  convert logo.png  -posterize 2                logo_posterize_2.gif
  convert logo.png  -ordered-dither o8x8,6      logo_o8x8_6.gif
  convert logo.png  -posterize 6                logo_posterize_6.gif
  convert logo.png  -ordered-dither o8x8,8,8,4  logo_o8x8_332.gif
  convert logo.png  -remap colormap_332.png     logo_remap_332.gif
[IM Output] [IM Output]     [IM Output] [IM Output]     [IM Output] [IM Output]

The first image of each pair in the above is mathematically ordered dithered, while the second is pseudo-randomly 'error correction' dithered.

The last pair uses the special '332 colormap' (See Generating Color Maps), which is regarded as probably the best posterize colormap for general images with a 256 color limit. The odd difference in channel levels producing a slightly better shading of colors for this cartoon like image.

It was to allow the production the '332 colormap' that the "-ordered-dither" operator, included the ability to specify separate levels for each color channel.

Better Ordered Dither Results

Lets take a closer look at the level 6 O-Dither we just generated.

  convert logo.png -ordered-dither o8x8,6 -format %k info:
[IM Text]

As you can see for this image we did not even come close to filling the GIF color table (256 limit). Basically as the image generally consists of mostly blue colors, very few shades of red or even green colors from a level 6 uniform colormap was even used.

However by increasing the number of posterization level we can fill the GIF color table better, so as to produce a better O-Dithered Image.

  convert logo.png -ordered-dither o8x8,13 -format %k info:
[IM Text]

This produces enough colors to be only slightly smaller than the GIF color table limits. With the increase in the number of colors the result looks a lot better that the results of a simple standard uniform colormap.

  convert logo.png -ordered-dither o8x8,13    logo_o8x8_13.gif
[IM Output]

As you can see with a high 'levels' value, "-ordered-dither" can produce images that are comparable to color quantized, to the specific color picking generated by color quantization and an error correction dither.

The major point about these images, is not that they are of high quality. After all a full Color Quantization can more easily produce better color map for the image. But that the low level dither pattern within the image is fixed, regardless of any small changes that may occur. Only the area changes will be changed in the Ordered Dithered Image.

That is, they do not have the E-Dither Sensitivity to change that causes problems for Frame Optimization in GIF animations. (see Optimization Problem)

Of course for an animation, you will need to "-append" all the images together, before checking how many colors is actually used. And you will need to use the special "+remap" option after using the "-ordered-dither" to force IM to generate a 'common global colormap' for ALL the images, even though you have already performed color reduction and dithering.

This method of determining the number of color levels is not simple to determine, but it does works. I hope to work out a way for IM to determine the best level automatically, especially for GIF animations.


DIY Dither Patterns and Threshold Maps

Previously I show you that the new "-ordered-dither" operator can accept a user defined dithering pattern. Here I am going to show you how you can create your own dither pattern. Specifically a special pattern I found useful for generating a shadows consisting of horizontal lines.

Multi-Image Dither Patterns

The first thing you need to do is create a set of images defining the pattern you want to create. The pattern should start with a solid black image of the right size as the first image (all pixels off) and a solid white image at the other end (all pixels on).

The next image should be the middle 50% grey pattern, defining the the basic style of the dithering you are trying to achieve.

For example here is my initial DIY dither pattern. Which I save into a multi-image GIF file (not a GIF animation)...

  convert -size 2x2 xc:black \
          \( +clone -draw 'fill white line 0,0 1,0' \) \
          xc:white     dpat_hlines2x2.gif
  montage dpat_hlines2x2.gif    -tile x1 -background none -frame 2 \
          -filter box  -geometry 32x32+5+0    dpat_hlines2x2_imgs.gif
[IM Output]

This is about the simplest set of dither pattern images you can get, and is very similar to the 'checks' or 'Checkerboard Dither', but with horizontal lines, rather than a checker pattern.

So you can see just what this dither pattern would look like, here is a rather simple DIY ordered dither, that makes direct use of the threshold dithering image set.

  convert gradient.png   dpat_hlines2x2.gif \
          -virtual-pixel tile  -fx 'u[floor((n-1)*u)+1]' dgrad_hlines2x2.gif
[IM Output]

As you can see the dither pattern is nothing fancy. The "-fx" function is a variation of the Color Lookup Tables function, namely, an IM Dither Lookup Patterns' type function. And with a "-virtual-pixel" setting of 'tile', the function does not even need to know the size of the dither pattern image you are using.

The use of "-virtual-pixel" by the "-fx" operator using calculated indexes like this was broken before IM version 6.2.9-2.

Lets try this dither pattern set again but using a simple shadowed image...

  convert shadow.png dpat_hlines2x2.gif  -channel A \
          -virtual-pixel tile  -fx 'u[floor((n-1)*u)+1].g' \
          shadow_dpat_hlines2x2.gif
[IM Output] ==> [IM Output]

DIY Ordered Dither Threshold Maps

The above DIY dither pattern is about as simple a dither pattern as you can get, and as such we can convert it directly into a XML threshold map, so that the fast built-in "-ordered-dither" operator can make use of it.

Here is the final XML definition, which I saved in my own personal threshold map file "~/.magick/thresholds.xml" of my "$HOME" directory.
[IM Output]

The XML format is very simple, and defines a 2x2 pixel map. The first black image is given a value of zero, and has no pixels, so no zero values is present. The pixels turned on (made white) in the middle image is set to '1' and the remaining or second image pixels are given a value of '2'.

The 'divisor=' defines the number of images, or pseudo-color levels (fake color levels) this dither pattern represents, so it has a value of '3'. It divides the pixel values, to defined the color level at which that pixel is to be turned on. As such the top two pixels are turned on for colors larger than 1/3, while the bottom two are turned on for color values larger that 2/3. That is each pixel value represents a 'threshold' level, and why dither patterns are also called threshold maps.

The rest of the definition defines the names (and optional alias) by which you can refer to the threshold map for the ordered dither operator.

So lets try it out...

  convert gradient.png  -ordered-dither hlines2x2  od_hlines2x2.gif
  convert shadow.png  -channel A \
          -ordered-dither hlines2x2   shadow_hlines2x2.gif
[IM Output]
[IM Output] ==> [IM Output]

As you can see the result is reasonably good, but we can do other things to improve the result.

By adjusting the threshold values in the map, we can change the boundary's, so it does not divide the color space into 3 equal areas...
[IM Output]

Note how I increased the divisor to '10', so as to divide the color levels into ten equal sections. I then changed the threshold settings so that the pattern starts at a 30% threshold at the transparent end (black), to 90% for fully opaque (white). And here is the results of changing the threshold map.

  convert gradient.png  -ordered-dither hlines2x2a  od_hlines2x2a.gif
  convert shadow.png -channel A \
          -ordered-dither hlines2x2a  shadow_hlines2x2a.gif
[IM Output]
[IM Output]

As you can see this widened the range of semi-transparent pixels that use pure horizontal lines as a dither pattern. This gives a better better shadow effect, though it probably should only be used with a less fuzzy shadow than the example used here.

Note however that this type of change to a threshold is very uncommon. Though justified for the intended use in this case. Basically it does not properly define a gradient, or allow for lighter and darker shades of patterns. For that we need to make a much more complex threshold map, with more pixels, and more patterns.

DIY Horizontal Line Dither

Here I expanded the simple horizontal lines dither pattern I created above into a set of patterns, to produce a smoother gradient from 'off' to 'on'. This was the result.

  montage dpat_hlines.gif   -filter box   -geometry 60x20+2+0 \
          -tile x1 -background none  -frame 2   dpat_hlines_images.gif
  convert gradient.png  dpat_hlines.gif  \
          -virtual-pixel tile  -fx 'u[floor((n-1)*u)+1]' \
          dgrad_dpat_hlines.gif
  convert shadow.png dpat_hlines.gif  -channel A \
          -virtual-pixel tile  -fx 'u[floor((n-1)*u)+1].g' \
          shadow_dpat_hlines.gif
[IM Output]
[IM Output]
[IM Output] ==> [IM Output]

As you can see it now consists of 9, 12x4 pixel images. It does not represent all the pixel patterns that you could have, but this enhances the effect of the lines. Also I doubled its height so as to jitter the gaps in the lines appropriately.

Here is another example of using this dither pattern...

  convert -size 120x55 xc:white  -draw 'fill #777 ellipse 50,43 30,5 0,360' \
          -motion-blur 0x15+180   -blur 0x2      sphere_shadow.png
  convert sphere_shadow.png dpat_hlines.gif \
          -virtual-pixel tile  -fx 'u[floor((n-1)*u)+1]' \
          sphere_shadow_dither.gif
  convert sphere_shadow_dither.gif   -fill red  -stroke firebrick \
          -draw 'circle 35,25 35,5'     sphere_shadow_hlines.gif
[IM Output] ==> [IM Output] ==> [IM Output]

The next step is to convert this set of dither patterns into single threshold map image, rather than a set of multiple images. This is achieved by using some fancy image manipulations to merge all the images together.

  convert -size 1x10 gradient: -flip -crop 1x1 +repage -delete 0,-1 \
          -scale 12x4\! null: \( dpat_hlines.gif -delete 0 \) \
          +matte -compose CopyOpacity -layers Composite \
          -reverse -compose Over -flatten +matte dmap_hlines.png
[IM Output]

The value '10' is one more than the number of images in the dither pattern while "-scale 12x4\!" is the size of the dither patterns being converted into a threshold map.

The result is a grey-scale map, with no pure black or white colors. The grey level used for a pixel means that if the color level is that grey value or higher, then that pixel should be turned on. That is each grey level is the 'threshold' level at which the color value goes from black to white.

If you like to look at the image in another way, dark pixels generally result in those pixels being turn on for more color levels. While light pixels are only turned on when the image color becomes very bright. This is almost a negation of what the image actually looks like, but if you think about it it does make sense.

I also used a PNG image rather than a GIF image for the map as only one image needs to be saved, and more importantly, to attempt to preserve 16 bit quality levels for the threshold values. GIF can only handle 8 bit color levels.

Now we can dither our images using, just a single image, and a much simpler threshold comparison of each pixel directly against the dithering threshold image (or map).

  convert gradient.png dmap_hlines.png \
          -virtual-pixel tile  -fx 'u>=v'   dgrad_dmap_hlines.gif
[IM Output] ==> [IM Output]

See how much simpler a threshold map is. You have only one image, and one direct comparison to do per pixel, for each channel that is being dithered. This makes dithering using a threshold map very very fast. Much faster that full color quantization. This simplicity is the why ImageMagick and most graphics software use a threshold map to hold various dither patterns.

The greater-then or equal ('>=') test was not added to the "-fx" operator, until IM version 6.2.9-2. If this is a problem, use the inverted test 'v<u' in the above.

This simplicity however becomes a lot more complicated if the user wanted to dither using multiple color levels. The proof of concept of this was first worked out in the examples on the page Posterized Ordered Dither before being incorporated into the IM core functions.

Now that we have a merged threshold image, we next need to convert the above image into a XML threshold map, that IM can directly read, and the "-ordered-dither" operator can use.

To do this we need to output our image as numbers representing the 9 grey levels it represents. This is best done using the NetPBM or PBMplus image format, with a depth adjustment using the "NetPbm" image processing software. This package is generally a standard linux install so most people will have it already, or can install it from there normal software distribution.

The "pnmdepth" number is again the number of gray levels that the threshold image contains.

  convert dmap_hlines.png pgm:- | pnmdepth 9 | pnmnoraw > dmap_hlines.pgm
[IM Text]

All the numbers (other than the 'P2' image magic identifier) in the above are the numbers needed to generate the appropriate 'threshold map', that you can add to your personal "thresholds.xml" file. For example here is the resulting threshold map entry created from the above.
[IM Output]

And here is an example of using this threshold map.

  convert shadow.png  -channel A  -ordered-dither hlines   shadow_hlines.gif
[IM Output] ==> [IM Output]

And that is how you can generate a complex threshold map from a progression of images.

Dithering with Symbol Patterns

Now while you can use a single threshold map or threshold image, instead of a multi-image pattern set for most dithering operations, that does not mean that multi-image maps don't have there own uses.

You can use a set of lookup image to tile multiple areas all at once, rather than one at a time. For example by scaling a simple image, and then replace each pixel in an image with a particular symbol.

For example, here I take the very small 'eyes' image [IM Output] and replace the individual pixels with various symbols, to produce such a pattern for each pixel in the original image.

  montage dpat_symbols.gif   -geometry +5+0 \
          -tile x1 -background none -mattecolor blue  -frame 3 \
          dpat_syms_images.gif
  convert eyes.gif -alpha off -colorspace sRGB -grayscale Average \
          +matte -scale 1600% -negate  \
          dpat_symbols.gif -virtual-pixel tile -fx 'u[floor(15.9999*u)+1]' \
          eyes_syms.gif
[IM Output]
[IM Output]

The montage is used to expand the multi-image GIF image so that you can see it's contents, without it being 'animated'.

You can adjust which "-grayscale" intensity method you want to use, from the normal 'Rec709Luminance' to a darker 'Rec709Luma', or use a 'average' of either a non-linear 'sRGB colorspace, or linear 'RGB' colorspace. You can even adjust the "-gamma" scaling of the values to get the best spread of colors.

There are a lot of posibilities, and what is good depends more on your symbol arrangement than actual method chosen. The key with the above is to somehow ensure each color in the input image produces a unqiue symbol, and that can be very tricky to achieve.

This example can be used for creating cross-stitch or knitting guides that hobbists can follow, generating larger scale artwork from smaller comuter images.

You can use this technique to tile greyscale images with a set of tiling color images. The result is a bit like the landscape maps seen in many old computer war games.

  montage dpat_map.gif   -geometry +5+0 -tile x1  -background none  \
          dpat_map_images.gif
  convert -seed 100 \
          -size 200x200 plasma:'gray(50%)-gray(50%)' -blur 0x15 \
          -channel G -auto-level +channel -set colorspace sRGB \
          dpat_map.gif -virtual-pixel tile  -fx 'u[floor(5.999*u.g)+1]' \
          map.gif
[IM Output]
[IM Output]

Note that I needed to ensure that IM thought of the grayscale image as being already in the final sRGB colorspace (as the tile images are), even though it is actually linear RGB data that is being used for FX index lookups. Without this the resulting 'map' is skewed toward wooded landscapes, with little chance of water areas.

As you can see any set of images can be used for the tiles, the images don't even need to line up with each other, or even be the same size tile. Of course if the tiles are the same size and are closely related to each other, as in the case of the 3 blue 'sea' tiles, the tiled pattern can 'flow' from one tiled area into another.

By replacing the tile with images of numbers you can also generate a sort of paint by numbers guide. However some extra processing may be needed to border the different areas. This is left as an exercise, mail your solutions to me, and you can have your name in IM examples as the author for this technique.


Ordered Dither Random notes and Future possibilities

Under Construction

Ordered Dither with small numbers of colors.

When using a small number of colors for a small image a pseudo-random dither like IM's hilbert curve error correction dither, or even a simpler Floyd-Steinberg error correction dither (see Error Correction Dither above ) produces a horrible looking result.

Ideally an Ordered Dither should be used with low color and small icon like images to produce a much better looking result. However at the moment ordered dither in IM can only used 'fixed' mathematically generated color tables, and not just a collection of 'best' colors.

Ordered Dither with a map of any colors

There is some algorithm that will allow you to use a set of specific colors for ordered dithering.

Basically by adding the 'pseudo-colors' that an ordered dither algorithm can generate (which could involve three color dithering), to the given color map, to 'fill it out'. You can then 'map' the individual pixels of an image to this 'expanded color map'.

From that initial mapping to a specific pseudo-color) the actual color can be picked from the threshold map that generates that pseudo-color, and thus order dither areas of that color, and thus the whole image to a given set of colors.

Because of my background in icons, and now GIF animations, I definitely would like to see an any color ordered dither implemented, but I have yet to find a practical reference to how to do ordered dithers with a fixed set of colors.


Created: 11 August 2006
Updated: 8 May 2010
Author: Anthony Thyssen, <A.Thyssen@griffith.edu.au>
Examples Generated with: [version image]
URL: http://www.imagemagick.org/Usage/quantize/