Architecture

The Pixel Cache • Image Properties and Profiles • Multispectral Imagery • Large Image Support • Streaming Pixels • Threads of Execution • Heterogeneous Distributed Processing • Custom Image Coders • Custom Image Filters

The citizens of Oz were quite content with their benefactor, the all-powerful Wizard. They accepted his wisdom and benevolence without ever questioning the who, why, and where of his power. Like the citizens of Oz, if you feel comfortable that ImageMagick can help you convert, edit, or compose your images without knowing what goes on behind the curtain, feel free to skip this section. However, if you want to know more about the software and algorithms behind ImageMagick, read on. To fully benefit from this discussion, you should be comfortable with image nomenclature and be familiar with computer programming.

Architecture Overview

An image typically consists of a rectangular region of pixels and metadata. To convert, edit, or compose an image in an efficient manner, we need convenient access to any pixel anywhere within the region (and sometimes outside the region). And in the case of an image sequence, we need access to any pixel of any region of any image in the sequence. However, there are hundreds of image formats such JPEG, TIFF, PNG, GIF, etc., that makes it difficult to access pixels on demand. Within these formats we find differences in:

• colorspace (e.g sRGB, linear RGB, linear GRAY, CMYK, YUV, Lab, etc.)
• bit depth (.e.g 1, 4, 8, 12, 16, etc.)
• storage format (e.g. unsigned, signed, float, double, etc.)
• compression (e.g. uncompressed, RLE, Zip, BZip, etc.)
• orientation (i.e. top-to-bottom, right-to-left, etc.),
• layout (.e.g. raw, interspersed with opcodes, etc.)

In addition, some image pixels may require attenuation, some formats permit more than one frame, and some formats contain vector graphics that must first be rasterized (converted from vector to pixels).

An efficient implementation of an image processing algorithm may require we get or set:

• one pixel a time (e.g. pixel at location 10,3)
• a single scanline (e.g. all pixels from row 4)
• a few scanlines at once (e.g. pixel rows 4-7)
• a single column or columns of pixels (e.g. all pixels from column 11)
• an arbitrary region of pixels from the image (e.g. pixels defined at 10,7 to 10,19)
• a pixel in random order (e.g. pixel at 14,15 and 640,480)
• pixels from two different images (e.g. pixel at 5,1 from image 1 and pixel at 5,1 from image 2)
• pixels outside the boundaries of the image (e.g. pixel at -1,-3)
• a pixel component that is unsigned (65311) or in a floating-point representation (e.g. 0.17836)
• a high-dynamic range pixel that can include negative values (e.g. -0.0072973525628) as well as values that exceed the quantum depth (e.g. 65931)
• one or more pixels simultaneously in different threads of execution
• all the pixels in memory to take advantage of speed-ups offered by executing in concert across heterogeneous platforms consisting of CPUs, GPUs, and other processors
• traits associated with each channel to specify whether the pixel channel is copied, updated, or blended
• masks that define which pixels are eligible to be updated
• extra channels that benefits the user but otherwise remain untouched by ImageMagick image processing algorithms

Given the varied image formats and image processing requirements, we implemented the ImageMagick pixel cache to provide convenient sequential or parallel access to any pixel on demand anywhere inside the image region (i.e. authentic pixels) and from any image in a sequence. In addition, the pixel cache permits access to pixels outside the boundaries defined by the image (i.e. virtual pixels).

In addition to pixels, images have a plethora of image properties and profiles. Properties include the well known attributes such as width, height, depth, and colorspace. An image may have optional properties which might include the image author, a comment, a create date, and others. Some images also include profiles for color management, or EXIF, IPTC, 8BIM, or XMP informational profiles. ImageMagick provides command line options and programming methods to get, set, or view image properties or profiles or apply profiles.

ImageMagick consists of nearly a half million lines of C code and optionally depends on several million lines of code in dependent libraries (e.g. JPEG, PNG, TIFF libraries). Given that, one might expect a huge architecture document. However, a great majority of image processing is simply accessing pixels and its metadata and our simple, elegant, and efficient implementation makes this easy for the ImageMagick developer. We discuss the implementation of the pixel cache and getting and setting image properties and profiles in the next few sections. Next, we discuss using ImageMagick within a thread of execution. In the final sections, we discuss image coders to read or write a particular image format followed by a few words on creating a filter to access or update pixels based on your custom requirements.

The Pixel Cache

The ImageMagick pixel cache is a repository for image pixels with up to 64 channels. The channels are stored contiguously at the depth specified when ImageMagick was built. The channel depths are 8 bits-per-pixel component for the Q8 version of ImageMagick, 16 bits-per-pixel component for the Q16 version, and 32 bits-per-pixel component for the Q32 version. By default pixel components are 32-bit floating-bit high dynamic-range quantities. The channels can hold any value but typically contain red, green, blue, and alpha intensities or cyan, magenta, yellow, black and alpha intensities. A channel might contain the colormap indexes for colormapped images or the black channel for CMYK images. The pixel cache storage may be heap memory, disk-backed memory mapped, or on disk. The pixel cache is reference-counted. Only the cache properties are copied when the cache is cloned. The cache pixels are subsequently copied only when you signal your intention to update any of the pixels.

Create the Pixel Cache

The pixel cache is associated with an image when it is created and it is initialized when you try to get or put pixels. Here are three common methods to associate a pixel cache with an image:

Create an image canvas initialized to the background color:

image=AllocateImage(image_info);
if (SetImageExtent(image,640,480) == MagickFalse)
  { /* an exception was thrown */ }
(void) QueryMagickColor("red",&image->background_color,&image->exception);
SetImageBackgroundColor(image);

Create an image from a JPEG image on disk:

(void) strcpy(image_info->filename,"image.jpg"):
image=ReadImage(image_info,exception);
if (image == (Image *) NULL)
  { /* an exception was thrown */ }

Create an image from a memory based image:

image=BlobToImage(blob_info,blob,extent,exception);
if (image == (Image *) NULL)
  { /* an exception was thrown */ }

In our discussion of the pixel cache, we use the MagickCore API to illustrate our points, however, the principles are the same for other program interfaces to ImageMagick.

When the pixel cache is initialized, pixels are scaled from whatever bit depth they originated from to that required by the pixel cache. For example, a 1-channel 1-bit monochrome PBM image is scaled to 8-bit gray image, if you are using the Q8 version of ImageMagick, and 16-bit RGBA for the Q16 version. You can determine which version you have with the -version option:

$ identify -version
Version: ImageMagick 7.1.1-30 2024-02-11 Q16 https://im.awm.jp

As you can see, the convenience of the pixel cache sometimes comes with a trade-off in storage (e.g. storing a 1-bit monochrome image as 16-bit is wasteful) and speed (i.e. storing the entire image in memory is generally slower than accessing one scanline of pixels at a time). In most cases, the benefits of the pixel cache typically outweigh any disadvantages.

Access the Pixel Cache

Once the pixel cache is associated with an image, you typically want to get, update, or put pixels into it. We refer to pixels inside the image region as authentic pixels and outside the region as virtual pixels. Use these methods to access the pixels in the cache:

• GetVirtualPixels(): gets pixels that you do not intend to modify or pixels that lie outside the image region (e.g. pixel @ -1,-3)
• GetAuthenticPixels(): gets pixels that you intend to modify
• QueueAuthenticPixels(): queue pixels that you intend to set
• SyncAuthenticPixels(): update the pixel cache with any modified pixels

Here is a typical MagickCore code snippet for manipulating pixels in the pixel cache. In our example, we copy pixels from the input image to the output image and decrease the intensity by 10%:

const Quantum
  *p;

Quantum
  *q;

ssize_t
  x,
  y;

destination=CloneImage(source,source->columns,source->rows,MagickTrue,exception);
if (destination == (Image *) NULL)
  { /* an exception was thrown */ }
for (y=0; y < (ssize_t) source->rows; y++)
{
  p=GetVirtualPixels(source,0,y,source->columns,1,exception);
  q=GetAuthenticPixels(destination,0,y,destination->columns,1,exception);
  if ((p == (const Quantum *) NULL) || (q == (Quantum *) NULL)
    break;
  for (x=0; x < (ssize_t) source->columns; x++)
  {
    SetPixelRed(image,90*p->red/100,q);
    SetPixelGreen(image,90*p->green/100,q);
    SetPixelBlue(image,90*p->blue/100,q);
    SetPixelAlpha(image,90*p->opacity/100,q);
    p+=GetPixelChannels(source);
    q+=GetPixelChannels(destination);
  }
  if (SyncAuthenticPixels(destination,exception) == MagickFalse)
    break;
}
if (y < (ssize_t) source->rows)
  { /* an exception was thrown */ }

When we first create the destination image by cloning the source image, the pixel cache pixels are not copied. They are only copied when you signal your intentions to modify or set the pixel cache by calling GetAuthenticPixels() or QueueAuthenticPixels(). Use QueueAuthenticPixels() if you want to set new pixel values rather than update existing ones. You could use GetAuthenticPixels() to set pixel values but it is slightly more efficient to use QueueAuthenticPixels() instead. Finally, use SyncAuthenticPixels() to ensure any updated pixels are pushed to the pixel cache.

You can associate arbitrary content with each pixel, called meta content. Use GetVirtualMetacontent() (to read the content) or GetAuthenticMetacontent() (to update the content) to gain access to this content. For example, to print the metacontent, use:

const void
  *metacontent;

for (y=0; y < (ssize_t) source->rows; y++)
{
  p=GetVirtualPixels(source,0,y,source->columns,1);
  if (p == (const Quantum *) NULL)
    break;
  metacontent=GetVirtualMetacontent(source);
  /* print meta content here */
}
if (y < (ssize_t) source->rows)
  /* an exception was thrown */

The pixel cache manager decides whether to give you direct or indirect access to the image pixels. In some cases the pixels are staged to an intermediate buffer-- and that is why you must call SyncAuthenticPixels() to ensure this buffer is pushed out to the pixel cache to guarantee the corresponding pixels in the cache are updated. For this reason we recommend that you only read or update a scanline or a few scanlines of pixels at a time. However, you can get any rectangular region of pixels you want. GetAuthenticPixels() requires that the region you request is within the bounds of the image area. For a 640 by 480 image, you can get a scanline of 640 pixels at row 479 but if you ask for a scanline at row 480, an exception is returned (rows are numbered starting at 0). GetVirtualPixels() does not have this constraint. For example,

p=GetVirtualPixels(source,-3,-3,source->columns+3,6,exception);

gives you the pixels you asked for without complaint, even though some are not within the confines of the image region.

Virtual Pixels

There are a plethora of image processing algorithms that require a neighborhood of pixels about a pixel of interest. The algorithm typically includes a caveat concerning how to handle pixels around the image boundaries, known as edge pixels. With virtual pixels, you do not need to concern yourself about special edge processing other than choosing which virtual pixel method is most appropriate for your algorithm.

Access to the virtual pixels are controlled by the SetImageVirtualPixelMethod() method from the MagickCore API or the -virtual-pixel option from the command line. The methods include:

Cache Storage and Resource Requirements

Recall that this simple and elegant design of the ImageMagick pixel cache comes at a cost in terms of storage and processing speed. The pixel cache storage requirements scales with the area of the image and the bit depth of the pixel components. For example, if we have a 640 by 480 image and we are using the non-HDRI Q16 version of ImageMagick, the pixel cache consumes image width * height * bit-depth / 8 * channels bytes or approximately 2.3 mebibytes (i.e. 640 * 480 * 2 * 4). Not too bad, but what if your image is 25000 by 25000 pixels? The pixel cache requires approximately 4.7 gibibytes of storage. Ouch. ImageMagick accounts for possible huge storage requirements by caching large images to disk rather than memory. Typically the pixel cache is stored in memory using heap memory. If heap memory is exhausted, we create the pixel cache on disk and attempt to memory-map it. If memory-map memory is exhausted, we simply use standard disk I/O. Disk storage is plentiful and cheap, but it is also very slow-- upwards of 1000 times slower than accessing pixels in memory. We can get some speed improvements, up to 5 times, if we memory-map the disk-based cache. These decisions about storage are made automagically by the pixel cache manager negotiating with the operating system. However, you can influence how the pixel cache manager allocates the pixel cache with cache resource limits. The limits include:

Note, these limits pertain to the ImageMagick pixel cache. Certain algorithms within ImageMagick do not respect these limits nor does any of the external delegate libraries (e.g. JPEG, TIFF, etc.).

To determine the current setting of these limits, use this command:

-> identify -list resource
Resource limits:
  Width: 100MP
  Height: 100MP
  Area: 25.181GB
  Memory: 11.726GiB
  Map: 23.452GiB
  Disk: unlimited
  File: 768
  Thread: 12
  Throttle: 0
  Time: unlimited

You can set these limits either as a security policy (see policy.xml), with an environment variable, with the -limit command line option, or with the SetMagickResourceLimit() MagickCore API method. As an example, our online web interface to ImageMagick, MagickStudio, includes these policy limits to help prevent a denial-of-service:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE policymap [
<!ELEMENT policymap (policy)*>
<!ATTLIST policymap xmlns CDATA #FIXED "">
<!ELEMENT policy EMPTY>
<!ATTLIST policy xmlns CDATA #FIXED "">
<!ATTLIST policy domain NMTOKEN #REQUIRED>
<!ATTLIST policy name NMTOKEN #IMPLIED>
<!ATTLIST policy pattern CDATA #IMPLIED>
<!ATTLIST policy rights NMTOKEN #IMPLIED>
<!ATTLIST policy stealth NMTOKEN #IMPLIED>
<!ATTLIST policy value CDATA #IMPLIED>
]>
<!--
  Creating a security policy that fits your specific local environment
  before making use of ImageMagick is highly advised. You can find guidance on
  setting up this policy at ../www/security-policy.html,
  and it's important to verify your policy using the validation tool located
  at https://imagemagick-secevaluator.doyensec.com/.


  Secure ImageMagick security policy:

  This stringent security policy prioritizes the implementation of
  rigorous controls and restricted resource utilization to establish a
  profoundly secure setting while employing ImageMagick. It deactivates
  conceivably hazardous functionalities, including specific coders like
  SVG or HTTP. The policy promotes the tailoring of security measures to
  harmonize with the requirements of the local environment and the guidelines
  of the organization. This protocol encompasses explicit particulars like
  limitations on memory consumption, sanctioned pathways for reading and
  writing, confines on image sequences, the utmost permissible duration of
  workflows, allocation of disk space intended for image data, and even an
  undisclosed passphrase for remote connections. By adopting this robust
  policy, entities can elevate their overall security stance and alleviate
  potential vulnerabilities.
-->
<policymap>
  <!-- Set maximum parallel threads. -->
  <policy domain="resource" name="thread" value="2"/>
  <!-- Set maximum time in seconds. When this limit is exceeded, an exception
       is thrown and processing stops. -->
  <policy domain="resource" name="time" value="120"/>
  <!-- Set maximum number of open pixel cache files. When this limit is
       exceeded, any subsequent pixels cached to disk are closed and reopened
       on demand. -->
  <policy domain="resource" name="file" value="768"/>
  <!-- Set maximum amount of memory in bytes to allocate for the pixel cache
       from the heap. When this limit is exceeded, the image pixels are cached
       to memory-mapped disk. -->
  <policy domain="resource" name="memory" value="256MiB"/>
  <!-- Set maximum amount of memory map in bytes to allocate for the pixel
       cache. When this limit is exceeded, the image pixels are cached to
       disk. -->
  <policy domain="resource" name="map" value="512MiB"/>
  <!-- Set the maximum width * height of an image that can reside in the pixel
       cache memory. Images that exceed the area limit are cached to disk. -->
  <policy domain="resource" name="area" value="16KP"/>
  <!-- Set maximum amount of disk space in bytes permitted for use by the pixel
       cache. When this limit is exceeded, the pixel cache is not be created
       and an exception is thrown. -->
  <policy domain="resource" name="disk" value="1GiB"/>
  <!-- Set the maximum length of an image sequence.  When this limit is
       exceeded, an exception is thrown. -->
  <policy domain="resource" name="list-length" value="32"/>
  <!-- Set the maximum width of an image.  When this limit is exceeded, an
       exception is thrown. -->
  <policy domain="resource" name="width" value="8KP"/>
  <!-- Set the maximum height of an image.  When this limit is exceeded, an
       exception is thrown. -->
  <policy domain="resource" name="height" value="8KP"/>
  <!-- Periodically yield the CPU for at least the time specified in
       milliseconds. -->
  <!--  -->
  <!-- Do not create temporary files in the default shared directories, instead
       specify a private area to store only ImageMagick temporary files. -->
  <!--  -->
  <!-- Force memory initialization by memory mapping select memory
       allocations. -->
  <policy domain="cache" name="memory-map" value="anonymous"/>
  <!-- Ensure all image data is fully flushed and synchronized to disk. -->
  <policy domain="cache" name="synchronize" value="true"/>
  <!-- Replace passphrase for secure distributed processing -->
  <!--  -->
  <!-- Do not permit any delegates to execute. -->
  <policy domain="delegate" rights="none" pattern="*"/>
  <!-- Do not permit any image filters to load. -->
  <policy domain="filter" rights="none" pattern="*"/>
  <!-- Don't read/write from/to stdin/stdout. -->
  <policy domain="path" rights="none" pattern="-"/>
  <!-- don't read sensitive paths. -->
  <policy domain="path" rights="none" pattern="/etc/*"/>
  <!-- Indirect reads are not permitted. -->
  <policy domain="path" rights="none" pattern="@*"/>
  <!-- These image types are security risks on read, but write is fine -->
  <policy domain="module" rights="write" pattern="{MSL,MVG,PS,SVG,URL,XPS}"/>
  <!-- This policy sets the number of times to replace content of certain
       memory buffers and temporary files before they are freed or deleted. -->
  <policy domain="system" name="shred" value="1"/>
  <!-- Enable the initialization of buffers with zeros, resulting in a minor
       performance penalty but with improved security. -->
  <policy domain="system" name="memory-map" value="anonymous"/>
  <!-- Set the maximum amount of memory in bytes that is permitted for
       allocation requests. -->
  <policy domain="system" name="max-memory-request" value="256MiB"/>
</policymap>

Since we process multiple simultaneous sessions, we don't want any one session consuming all the available memory. With this policy, large images are cached to disk. If the image is too large and exceeds the pixel cache disk limit, the program exits. In addition, we place a time limit to prevent any run-away processing tasks. If any one image has a width or height that exceeds 8192 pixels, an exception is thrown and processing stops. As of ImageMagick 7.0.1-8, you can prevent the use of any delegate or all delegates (set the pattern to "*"). Note, prior to this release, use a domain of "coder" to prevent delegate usage (e.g. domain="coder" rights="none" pattern="HTTPS"). The policy also prevents indirect reads. If you want to, for example, read text from a file (e.g. caption:@myCaption.txt), you'll need to remove this policy.

Note, the cache limits are global to each invocation of ImageMagick, meaning if you create several images, the combined resource requirements are compared to the limit to determine the pixel cache storage disposition.

To determine which type and how much resources are consumed by the pixel cache, add the -debug cache option to the command-line:

$ magick -debug cache logo: -sharpen 3x2 null:
2016-12-17T13:33:42-05:00 0:00.000 0.000u 7.0.0 Cache magick: cache.c/DestroyPixelCache/1275/Cache
  destroy 
2016-12-17T13:33:42-05:00 0:00.000 0.000u 7.0.0 Cache magick: cache.c/OpenPixelCache/3834/Cache
  open LOGO[0] (Heap Memory, 640x480x4 4.688MiB)
2016-12-17T13:33:42-05:00 0:00.010 0.000u 7.0.0 Cache magick: cache.c/OpenPixelCache/3834/Cache
  open LOGO[0] (Heap Memory, 640x480x3 3.516MiB)
2016-12-17T13:33:42-05:00 0:00.010 0.000u 7.0.0 Cache magick: cache.c/ClonePixelCachePixels/1044/Cache
  Memory => Memory
2016-12-17T13:33:42-05:00 0:00.020 0.010u 7.0.0 Cache magick: cache.c/ClonePixelCachePixels/1044/Cache
  Memory => Memory
2016-12-17T13:33:42-05:00 0:00.020 0.010u 7.0.0 Cache magick: cache.c/OpenPixelCache/3834/Cache
  open LOGO[0] (Heap Memory, 640x480x3 3.516MiB)
2016-12-17T13:33:42-05:00 0:00.050 0.100u 7.0.0 Cache magick: cache.c/DestroyPixelCache/1275/Cache
  destroy LOGO[0]
2016-12-17T13:33:42-05:00 0:00.050 0.100u 7.0.0 Cache magick: cache.c/DestroyPixelCache/1275/Cache
  destroy LOGO[0]

This command utilizes a pixel cache in memory. The logo consumed 4.688MiB and after it was sharpened, 3.516MiB.

Distributed Pixel Cache

A distributed pixel cache is an extension of the traditional pixel cache available on a single host. The distributed pixel cache may span multiple servers so that it can grow in size and transactional capacity to support very large images. Start up the pixel cache server on one or more machines. When you read or operate on an image and the local pixel cache resources are exhausted, ImageMagick contacts one or more of these remote pixel servers to store or retrieve pixels. The distributed pixel cache relies on network bandwidth to marshal pixels to and from the remote server. As such, it will likely be significantly slower than a pixel cache utilizing local storage (e.g. memory, disk, etc.).

magick -distribute-cache 6668 &  // start on 192.168.100.50
magick -define registry:cache:hosts=192.168.100.50:6668 myimage.jpg -sharpen 5x2 mimage.png

Cache Views

GetVirtualPixels(), GetAuthenticPixels(), QueueAuthenticPixels(), and SyncAuthenticPixels(), from the MagickCore API, can only deal with one pixel cache area per image at a time. Suppose you want to access the first and last scanline from the same image at the same time? The solution is to use a cache view. A cache view permits you to access as many areas simultaneously in the pixel cache as you require. The cache view methods are analogous to the previous methods except you must first open a view and close it when you are finished with it. Here is a snippet of MagickCore code that permits us to access the first and last pixel row of the image simultaneously:

CacheView
  *first_row,
  *last_row;

first_row=AcquireVirtualCacheView(source,exception);
last_row=AcquireVirtualCacheView(source,exception);
for (y=0; y < (ssize_t) source->rows; y++)
{
  const Quantum
    *p,
    *q;

  p=GetCacheViewVirtualPixels(first_row,0,y,source->columns,1,exception);
  q=GetCacheViewVirtualPixels(last_row,0,source->rows-y-1,source->columns,1,exception);
  if ((p == (const Quantum *) NULL) || (q == (const Quantum *) NULL))
    break;
  for (x=0; x < (ssize_t) source->columns; x++)
  {
    /* do something with p & q here */
  }
}
last_row=DestroyCacheView(last_row);
first_row=DestroyCacheView(first_row);
if (y < (ssize_t) source->rows)
  { /* an exception was thrown */ }

Magick Pixel Cache Format

Recall that each image format is decoded by ImageMagick and the pixels are deposited in the pixel cache. If you write an image, the pixels are read from the pixel cache and encoded as required by the format you are writing (e.g. GIF, PNG, etc.). The Magick Pixel Cache (MPC) format is designed to eliminate the overhead of decoding and encoding pixels to and from an image format. MPC writes two files. One, with the extension .mpc, retains all the properties associated with the image or image sequence (e.g. width, height, colorspace, etc.) and the second, with the extension .cache, is the pixel cache in the native raw format. When reading an MPC image file, ImageMagick reads the image properties and memory maps the pixel cache on disk eliminating the need for decoding the image pixels. The trade-off is in disk space. MPC is generally larger in file size than most other image formats.

The most efficient use of MPC image files is a write-once, read-many-times pattern. For example, your workflow requires extracting random blocks of pixels from the source image. Rather than re-reading and possibly decompressing the source image each time, we use MPC and map the image directly to memory.

Pixel Cache Recommended Practices

Although you can request any pixel from the pixel cache, any block of pixels, any scanline, multiple scanlines, any row, or multiple rows with the GetVirtualPixels(), GetAuthenticPixels(), QueueAuthenticPixels, GetCacheViewVirtualPixels(), GetCacheViewAuthenticPixels(), and QueueCacheViewAuthenticPixels() methods, ImageMagick is optimized to return a few pixels or a few pixels rows at time. There are additional optimizations if you request a single scanline or a few scanlines at a time. These methods also permit random access to the pixel cache, however, ImageMagick is optimized for sequential access. Although you can access scanlines of pixels sequentially from the last row of the image to the first, you may get a performance boost if you access scanlines from the first row of the image to the last, in sequential order.

You can get, modify, or set pixels in row or column order. However, it is more efficient to access the pixels by row rather than by column.

If you update pixels returned from GetAuthenticPixels() or GetCacheViewAuthenticPixels(), don't forget to call SyncAuthenticPixels() or SyncCacheViewAuthenticPixels() respectively to ensure your changes are synchronized with the pixel cache.

Use QueueAuthenticPixels() or QueueCacheViewAuthenticPixels() if you are setting an initial pixel value. The GetAuthenticPixels() or GetCacheViewAuthenticPixels() method reads pixels from the cache and if you are setting an initial pixel value, this read is unnecessary. Don't forget to call SyncAuthenticPixels() or SyncCacheViewAuthenticPixels() respectively to push any pixel changes to the pixel cache.

GetVirtualPixels(), GetAuthenticPixels(), QueueAuthenticPixels(), and SyncAuthenticPixels() are slightly more efficient than their cache view counter-parts. However, cache views are required if you need access to more than one region of the image simultaneously or if more than one thread of execution is accessing the image.

You can request pixels outside the bounds of the image with GetVirtualPixels() or GetCacheViewVirtualPixels(), however, it is more efficient to request pixels within the confines of the image region.

Although you can force the pixel cache to disk using appropriate resource limits, disk access can be upwards of 1000 times slower than memory access. For fast, efficient, access to the pixel cache, try to keep the pixel cache in heap memory.

The ImageMagick Q16 version of ImageMagick permits you to read and write 16 bit images without scaling but the pixel cache consumes twice as many resources as the Q8 version. If your system has constrained memory or disk resources, consider the Q8 version of ImageMagick. In addition, the Q8 version typically executes faster than the Q16 version.

A great majority of image formats and algorithms restrict themselves to a fixed range of pixel values from 0 to some maximum value, for example, the Q16 version of ImageMagick permit intensities from 0 to 65535. High dynamic-range imaging (HDRI), however, permits a far greater dynamic range of exposures (i.e. a large difference between light and dark areas) than standard digital imaging techniques. HDRI accurately represents the wide range of intensity levels found in real scenes ranging from the brightest direct sunlight to the deepest darkest shadows. Enable HDRI at ImageMagick build time to deal with high dynamic-range images, but be mindful that each pixel component is a 32-bit floating point value. In addition, pixel values are not clamped by default so some algorithms may have unexpected results due to out-of-band pixel values than the non-HDRI version.

If you are dealing with large images, make sure the pixel cache is written to a disk area with plenty of free space. Under Linux, this is typically /tmp and for Windows, c:/temp. You can tell ImageMagick to write the pixel cache to an alternate location and conserve memory with these options:

magick -limit memory 2GB -limit map 4GB -define registry:temporary-path=/data/tmp ...

Set global resource limits for your environment in the policy.xml configuration file.

If you plan on processing the same image many times, consider the MPC format. Reading a MPC image has near-zero overhead because its in the native pixel cache format eliminating the need for decoding the image pixels. Here is an example:

magick image.tif image.mpc
magick image.mpc -crop 100x100+0+0 +repage 1.png
magick image.mpc -crop 100x100+100+0 +repage 2.png
magick image.mpc -crop 100x100+200+0 +repage 3.png

MPC is ideal for web sites. It reduces the overhead of reading and writing an image. We use it exclusively at our online image studio.

Image Properties and Profiles

Images have metadata associated with them in the form of properties (e.g. width, height, description, etc.) and profiles (e.g. EXIF, IPTC, color management). ImageMagick provides convenient methods to get, set, or update image properties and get, set, update, or apply profiles. Some of the more popular image properties are associated with the Image structure in the MagickCore API. For example:

(void) printf("image width: %lu, height: %lu\n",image->columns,image->rows);

For a great majority of image properties, such as an image comment or description, we use the GetImageProperty() and SetImageProperty() methods. Here we set a property and fetch it right back:

const char
  *comment;

(void) SetImageProperty(image,"comment","This space for rent");
comment=GetImageProperty(image,"comment");
if (comment == (const char *) NULL)
  (void) printf("Image comment: %s\n",comment);

ImageMagick supports artifacts with the GetImageArtifact() and SetImageArtifact() methods. Artifacts are stealth properties that are not exported to image formats (e.g. PNG).

Image profiles are handled with GetImageProfile(), SetImageProfile(), and ProfileImage() methods. Here we set a profile and fetch it right back:

StringInfo
  *profile;

profile=AcquireStringInfo(length);
SetStringInfoDatum(profile,my_exif_profile);
(void) SetImageProfile(image,"EXIF",profile);
DestroyStringInfo(profile);
profile=GetImageProfile(image,"EXIF");
if (profile != (StringInfo *) NULL)
  (void) PrintStringInfo(stdout,"EXIF",profile);

Multispectral Imagery

ImageMagick supports multispectral imagery where all channels have the same dimensions and number of pixels as the original image. However, not all image formats support multispectral images. PSD, TIFF, MIFF, MPC, and FTXT have full support for multispectral images up to 31 bands, 21 of them meta channels. Note, if you build ImageMagick with the configure script --enable-64bit-channel-masks option, you can process 62 band multispectral images with up to 52 meta channels.

If you have a use case that is not currently supported by an image format, post it to the discussion forum. There is a good chance, we can support your use case in a future release of ImageMagick.

Streaming Pixels

ImageMagick provides for streaming pixels as they are read from or written to an image. This has several advantages over the pixel cache. The time and resources consumed by the pixel cache scale with the area of an image, whereas the pixel stream resources scale with the width of an image. The disadvantage is the pixels must be consumed as they are streamed so there is no persistence.

Use ReadStream() or WriteStream() with an appropriate callback method in your MagickCore program to consume the pixels as they are streaming. Here's an abbreviated example of using ReadStream:

static size_t StreamPixels(const Image *image,const void *pixels,const size_t columns)
{
  register const Quantum
    *p;

  MyData
    *my_data;

  my_data=(MyData *) image->client_data;
  p=(Quantum *) pixels;
  if (p != (const Quantum *) NULL)
    {
      /* process pixels here */
    }
  return(columns);
}

...

/* invoke the pixel stream here */
image_info->client_data=(void *) MyData;
image=ReadStream(image_info,&StreamPixels,exception);

We also provide a lightweight tool, stream, to stream one or more pixel components of the image or portion of the image to your choice of storage formats. It writes the pixel components as they are read from the input image a row at a time making stream desirable when working with large images or when you require raw pixel components. A majority of the image formats stream pixels (red, green, and blue) from left to right and top to bottom. However, a few formats do not support this common ordering (e.g. the PSD format).

Large Image Support

ImageMagick has the capability to handle image sizes spanning from mega- to tera-pixels, encompassing reading, processing, and writing operations. In theory, image dimensions can extend up to 31 million rows/columns on a 32-bit operating system and up to a whopping 31 trillion on a 64-bit OS. However, the actual achievable dimensions are substantially less, contingent on the resources available on your host computer. It's essential to be aware that certain image formats impose limitations on image size. For instance, Photoshop images are constrained to a maximum of 300,000 pixels in width or height. Here we resize an image to a quarter million pixels square:

magick logo: -resize 250000x250000 logo.miff

For large images, memory resources will likely be exhausted and ImageMagick will instead create a pixel cache on disk. Make sure you have plenty of temporary disk space. If your default temporary disk partition is too small, tell ImageMagick to use another partition with plenty of free space. For example:

magick -define registry:temporary-path=/data/tmp logo:  \ 
     -resize 250000x250000 logo.miff

To ensure large images do not consume all the memory on your system, force the image pixels to memory-mapped disk with resource limits:

magick -define registry:temporary-path=/data/tmp -limit memory 16mb \
  logo: -resize 250000x250000 logo.miff

Here we force all image pixels to disk:

magick -define registry:temporary-path=/data/tmp -limit area 0 \
  logo: -resize 250000x250000 logo.miff

Caching pixels to disk is about 1000 times slower than memory. Expect long run times when processing large images on disk with ImageMagick. You can monitor progress with this command:

magick -monitor -limit memory 2GiB -limit map 4GiB -define registry:temporary-path=/data/tmp \
  logo: -resize 250000x250000 logo.miff

For really large images, or if there is limited resources on your host, you can utilize a distributed pixel cache on one or more remote hosts:

magick -distribute-cache 6668 &  // start on 192.168.100.50
magick -distribute-cache 6668 &  // start on 192.168.100.51
magick -limit memory 2mb -limit map 2mb -limit disk 2gb \
  -define registry:cache:hosts=192.168.100.50:6668,192.168.100.51:6668 \
  myhugeimage.jpg -sharpen 5x2 myhugeimage.png

Due to network latency, expect a substantial slow-down in processing your workflow.

Threads of Execution

Many of ImageMagick's internal algorithms are threaded to take advantage of speed-ups offered by the multicore processor chips. However, you are welcome to use ImageMagick algorithms in your threads of execution with the exception of the MagickCore's GetVirtualPixels(), GetAuthenticPixels(), QueueAuthenticPixels(), or SyncAuthenticPixels() pixel cache methods. These methods are intended for one thread of execution only with the exception of an OpenMP parallel section. To access the pixel cache with more than one thread of execution, use a cache view. We do this for the CompositeImage() method, for example. Suppose we want to composite a single source image over a different destination image in each thread of execution. If we use GetVirtualPixels(), the results are unpredictable because multiple threads would likely be asking for different areas of the pixel cache simultaneously. Instead we use GetCacheViewVirtualPixels() which creates a unique view for each thread of execution ensuring our program behaves properly regardless of how many threads are invoked. The other program interfaces, such as the MagickWand API, are completely thread safe so there are no special precautions for threads of execution.

Here is an MagickCore code snippet that takes advantage of threads of execution with the OpenMP programming paradigm:

CacheView
  *image_view;

MagickBooleanType
  status;

ssize_t
  y;

/*
  Acquire a cache view to enable parallelism.
*/
status=MagickTrue;
image_view=AcquireVirtualCacheView(image,exception);
#pragma omp parallel for schedule(static,4) shared(status)
for (y=0; y < (ssize_t) image->rows; y++)
{
  register Quantum
    *q;

  register ssize_t
    x;

  register void
    *metacontent;

  if (status == MagickFalse)
    continue;
  /*
    Get a row of pixels.
  */
  q=GetCacheViewAuthenticPixels(image_view,0,y,image->columns,1,exception);
  if (q == (Quantum *) NULL)
    {
      status=MagickFalse;
      continue;
    }
  metacontent=GetCacheViewAuthenticMetacontent(image_view);
  for (x=0; x < (ssize_t) image->columns; x++)
  {
    /*
      Set the pixel color.
    */
    SetPixelRed(image,...,q);
    SetPixelGreen(image,...,q);
    SetPixelBlue(image,...,q);
    SetPixelAlpha(image,...,q);
    if (metacontent != NULL)
      metacontent[indexes+x]=...;
    q+=GetPixelChannels(image);
  }
  /*
    Sync the updated pixels to the pixel cache.
  */
  if (SyncCacheViewAuthenticPixels(image_view,exception) == MagickFalse)
    status=MagickFalse;
}
/*
  Destroy the cache view.
*/
image_view=DestroyCacheView(image_view);
if (status == MagickFalse)
  perror("something went wrong");

This code snippet converts an uncompressed Windows bitmap to a Magick++ image:

#include "Magick++.h"
#include <assert.h>
#include "omp.h"

void ConvertBMPToImage(const BITMAPINFOHEADER *bmp_info,
  const unsigned char *restrict pixels,Magick::Image *image)
{
  /*
    Prepare the image so that we can modify the pixels directly.
  */
  assert(bmp_info->biCompression == BI_RGB);
  assert(bmp_info->biWidth == image->columns());
  assert(abs(bmp_info->biHeight) == image->rows());
  image->modifyImage();
  if (bmp_info->biBitCount == 24)
    image->type(MagickCore::TrueColorType);
  else
    image->type(MagickCore::TrueColorMatteType);
  register unsigned int bytes_per_row=bmp_info->biWidth*bmp_info->biBitCount/8;
  if (bytes_per_row % 4 != 0) {
    bytes_per_row=bytes_per_row+(4-bytes_per_row % 4);  // divisible by 4.
  }
  /*
    Copy all pixel data, row by row.
  */
  #pragma omp parallel for
  for (int y=0; y < int(image->rows()); y++)
  {
    int
      row;

    register const unsigned char
      *restrict p;

    register MagickCore::Quantum
      *restrict q;

    row=(bmp_info->biHeight > 0) ? (image->rows()-y-1) : y;
    p=pixels+row*bytes_per_row;
    q=image->setPixels(0,y,image->columns(),1);
    for (int x=0; x < int(image->columns()); x++)
    {
      SetPixelBlue(image,p[0],q);
      SetPixelGreen(image,p[1],q);
      SetPixelRed(image,p[2],q);
      if (bmp_info->biBitCount == 32) {
        SetPixelAlpha(image,p[3],q);
      }
      q+=GetPixelChannels(image);
      p+=bmp_info->biBitCount/8;
    }
    image->syncPixels();  // sync pixels to pixel cache.
  }
  return;
}

If you call the ImageMagick API from your OpenMP-enabled application and you intend to dynamically increase the number of threads available in subsequent parallel regions, be sure to perform the increase before you call the API otherwise ImageMagick may fault.

MagickWand supports wand views. A view iterates over the entire, or portion, of the image in parallel and for each row of pixels, it invokes a callback method you provide. This limits most of your parallel programming activity to just that one module. There are similar methods in MagickCore. For an example, see the same sigmoidal contrast algorithm implemented in both MagickWand and MagickCore.

In most circumstances, the default number of threads is set to the number of processor cores on your system for optimal performance. However, if your system is hyperthreaded or if you are running on a virtual host and only a subset of the processors are available to your server instance, you might get an increase in performance by setting the thread policy or the MAGICK_THREAD_LIMIT environment variable. For example, your virtual host has 8 processors but only 2 are assigned to your server instance. The default of 8 threads can cause severe performance problems. One solution is to limit the number of threads to the available processors in your policy.xml configuration file:

<policy domain="resource" name="thread" value="2"/>

Or suppose your 12 core hyperthreaded computer defaults to 24 threads. Set the MAGICK_THREAD_LIMIT environment variable and you will likely get improved performance:

export MAGICK_THREAD_LIMIT=12

The OpenMP committee has not defined the behavior of mixing OpenMP with other threading models such as Posix threads. However, using modern releases of Linux, OpenMP and Posix threads appear to interoperate without complaint. If you want to use Posix threads from a program module that calls one of the ImageMagick application programming interfaces (e.g. MagickCore, MagickWand, Magick++, etc.) from Mac OS X or an older Linux release, you may need to disable OpenMP support within ImageMagick. Add the --disable-openmp option to the configure script command line and rebuild and reinstall ImageMagick.

You can further increase performance by reducing lock contention with the tcmalloc memory allocation library. To enable, add --with-tcmalloc to the configure command-line when you build ImageMagick.

Threading Performance

It can be difficult to predict behavior in a parallel environment. Performance might depend on a number of factors including the compiler, the version of the OpenMP library, the processor type, the number of cores, the amount of memory, whether hyperthreading is enabled, the mix of applications that are executing concurrently with ImageMagick, or the particular image-processing algorithm you utilize. The only way to be certain of optimal performance, in terms of the number of threads, is to benchmark. ImageMagick includes progressive threading when benchmarking a command and returns the elapsed time and efficiency for one or more threads. This can help you identify how many threads is the most efficient in your environment. For this benchmark we sharpen a 1920x1080 image of a model 10 times with 1 to 12 threads:

$ magick -bench 10 model.png -sharpen 5x2 null:
Performance[1]: 10i 1.135ips 1.000e 8.760u 0:08.810
Performance[2]: 10i 2.020ips 0.640e 9.190u 0:04.950
Performance[3]: 10i 2.786ips 0.710e 9.400u 0:03.590
Performance[4]: 10i 3.378ips 0.749e 9.580u 0:02.960
Performance[5]: 10i 4.032ips 0.780e 9.580u 0:02.480
Performance[6]: 10i 4.566ips 0.801e 9.640u 0:02.190
Performance[7]: 10i 3.788ips 0.769e 10.980u 0:02.640
Performance[8]: 10i 4.115ips 0.784e 12.030u 0:02.430
Performance[9]: 10i 4.484ips 0.798e 12.860u 0:02.230
Performance[10]: 10i 4.274ips 0.790e 14.830u 0:02.340
Performance[11]: 10i 4.348ips 0.793e 16.500u 0:02.300
Performance[12]: 10i 4.525ips 0.799e 18.320u 0:02.210

The sweet spot for this example is 6 threads. This makes sense since there are 6 physical cores. The other 6 are hyperthreads. It appears that sharpening does not benefit from hyperthreading.

In certain cases, it might be optimal to set the number of threads to 1 or to disable OpenMP completely with the MAGICK_THREAD_LIMIT environment variable, -limit command line option, or the policy.xml configuration file.

Heterogeneous Distributed Processing

ImageMagick includes support for heterogeneous distributed processing with the OpenCL framework. OpenCL kernels within ImageMagick permit image processing algorithms to execute across heterogeneous platforms consisting of CPUs, GPUs, and other processors. Depending on your platform, speed-ups can be an order of magnitude faster than the traditional single CPU.

First verify that your version of ImageMagick includes support for the OpenCL feature:

magick identify -version
Features: DPC Cipher Modules OpenCL OpenMP(4.5)

If so, run this command to realize a significant speed-up for image convolution:

magick image.png -convolve '-1, -1, -1, -1, 9, -1, -1, -1, -1' convolve.png

If an accelerator is not available or if the accelerator fails to respond, ImageMagick reverts to the non-accelerated convolution algorithm.

Here is an example OpenCL kernel that convolves an image:

static inline long ClampToCanvas(const long offset,const ulong range)
{
  if (offset < 0L)
    return(0L);
  if (offset >= range)
    return((long) (range-1L));
  return(offset);
}

static inline CLQuantum ClampToQuantum(const float value)
{
  if (value < 0.0)
    return((CLQuantum) 0);
  if (value >= (float) QuantumRange)
    return((CLQuantum) QuantumRange);
  return((CLQuantum) (value+0.5));
}

__kernel void Convolve(const __global CLPixelType *source,__constant float *filter,
  const ulong width,const ulong height,__global CLPixelType *destination)
{
  const ulong columns = get_global_size(0);
  const ulong rows = get_global_size(1);

  const long x = get_global_id(0);
  const long y = get_global_id(1);

  const float scale = (1.0/QuantumRange);
  const long mid_width = (width-1)/2;
  const long mid_height = (height-1)/2;
  float4 sum = { 0.0, 0.0, 0.0, 0.0 };
  float gamma = 0.0;
  register ulong i = 0;

  for (long v=(-mid_height); v <= mid_height; v++)
  {
    for (long u=(-mid_width); u <= mid_width; u++)
    {
      register const ulong index=ClampToCanvas(y+v,rows)*columns+ClampToCanvas(x+u,
        columns);
      const float alpha=scale*(QuantumRange-source[index].w);
      sum.x+=alpha*filter[i]*source[index].x;
      sum.y+=alpha*filter[i]*source[index].y;
      sum.z+=alpha*filter[i]*source[index].z;
      sum.w+=filter[i]*source[index].w;
      gamma+=alpha*filter[i];
      i++;
    }
  }

  gamma=1.0/(fabs(gamma) <= MagickEpsilon ? 1.0 : gamma);
  const ulong index=y*columns+x;
  destination[index].x=ClampToQuantum(gamma*sum.x);
  destination[index].y=ClampToQuantum(gamma*sum.y);
  destination[index].z=ClampToQuantum(gamma*sum.z);
  destination[index].w=ClampToQuantum(sum.w);
};

See MagickCore/accelerate.c for a complete implementation of image convolution with an OpenCL kernel.

Note, that under Windows, you might have an issue with TDR (Timeout Detection and Recovery of GPUs). Its purpose is to detect runaway tasks hanging the GPU by using an execution time threshold. For some older low-end GPUs running the OpenCL filters in ImageMagick, longer execution times might trigger the TDR mechanism and pre-empt the GPU image filter. When this happens, ImageMagick automatically falls back to the CPU code path and returns the expected results. To avoid pre-emption, increase the TdrDelay registry key.

Custom Image Coders

An image coder (i.e. encoder / decoder) is responsible for registering, optionally classifying, optionally reading, optionally writing, and unregistering one image format (e.g. PNG, GIF, JPEG, etc.). Registering an image coder alerts ImageMagick a particular format is available to read or write. While unregistering tells ImageMagick the format is no longer available. The classifying method looks at the first few bytes of an image and determines if the image is in the expected format. The reader sets the image size, colorspace, and other properties and loads the pixel cache with the pixels. The reader returns a single image or an image sequence (if the format supports multiple images per file), or if an error occurs, an exception and a null image. The writer does the reverse. It takes the image properties and unloads the pixel cache and writes them as required by the image format.

Here is a listing of a sample custom coder. It reads and writes images in the MGK image format which is simply an ID followed by the image width and height followed by the RGB pixel values.

#include <MagickCore/studio.h>
#include <MagickCore/blob.h>
#include <MagickCore/cache.h>
#include <MagickCore/colorspace.h>
#include <MagickCore/exception.h>
#include <MagickCore/image.h>
#include <MagickCore/list.h>
#include <MagickCore/magick.h>
#include <MagickCore/memory_.h>
#include <MagickCore/monitor.h>
#include <MagickCore/pixel-accessor.h>
#include <MagickCore/string_.h>
#include <MagickCore/module.h>
#include "filter/blob-private.h"
#include "filter/exception-private.h"
#include "filter/image-private.h"
#include "filter/monitor-private.h"
#include "filter/quantum-private.h"

/*
  Forward declarations.
*/
static MagickBooleanType
  WriteMGKImage(const ImageInfo *,Image *,ExceptionInfo *);

/*
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                                                                             %
%                                                                             %
%                                                                             %
%   I s M G K                                                                 %
%                                                                             %
%                                                                             %
%                                                                             %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%  IsMGK() returns MagickTrue if the image format type, identified by the
%  magick string, is MGK.
%
%  The format of the IsMGK method is:
%
%      MagickBooleanType IsMGK(const unsigned char *magick,const size_t length)
%
%  A description of each parameter follows:
%
%    o magick: This string is generally the first few bytes of an image file
%      or blob.
%
%    o length: Specifies the length of the magick string.
%
*/
static MagickBooleanType IsMGK(const unsigned char *magick,const size_t length)
{
  if (length < 7)
    return(MagickFalse);
  if (LocaleNCompare((char *) magick,"id=mgk",7) == 0)
    return(MagickTrue);
  return(MagickFalse);
}

/*
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                                                                             %
%                                                                             %
%                                                                             %
%   R e a d M G K I m a g e                                                   %
%                                                                             %
%                                                                             %
%                                                                             %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%  ReadMGKImage() reads a MGK image file and returns it.  It allocates the
%  memory necessary for the new Image structure and returns a pointer to the
%  new image.
%
%  The format of the ReadMGKImage method is:
%
%      Image *ReadMGKImage(const ImageInfo *image_info,
%        ExceptionInfo *exception)
%
%  A description of each parameter follows:
%
%    o image_info: the image info.
%
%    o exception: return any errors or warnings in this structure.
%
*/
static Image *ReadMGKImage(const ImageInfo *image_info,ExceptionInfo *exception)
{
  char
    buffer[MaxTextExtent];

  Image
    *image;

  long
    y;

  MagickBooleanType
    status;

  register long
    x;

  register Quantum
    *q;

  register unsigned char
    *p;

  ssize_t
    count;

  unsigned char
    *pixels;

  unsigned long
    columns,
    rows;

  /*
    Open image file.
  */
  assert(image_info != (const ImageInfo *) NULL);
  assert(image_info->signature == MagickCoreSignature);
  if (image_info->debug != MagickFalse)
    (void) LogMagickEvent(TraceEvent,GetMagickModule(),"%s",
      image_info->filename);
  assert(exception != (ExceptionInfo *) NULL);
  assert(exception->signature == MagickCoreSignature);
  image=AcquireImage(image_info,exception);
  status=OpenBlob(image_info,image,ReadBinaryBlobMode,exception);
  if (status == MagickFalse)
    {
      image=DestroyImageList(image);
      return((Image *) NULL);
    }
  /*
    Read MGK image.
  */
  (void) ReadBlobString(image,buffer);  /* read magic number */
  if (IsMGK(buffer,7) == MagickFalse)
    ThrowReaderException(CorruptImageError,"ImproperImageHeader");
  (void) ReadBlobString(image,buffer);
  count=(ssize_t) sscanf(buffer,"%lu %lu\n",&columns,&rows);
  if (count <= 0)
    ThrowReaderException(CorruptImageError,"ImproperImageHeader");
  do
  {
    /*
      Initialize image structure.
    */
    image->columns=columns;
    image->rows=rows;
    image->depth=8;
    if ((image_info->ping != MagickFalse) && (image_info->number_scenes != 0))
      if (image->scene >= (image_info->scene+image_info->number_scenes-1))
        break;
    /*
      Convert MGK raster image to pixel packets.
    */
    if (SetImageExtent(image,image->columns,image->rows,exception) == MagickFalse)
      return(DestroyImageList(image));
    pixels=(unsigned char *) AcquireQuantumMemory((size_t) image->columns,
      3UL*sizeof(*pixels));
    if (pixels == (unsigned char *) NULL)
      ThrowReaderException(ResourceLimitError,"MemoryAllocationFailed");
    for (y=0; y < (long) image->rows; y++)
    {
      count=(ssize_t) ReadBlob(image,(size_t) (3*image->columns),pixels);
      if (count != (ssize_t) (3*image->columns))
        ThrowReaderException(CorruptImageError,"UnableToReadImageData");
      p=pixels;
      q=QueueAuthenticPixels(image,0,y,image->columns,1,exception);
      if (q == (Quantum *) NULL)
        break;
      for (x=0; x < (long) image->columns; x++)
      {
        SetPixelRed(image,ScaleCharToQuantum(*p++),q);
        SetPixelGreen(image,ScaleCharToQuantum(*p++),q);
        SetPixelBlue(image,ScaleCharToQuantum(*p++),q);
        q+=GetPixelChannels(image);
      }
      if (SyncAuthenticPixels(image,exception) == MagickFalse)
        break;
      if (image->previous == (Image *) NULL)
        if ((image->progress_monitor != (MagickProgressMonitor) NULL) &&
            (QuantumTick(y,image->rows) != MagickFalse))
          {
            status=image->progress_monitor(LoadImageTag,y,image->rows,
              image->client_data);
            if (status == MagickFalse)
              break;
          }
    }
    pixels=(unsigned char *) RelinquishMagickMemory(pixels);
    if (EOFBlob(image) != MagickFalse)
      {
        ThrowFileException(exception,CorruptImageError,"UnexpectedEndOfFile",
          image->filename);
        break;
      }
    /*
      Proceed to next image.
    */
    if (image_info->number_scenes != 0)
      if (image->scene >= (image_info->scene+image_info->number_scenes-1))
        break;
    *buffer='\0';
    (void) ReadBlobString(image,buffer);
    count=(ssize_t) sscanf(buffer,"%lu %lu\n",&columns,&rows);
    if (count > 0)
      {
        /*
          Allocate next image structure.
        */
        AcquireNextImage(image_info,image,exception);
        if (GetNextImageInList(image) == (Image *) NULL)
          {
            image=DestroyImageList(image);
            return((Image *) NULL);
          }
        image=SyncNextImageInList(image);
        if (image->progress_monitor != (MagickProgressMonitor) NULL)
          {
            status=SetImageProgress(image,LoadImageTag,TellBlob(image),
              GetBlobSize(image));
            if (status == MagickFalse)
              break;
          }
      }
  } while (count > 0);
  (void) CloseBlob(image);
  return(GetFirstImageInList(image));
}

/*
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                                                                             %
%                                                                             %
%                                                                             %
%   R e g i s t e r M G K I m a g e                                           %
%                                                                             %
%                                                                             %
%                                                                             %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%  RegisterMGKImage() adds attributes for the MGK image format to
%  the list of supported formats.  The attributes include the image format
%  tag, a method to read and/or write the format, whether the format
%  supports the saving of more than one frame to the same file or blob,
%  whether the format supports native in-memory I/O, and a brief
%  description of the format.
%
%  The format of the RegisterMGKImage method is:
%
%      unsigned long RegisterMGKImage(void)
%
*/
ModuleExport unsigned long RegisterMGKImage(void)
{
  MagickInfo
    *entry;

  entry=AcquireMagickInfo("MGK","MGK","MGK image");
  entry->decoder=(DecodeImageHandler *) ReadMGKImage;
  entry->encoder=(EncodeImageHandler *) WriteMGKImage;
  entry->magick=(IsImageFormatHandler *) IsMGK;
  (void) RegisterMagickInfo(entry);
  return(MagickImageCoderSignature);
}

/*
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                                                                             %
%                                                                             %
%                                                                             %
%   U n r e g i s t e r M G K I m a g e                                       %
%                                                                             %
%                                                                             %
%                                                                             %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%  UnregisterMGKImage() removes format registrations made by the
%  MGK module from the list of supported formats.
%
%  The format of the UnregisterMGKImage method is:
%
%      UnregisterMGKImage(void)
%
*/
ModuleExport void UnregisterMGKImage(void)
{
  (void) UnregisterMagickInfo("MGK");
}

/*
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                                                                             %
%                                                                             %
%                                                                             %
%   W r i t e M G K I m a g e                                                 %
%                                                                             %
%                                                                             %
%                                                                             %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%  WriteMGKImage() writes an image to a file in red, green, and blue MGK
%  rasterfile format.
%
%  The format of the WriteMGKImage method is:
%
%      MagickBooleanType WriteMGKImage(const ImageInfo *image_info,
%        Image *image)
%
%  A description of each parameter follows.
%
%    o image_info: the image info.
%
%    o image:  The image.
%
%    o exception:  return any errors or warnings in this structure.
%
*/
static MagickBooleanType WriteMGKImage(const ImageInfo *image_info,Image *image,
  ExceptionInfo *exception)
{
  char
    buffer[MaxTextExtent];

  long
    y;

  MagickBooleanType
    status;

  MagickOffsetType
    scene;

  register const Quantum
    *p;

  register long
    x;

  register unsigned char
    *q;

  unsigned char
    *pixels;

  /*
    Open output image file.
  */
  assert(image_info != (const ImageInfo *) NULL);
  assert(image_info->signature == MagickCoreSignature);
  assert(image != (Image *) NULL);
  assert(image->signature == MagickCoreSignature);
  if (image->debug != MagickFalse)
    (void) LogMagickEvent(TraceEvent,GetMagickModule(),"%s",image->filename);
  status=OpenBlob(image_info,image,WriteBinaryBlobMode,exception);
  if (status == MagickFalse)
    return(status);
  scene=0;
  do
  {
    /*
      Allocate memory for pixels.
    */
    if (image->colorspace != RGBColorspace)
      (void) SetImageColorspace(image,RGBColorspace,exception);
    pixels=(unsigned char *) AcquireQuantumMemory((size_t) image->columns,
      3UL*sizeof(*pixels));
    if (pixels == (unsigned char *) NULL)
      ThrowWriterException(ResourceLimitError,"MemoryAllocationFailed");
    /*
      Initialize raster file header.
    */
    (void) WriteBlobString(image,"id=mgk\n");
    (void) FormatLocaleString(buffer,MaxTextExtent,"%lu %lu\n",image->columns,
       image->rows);
    (void) WriteBlobString(image,buffer);
    for (y=0; y < (long) image->rows; y++)
    {
      p=GetVirtualPixels(image,0,y,image->columns,1,exception);
      if (p == (const Quantum *) NULL)
        break;
      q=pixels;
      for (x=0; x < (long) image->columns; x++)
      {
        *q++=ScaleQuantumToChar(GetPixelRed(image,p));
        *q++=ScaleQuantumToChar(GetPixelGreen(image,p));
        *q++=ScaleQuantumToChar(GetPixelBlue(image,p));
        p+=GetPixelChannels(image);
      }
      (void) WriteBlob(image,(size_t) (q-pixels),pixels);
      if (image->previous == (Image *) NULL)
        if ((image->progress_monitor != (MagickProgressMonitor) NULL) &&
            (QuantumTick(y,image->rows) != MagickFalse))
          {
            status=image->progress_monitor(SaveImageTag,y,image->rows,
              image->client_data);
            if (status == MagickFalse)
              break;
          }
    }
    pixels=(unsigned char *) RelinquishMagickMemory(pixels);
    if (GetNextImageInList(image) == (Image *) NULL)
      break;
    image=SyncNextImageInList(image);
    status=SetImageProgress(image,SaveImagesTag,scene,
      GetImageListLength(image));
    if (status == MagickFalse)
      break;
    scene++;
  } while (image_info->adjoin != MagickFalse);
  (void) CloseBlob(image);
  return(MagickTrue);
}

To invoke the custom coder from the command line, use these commands:

magick logo: logo.mgk
display logo.mgk

We provide the Magick Coder Kit to help you get started writing your own custom coder.

Before you build, set the PKG_CONFIG_PATH environment variable if ImageMagick is not in your default system path:

export PKG_CONFIG_PATH=/usr/local/lib/pkgconfig 

Custom Image Filters

ImageMagick provides a convenient mechanism for adding your own custom image processing algorithms. We call these image filters and they are invoked from the command line with the -process option or from the MagickCore API method ExecuteModuleProcess().

Here is a listing of a sample custom image filter. It computes a few statistics such as the pixel brightness and saturation mean and standard-deviation.

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <assert.h>
#include <math.h>
#include "MagickCore/studio.h"
#include "MagickCore/MagickCore.h"

/*
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                                                                             %
%                                                                             %
%                                                                             %
%   a n a l y z e I m a g e                                                   %
%                                                                             %
%                                                                             %
%                                                                             %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%  analyzeImage() computes the brightness and saturation mean,  standard
%  deviation, kurtosis and skewness and stores these values as attributes 
%  of the image.
%
%  The format of the analyzeImage method is:
%
%      size_t analyzeImage(Image *images,const int argc,char **argv,
%        ExceptionInfo *exception)
%
%  A description of each parameter follows:
%
%    o image: the address of a structure of type Image.
%
%    o argc: Specifies a pointer to an integer describing the number of
%      elements in the argument vector.
%
%    o argv: Specifies a pointer to a text array containing the command line
%      arguments.
%
%    o exception: return any errors or warnings in this structure.
%
*/

typedef struct _StatisticsInfo
{
  double
    area,
    brightness,
    mean,
    standard_deviation,
    sum[5],
    kurtosis,
    skewness;
} StatisticsInfo;

static inline int GetMagickNumberThreads(const Image *source,
  const Image *destination,const size_t chunk,int multithreaded)
{
#define MagickMax(x,y)  (((x) > (y)) ? (x) : (y))
#define MagickMin(x,y)  (((x) < (y)) ? (x) : (y))

  /*
    Number of threads bounded by the amount of work and any thread resource
    limit.  The limit is 2 if the pixel cache type is not memory or
    memory-mapped.
  */
  if (multithreaded == 0)
    return(1);
  if (((GetImagePixelCacheType(source) != MemoryCache) &&
       (GetImagePixelCacheType(source) != MapCache)) ||
      ((GetImagePixelCacheType(destination) != MemoryCache) &&
       (GetImagePixelCacheType(destination) != MapCache)))
    return(MagickMax(MagickMin(GetMagickResourceLimit(ThreadResource),2),1));
  return(MagickMax(MagickMin((ssize_t) GetMagickResourceLimit(ThreadResource),
    (ssize_t) (chunk)/64),1));
}

ModuleExport size_t analyzeImage(Image **images,const int argc,
  const char **argv,ExceptionInfo *exception)
{
#define AnalyzeImageFilterTag  "Filter/Analyze"
#define magick_number_threads(source,destination,chunk,multithreaded) \
  num_threads(GetMagickNumberThreads(source,destination,chunk,multithreaded))

  char
    text[MagickPathExtent];

  Image
    *image;

  MagickBooleanType
    status;

  MagickOffsetType
    progress;

  assert(images != (Image **) NULL);
  assert(*images != (Image *) NULL);
  assert((*images)->signature == MagickCoreSignature);
  (void) argc;
  (void) argv;
  status=MagickTrue;
  progress=0;
  for (image=(*images); image != (Image *) NULL; image=GetNextImageInList(image))
  {
    CacheView
      *image_view;

    double
      area;

    ssize_t
      y;

    StatisticsInfo
      brightness,
      saturation;

    if (status == MagickFalse)
      continue;
    (void) memset(&brightness,0,sizeof(brightness));
    (void) memset(&saturation,0,sizeof(saturation));
    status=MagickTrue;
    image_view=AcquireVirtualCacheView(image,exception);
#if defined(MAGICKCORE_OPENMP_SUPPORT)
  #pragma omp parallel for schedule(static) \
    shared(progress,status,brightness,saturation) \
    magick_number_threads(image,image,image->rows,1)
#endif
    for (y=0; y < (ssize_t) image->rows; y++)
    {
      const Quantum
        *p;

      ssize_t
        i,
        x;

      StatisticsInfo
        local_brightness,
        local_saturation;

      if (status == MagickFalse)
        continue;
      p=GetCacheViewVirtualPixels(image_view,0,y,image->columns,1,exception);
      if (p == (const Quantum *) NULL)
        {
          status=MagickFalse;
          continue;
        }
      (void) memset(&local_brightness,0,sizeof(local_brightness));
      (void) memset(&local_saturation,0,sizeof(local_saturation));
      for (x=0; x < (ssize_t) image->columns; x++)
      {
        double
          b,
          h,
          s;

        ConvertRGBToHSL(GetPixelRed(image,p),GetPixelGreen(image,p),
          GetPixelBlue(image,p),&h,&s,&b);
        b*=QuantumRange;
        for (i=1; i <= 4; i++)
          local_brightness.sum[i]+=pow(b,(double) i);
        s*=QuantumRange;
        for (i=1; i <= 4; i++)
          local_saturation.sum[i]+=pow(s,(double) i);
        p+=GetPixelChannels(image);
      }
#if defined(MAGICKCORE_OPENMP_SUPPORT)
      #pragma omp critical (analyzeImage)
#endif
      for (i=1; i <= 4; i++)
      {
        brightness.sum[i]+=local_brightness.sum[i];
        saturation.sum[i]+=local_saturation.sum[i];
      }
    }
    image_view=DestroyCacheView(image_view);
    area=(double) image->columns*image->rows;
    brightness.mean=brightness.sum[1]/area;
    (void) FormatLocaleString(text,MagickPathExtent,"%g",brightness.mean);
    (void) SetImageProperty(image,"filter:brightness:mean",text,exception);
    brightness.standard_deviation=sqrt(brightness.sum[2]/area-
      (brightness.sum[1]/area*brightness.sum[1]/area));
    (void) FormatLocaleString(text,MagickPathExtent,"%g",
      brightness.standard_deviation);
    (void) SetImageProperty(image,"filter:brightness:standard-deviation",text,
      exception);
    if (fabs(brightness.standard_deviation) >= MagickEpsilon)
      brightness.kurtosis=(brightness.sum[4]/area-4.0*brightness.mean*
        brightness.sum[3]/area+6.0*brightness.mean*brightness.mean*
        brightness.sum[2]/area-3.0*brightness.mean*brightness.mean*
        brightness.mean*brightness.mean)/(brightness.standard_deviation*
        brightness.standard_deviation*brightness.standard_deviation*
        brightness.standard_deviation)-3.0;
    (void) FormatLocaleString(text,MagickPathExtent,"%g",brightness.kurtosis);
    (void) SetImageProperty(image,"filter:brightness:kurtosis",text,exception);
    if (brightness.standard_deviation != 0)
      brightness.skewness=(brightness.sum[3]/area-3.0*brightness.mean*
        brightness.sum[2]/area+2.0*brightness.mean*brightness.mean*
        brightness.mean)/(brightness.standard_deviation*
        brightness.standard_deviation*brightness.standard_deviation);
    (void) FormatLocaleString(text,MagickPathExtent,"%g",brightness.skewness);
    (void) SetImageProperty(image,"filter:brightness:skewness",text,exception);
    saturation.mean=saturation.sum[1]/area;
    (void) FormatLocaleString(text,MagickPathExtent,"%g",saturation.mean);
    (void) SetImageProperty(image,"filter:saturation:mean",text,exception);
    saturation.standard_deviation=sqrt(saturation.sum[2]/area-
      (saturation.sum[1]/area*saturation.sum[1]/area));
    (void) FormatLocaleString(text,MagickPathExtent,"%g",
      saturation.standard_deviation);
    (void) SetImageProperty(image,"filter:saturation:standard-deviation",text,
      exception);
    if (fabs(saturation.standard_deviation) >= MagickEpsilon)
      saturation.kurtosis=(saturation.sum[4]/area-4.0*saturation.mean*
        saturation.sum[3]/area+6.0*saturation.mean*saturation.mean*
        saturation.sum[2]/area-3.0*saturation.mean*saturation.mean*
        saturation.mean*saturation.mean)/(saturation.standard_deviation*
        saturation.standard_deviation*saturation.standard_deviation*
        saturation.standard_deviation)-3.0;
    (void) FormatLocaleString(text,MagickPathExtent,"%g",saturation.kurtosis);
    (void) SetImageProperty(image,"filter:saturation:kurtosis",text,exception);
    if (fabs(saturation.standard_deviation) >= MagickEpsilon)
      saturation.skewness=(saturation.sum[3]/area-3.0*saturation.mean*
        saturation.sum[2]/area+2.0*saturation.mean*saturation.mean*
        saturation.mean)/(saturation.standard_deviation*
        saturation.standard_deviation*saturation.standard_deviation);
    (void) FormatLocaleString(text,MagickPathExtent,"%g",saturation.skewness);
    (void) SetImageProperty(image,"filter:saturation:skewness",text,exception);
    if (image->progress_monitor != (MagickProgressMonitor) NULL)
      {
        MagickBooleanType
          proceed;

#if defined(MAGICKCORE_OPENMP_SUPPORT)
        #pragma omp atomic
#endif
        progress++;
        proceed=SetImageProgress(image,AnalyzeImageFilterTag,progress,
          GetImageListLength(image));
        if (proceed == MagickFalse)
          status=MagickFalse;
      }
  }
  return(MagickImageFilterSignature);
}

To invoke the custom filter from the command line, use this command:

magick logo: -process \"analyze\" -verbose info:
Image: logo:
  Format: LOGO (ImageMagick Logo)
  Class: PseudoClass
  Geometry: 640x480
  ...
  filter:brightness:kurtosis: 3.97886
  filter:brightness:mean: 58901.3
  filter:brightness:skewness: -2.30827
  filter:brightness:standard-deviation: 16179.8
  filter:saturation:kurtosis: 6.59719
  filter:saturation:mean: 5321.05
  filter:saturation:skewness: 2.75679
  filter:saturation:standard-deviation: 14484.7

We provide the Magick Filter Kit to help you get started writing your own custom image filter.

Before you build, set the PKG_CONFIG_PATH environment variable if ImageMagick is not in your default system path:

export PKG_CONFIG_PATH=/usr/local/lib/pkgconfig