Provided by: libprima-perl_1.28-1.1_amd64 bug

NAME

       Prima::Image - Bitmap routines

SYNOPSIS

          use Prima qw(Application);

          # create a new image from scratch
          my $i = Prima::Image-> new(
             width => 32,
             height => 32,
             type   => im::BW, # same as im::bpp1 | im::GrayScale
          );

          # draw something
          $i-> begin_paint;
          $i-> color( cl::White);
          $i-> ellipse( 5, 5, 10, 10);
          $i-> end_paint;

          # mangle
          $i-> size( 64, 64);

          # file operations
          $i-> save('a.gif') or die "Error saving:$@\n";
          $i-> load('a.gif') or die "Error loading:$@\n";

          # draw on screen
          $::application-> begin_paint;

          # an image is drawn as specified by its palette
          $::application-> set( color => cl::Red, backColor => cl::Green);
          $::application-> put_image( 100, 100, $i);

          # a bitmap is drawn as specified by destination device colors
          $::application-> put_image( 200, 100, $i-> bitmap);

DESCRIPTION

       Prima::Image, Prima::Icon and Prima::DeviceBitmap are classes for bitmap handling,
       including file and graphic input and output. Prima::Image and Prima::DeviceBitmap are
       descendants of Prima::Drawable and represent bitmaps, stored in memory.  Prima::Icon is a
       descendant of Prima::Image and contains a transparency mask along with the regular data.

USAGE

       Images usually are represented as a memory area, where pixel data are stored row-wise. The
       Prima toolkit is no exception, however, it does not assume that the GUI system uses the
       same memory format.  The implicit conversion routines are called when Prima::Image is
       about to be drawn onto the screen, for example. The conversions are not always efficient,
       therefore the Prima::DeviceBitmap class is introduced to represent a bitmap, stored in the
       system memory in the system pixel format. These two basic classes serve the different
       needs, but can be easily converted to each other, with "image" and "bitmap" methods.
       Prima::Image is a more general bitmap representation, capable of file and graphic input
       and output, plus it is supplied with number of conversion and scaling functions. The
       Prima::DeviceBitmap class has almost none of additional functionality, and is targeted to
       efficient graphic input and output.

   Graphic input and output
       As descendants of Prima::Drawable, all Prima::Image, Prima::Icon and Prima::DeviceBitmap
       objects are subject to three-state painting mode - normal ( disabled ), painting ( enabled
       ) and informational.  Prima::DeviceBitmap is, however, exists only in the enabled state,
       and can not be switched to the other two.

       When an object enters the enabled state, it serves as a canvas, and all Prima::Drawable
       operations can be performed on it. When the object is back to the disabled state, the
       graphic information is stored into the object associated memory, in the pixel format,
       supported by the toolkit.  This information can be visualized by using one of
       "Prima::Drawable::put_image" group methods. If the object enters the enabled state again,
       the graphic information is presented as an initial state of a bitmap.

       It must be noted, that if an implicit conversion takes place after an object enters and
       before it leaves the enabled state, as it is with Prima::Image and Prima::Icon, the bitmap
       is converted to the system pixel format. During such conversion some information can be
       lost, due to down-sampling, and there is no way to preserve the information. This does not
       happen with Prima::DeviceBitmap.

       Image objects can be drawn upon images, as well as on the screen and Prima::Widget
       objects. This operation is performed via one of Prima::Drawable::put_image group methods (
       see Prima::Drawable), and can be called with the image object disregarding the paint
       state. The following code illustrates the dualism of an image object, where it can serve
       both as a drawing surface and as a drawing tool:

           my $a = Prima::Image-> create( width => 100, height => 100, type => im::RGB);
           $a-> begin_paint;
           $a-> clear;
           $a-> color( cl::Green);
           $a-> fill_ellipse( 50, 50, 30, 30);
           $a-> end_paint;
           $a-> rop( rop::XorPut);
           $a-> put_image( 10, 10, $a);
           $::application-> begin_paint;
           $::application-> put_image( 0, 0, $a);
           $::application-> end_paint;

       It must be noted, that "put_image", "stretch_image" and "put_image_indirect" are only
       painting methods that allow drawing on an image that is in its paint-disabled state.
       Moreover, in such context they only allow "Prima::Image" descendants to be passed as a
       source image object. This functionality does not imply that the image is internally
       switched to the paint-enabled state and back; the painting is performed without switching
       and without interference with the system's graphical layer.

       Another special case is a 1-bit ( monochrome ) DeviceBitmap. When it is drawn upon a
       drawable with bit depth greater than 1, the drawable's color and backColor properties are
       used to reflect 1 and 0 bits, respectively. On a 1-bit drawable this does not happen, and
       the color properties are not used.

   File input and output
       Depending on the toolkit configuration, images can be read and written in different
       formats. This functionality in accessible via "load()" and "save()" methods.
       Prima::image-load is dedicated to the description of loading and saving parameters, that
       can be passed to the methods, so they can handle different aspects of file format-specific
       options, such as multi-frame operations, auto conversion when a format does not support a
       particular pixel format etc. In this document, "load()" and "save()" methods are
       illustrated only in their basic, single-frame functionality. When called with no extra
       parameters, these methods fail only if a disk I/O error occurred or an unknown image
       format was used.

       When an image is loaded, the old bitmap memory content is discarded, and the image
       attributes are changed accordingly to the loaded image.  Along with these, an image
       palette is loaded, if available, and a pixel format is assigned, closest or identical to
       the pixel format in the image file.

   Pixel formats
       Prima::Image supports a number of pixel formats, governed by the "::type" property. It is
       reflected by an integer value, a combination of "im::XXX" constants. The whole set of
       pixel formats is represented by colored formats, like, 16-color, 256-color and 16M-color,
       and by gray-scale formats, mapped to C data types - unsigned char, unsigned short,
       unsigned long, float and double.  The gray-scale formats are further subdivided to real-
       number formats and complex-number format; the last ones are represented by two real values
       per pixel, containing the real and the imaginary values.

       Prima::Image can also be initialized from other formats, that it does not support, but can
       convert data from. Currently these are represented by a set of permutations of 32-bit RGBA
       format, and 24-bit BGR format.  These formats can only be used in conjunction with
       "::data" property.

       The conversions can be performed between any of the supported formats ( to do so, "::type"
       property is to be set-called ). An image of any of these formats can be drawn on the
       screen, but if the system can not accept the pixel format ( as it is with non-integer or
       complex formats ), the bitmap data are implicitly converted. The conversion does not
       change the data if the image is about to be drawn; the conversion is performed only when
       the image is about to be served as a drawing surface. If, by any reason, it is desired
       that the pixel format is not to be changed, the "::preserveType" property must be set to
       1. It does not prevent the conversion, but it detects if the image was implicitly
       converted inside "end_paint()" call, and reverts it to its previous pixel format.

       There are situations, when pixel format must be changed together while down-sampling the
       image. One of four down-sampling methods can be selected - normal, 8x8 ordered halftoning,
       error diffusion, and error diffusion combined with optimized palette. These can be set to
       the "::conversion" property with one of "ict::XXX" constants.  When there is no
       information loss, "::conversion" property is not used.

       Another special case of conversion is a conversion with a palette. The following calls,

         $image-> type( im::bpp4);
         $image-> palette( $palette);

       and

         $image-> palette( $palette);
         $image-> type( im::bpp4);

       produce different results, but none of these takes into account eventual palette
       remapping, because "::palette" property does not change bitmap pixel data, but overwrites
       palette information. A proper call syntax here would be

         $image-> set(
            palette => $palette,
            type    => im::bpp4,
         );

       This call produces also palette pixel mapping.  This syntax is most powerful when
       conversion is set to "ict::Optimized" ( by default ). It not only allows remapping or
       downsampling to a predefined colors set, but also can be used to limit palette size to a
       particular number, without knowing the actual values of the final color palette. For
       example, for an 24-bit image,

         $image-> set( type => im::bpp8, palette => 32);

       call would calculate colors in the image, compress them to an optimized palette of 32
       cells and finally converts to a 8-bit format.

       Instead of "palette" property, "colormap" can also be used.

   Data access
       The pixel values can be accessed in Prima::Drawable style, via "::pixel" property.
       However, Prima::Image introduces several helper functions, for different aims. The
       "::data" property is used to set or retrieve a scalar representation of bitmap data. The
       data are expected to be lined up to a 'line size' margin ( 4-byte boundary ), which is
       calculated as

         $lineSize = int(( $image->width * ( $image-> type & im::BPP) + 31) / 32) * 4;

       or returned from the read-only property "::lineSize".

       This is the line size for the data as lined up internally in memory, however "::data"
       should not necessarily should be aligned like this, and can be accompanied with a write-
       only flag 'lineSize' if pixels are aligned differently:

         $image-> set( width => 1, height=> 2);
         $image-> type( im::RGB);
         $image-> set(
            data => 'RGB----RGB----',
            lineSize => 7,
         );
         print $image-> data, "\n";


         output: RGB-RGB-
       Internally, Prima contains images in memory so that the first scanline is the farthest
       away from the memory start; this is consistent with general Y-axis orientation in Prima
       drawable terminology, but might be inconvenient when importing data organized otherwise.
       Another write-only boolean flag "reverse" can be set to 1 so data then are treated as if
       the first scanline of the image is the closest to the start of data:

         $image-> set( width => 1, height=> 2, type => im::RGB);
         $image-> set(
            data => 'RGB-123-',
            reverse => 1,
         );
         print $image-> data, "\n";


         output: RGB-123-
       Although it is possible to perform all kinds of calculations and modification with the
       pixels, returned by "::data", it is not advisable unless the speed does not matter.
       Standalone PDL package with help of PDL::PrimaImage package, and Prima-derived IPA package
       provide routines for data and image analysis.  Also, Prima::Image::Magick connects
       ImageMagick with Prima.  Prima::Image itself provides only the simplest statistic
       information, namely: lowest and highest pixel values, pixel sum, sum of square pixels,
       mean, variance, and standard deviation.

   Standalone usage
       The image functionality can be used standalone, with all other parts of the toolkit being
       uninitialized. This is useful in non-interactive programs, running in evnironments with no
       GUI access, a cgi-script with no access to X11 display, for example. Normally, Prima fails
       to start in such situations, but can be told not to initialize its GUI part by explicitly
       operating system-dependent options. To do so, invoke

         use Prima::noX11;

       in the beginning of your program. See Prima::noX11 for more.

   Prima::Icon
       Prima::Icon inherits all properties of Prima::Image, and it also provides a 1-bit depth
       transparency mask.  This mask can also be loaded and saved into image files, if the format
       supports a transparency information.

       Similar to Prima::Image::data property, Prima::Icon::mask property provides access to the
       binary mask data.  The mask can be updated automatically, after an icon object was subject
       to painting, resizing, or other destructive change.  The auxiliary properties
       "::autoMasking" and "::maskColor"/"::maskIndex" regulate  mask update procedure. For
       example, if an icon was loaded with the color ( vs. bitmap ) transparency information, the
       binary mask will be generated anyway, but it will be also recorded that a particular color
       serves as a transparent indicator, so eventual conversions can rely on the color value,
       instead of the mask bitmap.

       If an icon is drawn upon a graphic canvas, the image output is constrained to the mask. On
       raster displays it is typically simulated by a combination of and- and xor- operation
       modes, therefore attempts to put an icon with "::rop", different from "rop::CopyPut",
       usually fail.

API

   Prima::Image properties
       colormap @PALETTE
           A color palette, used for representing 1, 4, and 8-bit bitmaps, when an image object
           is to be visualized. @PALETTE contains individual colors component triplets, in RGB
           format. For example, black-and-white monochrome image may contain colormap as
           "0,0xffffff".

           See also "palette".

       conversion TYPE
           Selects the type of dithering algorithm to be used for pixel down-sampling.  TYPE is
           one of "ict::XXX" constants:

              ict::None            - no dithering
              ict::Halftone        - 8x8 ordered halftone dithering
              ict::ErrorDiffusion  - error diffusion dithering with static palette
              ict::Optimized       - error diffusion dithering with optimized palette

           As an example, if a 4x4 color image with every pixel set to RGB(32,32,32), converted
           to a 1-bit image, the following results occur:

              ict::None:
                [ 0 0 0 0 ]
                [ 0 0 0 0 ]
                [ 0 0 0 0 ]
                [ 0 0 0 0 ]

              ict::Halftone:
                [ 0 0 0 0 ]
                [ 0 0 1 0 ]
                [ 0 0 0 0 ]
                [ 1 0 0 0 ]

              ict::ErrorDiffusion, ict::Ordered:
                [ 0 0 1 0 ]
                [ 0 0 0 1 ]
                [ 0 0 0 0 ]
                [ 0 0 0 0 ]

       data SCALAR
           Provides access to the bitmap data. On get-call, returns all bitmap pixels, aligned to
           4-byte boundary. On set-call, stores the provided data with same alignment. The
           alignment can be altered by submitting 'lineSize' write-only flag to set call; the
           ordering of scan lines can be altered by setting 'reverse' write-only flag ( see "Data
           access" ).

       height INTEGER
           Manages the vertical dimension of the image data.  On set-call, the image data are
           changed accordingly to the new height, and depending on "::vScaling" property, the
           pixel values are either scaled or truncated.

       hScaling BOOLEAN
           If 1, the bitmap data will be scaled when image changes its horizontal extent. If 0,
           the data will be stripped or padded with zeros.

       lineSize INTEGER
           A read-only property, returning the length of an image row in bytes, as represented
           internally in memory. Data returned by "::data" property are aligned with "::lineSize"
           bytes per row, and setting "::data" expects data aligned with this value, unless
           "lineSize" is set together with "data" to indicate another alignment. See "Data
           access" for more.

       mean
           Returns mean value of pixels.  Mean value is "::sum" of pixel values, divided by
           number of pixels.

       palette [ @PALETTE ]
           A color palette, used for representing 1, 4, and 8-bit bitmaps, when an image object
           is to be visualized. @PALETTE contains individual color component triplets, in BGR
           format. For example, black-and-white monochrome image may contain palette as
           "[0,0,0,255,255,255]".

           See also "colormap".

       pixel ( X_OFFSET, Y_OFFSET ) PIXEL
           Provides per-pixel access to the image data when image object is in disabled paint
           state. Otherwise, same as "Prima::Drawable::pixel".

       preserveType BOOLEAN
           If 1, reverts the image type to its old value if an implicit conversion was called
           during "end_paint()".

       rangeHi
           Returns maximum pixel value in the image data.

       rangeLo
           Returns minimum pixel value in the image data.

       size WIDTH, HEIGHT
           Manages dimensions of the image. On set-call, the image data are changed accordingly
           to the new dimensions, and depending on "::vScaling" and "::hScaling" properties, the
           pixel values are either scaled or truncated.

       stats ( INDEX ) VALUE
           Returns one of calculated values, that correspond to INDEX, which is one of the
           following "is::XXX" constants:

              is::RangeLo  - minimum pixel value
              is::RangeHi  - maximum pixel value
              is::Mean     - mean value
              is::Variance - variance
              is::StdDev   - standard deviation
              is::Sum      - sum of pixel values
              is::Sum2     - sum of squares of pixel values

           The values are re-calculated on request and cached.  On set-call VALUE is stored in
           the cache, and is returned on next get-call.  The cached values are discarded every
           time the image data changes.

           These values are also accessible via set of alias properties: "::rangeLo",
           "::rangeHi", "::mean", "::variance", "::stdDev", "::sum", "::sum2".

       stdDev
           Returns standard deviation of the image data.  Standard deviation is the square root
           of "::variance".

       sum Returns sum of pixel values of the image data

       sum2
           Returns sum of squares of pixel values of the image data

       type TYPE
           Governs the image pixel format type. TYPE is a combination of "im::XXX" constants. The
           constants are collected in groups:

           Bit-depth constants provide size of pixel is bits. Their actual value is same as
           number of bits, so "im::bpp1" value is 1, "im::bpp4" - 4, etc. The valid constants
           represent bit depths from 1 to 128:

              im::bpp1
              im::bpp4
              im::bpp8
              im::bpp16
              im::bpp24
              im::bpp32
              im::bpp64
              im::bpp128

           The following values designate the pixel format category:

              im::Color
              im::GrayScale
              im::RealNumber
              im::ComplexNumber
              im::TrigComplexNumber

           Value of "im::Color" is 0, whereas other category constants represented by unique bit
           value, so combination of "im::RealNumber" and "im::ComplexNumber" is possible.

           There also several mnemonic constants defined:

              im::Mono          - im::bpp1
              im::BW            - im::bpp1 | im::GrayScale
              im::16            - im::bpp4
              im::Nibble        - im::bpp4
              im::256           - im::bpp8
              im::RGB           - im::bpp24
              im::Triple        - im::bpp24
              im::Byte          - gray 8-bit unsigned integer
              im::Short         - gray 16-bit unsigned integer
              im::Long          - gray 32-bit unsigned integer
              im::Float         - float
              im::Double        - double
              im::Complex       - dual float
              im::DComplex      - dual double
              im::TrigComplex   - dual float
              im::TrigDComplex  - dual double

           Bit depths of float- and double- derived pixel formats depend on a platform.

           The groups can be masked out with the mask values:

              im::BPP      - bit depth constants
              im::Category - category constants
              im::FMT      - extra format constants

           The extra formats are the pixel formats, not supported by "::type", but recognized
           within the combined set-call, like

              $image-> set(
                 type => im::fmtBGRI,
                 data => 'BGR-BGR-',
              );

           The data, supplied with the extra image format specification will be converted to the
           closest supported format. Currently, the following extra pixel formats are recognized:

              im::fmtBGR
              im::fmtRGBI
              im::fmtIRGB
              im::fmtBGRI
              im::fmtIBGR

       variance
           Returns variance of pixel values of the image data.  Variance is "::sum2", divided by
           number of pixels minus square of "::sum" of pixel values.

       vScaling BOOLEAN
           If 1, the bitmap data will be scaled when image changes its vertical extent. If 0, the
           data will be stripped or padded with zeros.

       width INTEGER
           Manages the horizontal dimension of the image data.  On set-call, the image data are
           changed accordingly to the new width, and depending on "::hScaling" property, the
           pixel values are either scaled or truncated.

   Prima::Icon properties
       autoMasking TYPE
           Selects whether the mask information should be updated automatically with "::data"
           change or not. Every "::data" change is mirrored in "::mask", using TYPE, one of
           "am::XXX" constants:

              am::None           - no mask update performed
              am::MaskColor      - mask update based on ::maskColor property
              am::MaskIndex      - mask update based on ::maskIndex property
              am::Auto           - mask update based on corner pixel values

           The "::maskColor" color value is used as a transparent color if TYPE is
           "am::MaskColor". The transparency mask generation algorithm, turned on by "am::Auto"
           checks corner pixel values, assuming that majority of the corner pixels represents a
           transparent color. Once such color is found, the mask is generated as in
           "am::MaskColor" case.

           "::maskIndex" is the same as "::maskColor", except that it points to a specific color
           index in the palette.

           When image "::data" is stretched, "::mask" is stretched accordingly, disregarding the
           "::autoMasking" value.

       mask SCALAR
           Provides access to the transparency bitmap. On get-call, returns all bitmap pixels,
           aligned to 4-byte boundary in 1-bit format. On set-call, stores the provided
           transparency data with same alignment.

       maskColor COLOR
           When "::autoMasking" set to "am::MaskColor", COLOR is used as a transparency value.

       maskIndex INDEX
           When "::autoMasking" set to "am::MaskIndex", INDEXth color in teh current palette is
           used as a transparency value.

   Prima::DeviceBitmap properties
       monochrome BOOLEAN
           A read-only property, that can only be set during creation, reflects whether the
           system bitmap is black-and-white 1-bit (monochrome) or not.  The color depth of a
           bitmap can be read via "get_bpp()" method; monochrome bitmaps always have bit depth of
           1.

   Prima::Image methods
       bitmap
           Returns newly created Prima::DeviceBitmap instance, with the image dimensions and with
           the bitmap pixel values copied to.

       codecs
           Returns array of hashes, each describing the supported image format. If the array is
           empty, the toolkit was set up so it can not load and save images.

           See Prima::image-load for details.

           This method can be called without object instance.

       dup Returns a duplicate of the object, a newly created Prima::Image, with all information
           copied to it.

       extract X_OFFSET, Y_OFFSET, WIDTH, HEIGHT
           Returns a newly created image object with WIDTH and HEIGHT dimensions, initialized
           with pixel data from X_OFFSET and Y_OFFSET in the bitmap.

       get_bpp
           Returns the bit depth of the pixel format. Same as "::type & im::BPP".

       get_handle
           Returns a system handle for an image object.

       load (FILENAME or FILEGLOB) [ %PARAMETERS ]
           Loads image from file FILENAME or stream FILEGLOB into an object, and returns the
           success flag.  The semantics of "load()" is extensive, and can be influenced by
           PARAMETERS hash. "load()" can be called either in a context of an existing object,
           then a boolean success flag is returned, or in a class context, then a newly created
           object ( or "undef" ) is returned. If an error occurs, $@ variable contains the error
           description string. These two invocation semantics are equivalent:

              my $x = Prima::Image-> create();
              die "$@" unless $x-> load( ... );

           and

              my $x = Prima::Image-> load( ... );
              die "$@" unless $x;

           See Prima::image-load for details.

           NB! When loading from streams on win32, mind "binmode".

       map COLOR
           Performs iterative mapping of bitmap pixels, setting every pixel to "::color" property
           with respect to "::rop" type if a pixel equals to COLOR, and to "::backColor" property
           with respect to "::rop2" type otherwise.

           "rop::NoOper" type can be used for color masking.

           Examples:

              width => 4, height => 1, data => [ 1, 2, 3, 4]
              color => 10, backColor => 20, rop => rop::CopyPut

              rop2 => rop::CopyPut
              input: map(2) output: [ 20, 10, 20, 20 ]

              rop2 => rop::NoOper
              input: map(2) output: [ 1, 10, 3, 4 ]

       resample SRC_LOW, SRC_HIGH, DEST_LOW, DEST_HIGH
           Performs linear scaling of gray pixel values from range (SRC_LOW - SRC_HIGH) to range
           (DEST_LOW - DEST_HIGH). Can be used to visualize gray non-8 bit pixel values, by the
           code:

              $image-> resample( $image-> rangeLo, $image-> rangeHi, 0, 255);

       save (FILENAME or FILEGLOB), [ %PARAMETERS ]
           Stores image data into image file FILENAME or stream FILEGLOB, and returns the success
           flag.  The semantics of "save()" is extensive, and can be influenced by PARAMETERS
           hash. If error occurs, $@ variable contains error description string.

           Note that when saving to a stream, "codecID" must be explicitly given in %PARAMETERS.

           See Prima::image-load for details.

           NB! When saving to streams on win32, mind "binmode".

   Prima::Image events
       "Prima::Image"-specific events occur only from inside load call, to report image loading
       progress. Not all codecs (currently JPEG,PNG,TIFF only) are able to report the progress to
       the caller. See "Loading with progress indicator" in Prima::image-load for details,
       "watch_load_progress" in Prima::ImageViewer and "load" in Prima::ImageDialog for suggested
       use.

       HeaderReady
           Called whenever image header is read, and image dimensions and pixel type is changed
           accordingly to accomodate image data.

       DataReady X, Y, WIDTH, HEIGHT
           Called whenever image data that cover area designated by X,Y,WIDTH,HEIGHT is acquired.
           Use "load" option "eventDelay" to limit the rate of "DataReady" event.

   Prima::Icon methods
       split
           Returns two new Prima::Image objects of same dimension.  Pixels in the first is are
           duplicated from "::data" storage, in the second - from "::mask" storage.

       combine DATA, MASK
           Copies information from DATA and MASK images into "::data" and "::mask" property. DATA
           and MASK are expected to be images of same dimension.

   Prima::DeviceBitmap methods
       icon
           Returns a newly created Prima::Icon object instance, with the pixel information copied
           from the object.

       image
           Returns a newly created Prima::Image object instance, with the pixel information
           copied from the object.

       get_handle
           Returns a system handle for a system bitmap object.

AUTHOR

       Dmitry Karasik, <dmitry@karasik.eu.org>.

SEE ALSO

       Prima, Prima::Drawable, Prima::image-load, Prima::codecs.

       PDL, PDL::PrimaImage, IPA

       ImageMagick, Prima::Image::Magick