Ubuntu Manpage: Prima::Image

Provided by: libprima-perl_1.28-1.4_amd64

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 "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>.

NAME

SYNOPSIS

DESCRIPTION

USAGE

API

AUTHOR

SEE ALSO