GSP
Quick Navigator

Search Site

Unix VPS
A - Starter
B - Basic
C - Preferred
D - Commercial
MPS - Dedicated
Previous VPSs
* Sign Up! *

Support
Contact Us
Online Help
Handbooks
Domain Status
Man Pages

FAQ
Virtual Servers
Pricing
Billing
Technical

Network
Facilities
Connectivity
Topology Map

Miscellaneous
Server Agreement
Year 2038
Credits
 

USA Flag

 

 

Man Pages
pod::Prima::image-load(3) User Contributed Perl Documentation pod::Prima::image-load(3)

Prima::image-load - Using image subsystem

Details on image subsystem - image loading, saving, and codec managements

Simplest case, loading a single image would look like:

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

Image functions can work being either invoked from package, or from existing Prima::Image object, in latter case the caller object itself is changing. The code above could be also written as

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

In both cases $x contains image data upon success. Error is returned into $@ variable ( see perldoc perlvar for more info).

"Prima::Image" can also load image by reading from a stream:

        open FILE, 'a.jpeg' or die "Cannot open:$!";
        binmode FILE;
        my $x = Prima::Image-> load( \*FILE);
        die "$@" unless $x;

Multiframe load call can be also issued in two ways:

        my @x = Prima::Image-> load( 'filename.duf', loadAll => 1);
        die "$@" unless $x[-1];
        my $x = Prima::Image-> create;
        my @x = $x-> load( 'filename.duf', loadAll => 1);
        die "$@" unless $x[-1];

In second case, the content of the first frame comes to $x and $x[0]. Sufficient check for error is whether last item of a returned array is defined. This check works also if an empty array is returned. Only this last item can be an undefined value, others are guaranteed to be valid objects.

Multiframe syntax is expressed in a set of extra hash keys. These keys are:

Request for loading all frames that can be read from a file. Example:

        loadAll => 1
    
If present, returns a single frame with index given. Example:

        index => 8
    
Contains an anonymous array of frame indices to load. Valid indices are above zero, negative ones can't be counted in a way perl array indices are. Example:

         map => [0, 10, 15..20]
    

By default Prima loads image data and palette only. For any other information that can be loaded, anonymous hash 'extras' can be defined. To notify a codec that this extra information is desired, loadExtras boolean value is used. Example:

        my $x = Prima::Image-> load( $f, loadExtras => 1);
        die "$@" unless $x;
        for ( keys %{$x-> {extras}}) {
           print " $_ : $x->{extras}->{$_}\n";
        }

The code above loads and prints extra information read from a file. Typical output, for example, from a gif codec based on libgif would look like:

    codecID : 1
    transparentColorIndex : 1
    comment : created by GIMP
    frames : 18

'codecID' is a Prima-defined extra field, which is an index of the codec which have loaded the file. This field's value is useful for explicit indication of codec on the save request.

'frames' is also a Prima-defined extra field, with integer value set to a number of frames in the image. It might be set to -1, signaling that codec is incapable of quick reading of the frame count. If, however, it is necessary to get actual frame count, a 'wantFrames' profile boolean value should be set to 1 - then frames is guaranteed to be set to a 0 or positive value, but the request may take longer time, especially on a large file with sequential access. Real life example is a gif file with more than thousand frames. 'wantFrames' is useful in null load requests.

The parameters that are accepted by load, are divided into several categories - first, those that apply to all loading process and those who apply only to a particular frame. Those who are defined by Prima, are enumerated above - loadExtras, loadAll etc. Only loadExtras, noImageData, noIncomplete and iconUnmask are applicable to a frame, other govern the loading process. A codec may as well define its own parameters, however it is not possible to tell what parameter belongs to what group - this information is to be found in codec documentation;

The parameters that applicable to any frame, can be specified separately to every desirable frame in single call. For that purpose, parameter 'profiles' is defined. 'profiles' is expected to be an anonymous array of hashes, each hash where corresponds to a request number. Example:

        $x-> load( $f, loadAll => 1, profiles => [
             {loadExtras => 0},
             {loadExtras => 1},
        ]);

First hash there applies to frame index 0, second - to frame index 1. Note that in code

        $x-> load( $f,
           map => [ 5, 10],
           profiles => [
             {loadExtras => 0},
             {loadExtras => 1},
        ]);

first hash applies to frame index 5, and second - to frame index 10.

If it is desired to peek into image, reading type and dimensions only, one should set 'noImageData' boolean value to 1. Using 'noImageData', empty objects with read type are returned, and with extras 'width' and 'height' set to image dimensions. Example:

        $x-> load( $f, noImageData => 1);
        die "$@" unless $x;
        print $x-> {extras}-> {width} , 'x' , $x-> {extras}-> {height}, 'x',
           $x-> type & im::BPP, "\n";

Some information about image can be loaded even without frame loading - if the codec provides such a functionality. This is the only request that cannot be issued on a package:

        $x-> load( $f, map => [], loadExtras => 1);

Since no frames are required to load, an empty array is returned upon success and an array with one undefined value on failure.

If Prima needs to create a storage object, it is by default Prima::Image, or a class name of an caller object, or a package the request was issued on. This behavior can be altered using parameter 'className', which defines the class to be used for the frame.

        my @x = Prima::Image-> load( $f,
            map => [ 1..3],
            className => 'Prima::Icon',
            profiles => [
                {},
                { className => 'Prima::Image' },
                {}
            ],

In this example @x will be ( Icon, Image, Icon) upon success.

When loading to an Icon object, the default toolkit action is to build the transparency mask based on image data. When it is not the desired behavior, e.g., there is no explicit knowledge of image, but the image may or may not contain transparency information, "iconUnmask" boolean option can be used. When set to a "true" value, and the object is "Prima::Icon" descendant, "Prima::Icon::autoMasking" is set to "am::None" prior to the file loading. By default this options is turned off.

Some codecs (PNG,TIFF,JPEG) can notify the caller as they read image data. For this purpose, "Prima::Image" has two events, "onHeaderReady" and "onDataReady". If either (or both) are present on image object that is issuing load call, and the codec supports progressive loading, these events are called. "onHeaderReady" is called when image header data is acquired, and empty image with the dimensions and pixel type is allocated. "onDataReady" is called whenever a part of image is ready and is loaded in the memory of the object; the position and dimensions of the loaded area is reported also. The format of the events is:

    onHeaderReady $OBJECT
    onDataReady   $OBJECT, $X, $Y, $WIDTH, $HEIGHT

"onHeaderReady" is called only once, but "onDataReady" is called as soon as new image data is available. To reduce frequency of these calls, that otherwise would be issued on every scanline loaded, "load" has parameter "eventDelay", a number of seconds, which limits event rate. The default "eventDelay" is 0.1 .

The handling on "onDataReady" must be performed with care. First, the image must be accessed read-only, which means no transformations with image size and type are allowed. Currently there is no protection for such actions ( because codec must perform these ), so a crash will most surely issue. Second, loading and saving of images is not in general reentrant, and although some codecs are reentrant, loading and saving images inside image events is not recommended.

There are two techniques to display partial image as it loads. All of these share overloading of "onHeaderReady" and "onDataReady". The simpler is to call "put_image" from inside "onDataReady":

        $i = Prima::Image-> new(
                onDataReady => sub {
                        $progress_widget-> put_image( 0, 0, $i);
                },
        );

but that will most probably loads heavily underlying OS-dependent conversion of image data to native display bitmap data. A more smarter, but more complex solution is to copy loaded (and only loaded) bits to a preexisting device bitmap:

        $i = Prima::Image-> new(
                onHeaderReady => sub {
                        $bitmap = Prima::DeviceBitmap-> new(
                                width    => $i-> width,
                                height   => $i-> height,
                        ));
                },
                onDataReady => sub {
                        my ( $i, $x, $y, $w, $h) = @_;
                        $bitmap-> put_image( $x, $y, $i-> extract( $x, $y, $w, $h));
                },
        );

The latter technique is used by "Prima::ImageViewer" when it is setup to monitor image loading progress. See "watch_load_progress" in Prima::ImageViewer for details.

By default, codecs are not specified whether they would fail on premature end of file or omit the error and return truncated image. "noIncomplete" boolean flag tells that a codec must always fail if the image cannot be red in full. It is off by default. If indeed the codec detected that the file was incomplete, it sets "truncated" error string in the "extras" profile, if "loadExtras" was requested.

Using "Prima::Image::base64" it is possible to convert images into an base64 format, and embed the result directly into the source. Assuming an appropriate codec was compiled, the following would work:

        my $icon = Prima::Icon->load_stream(<<~'ICON');
                R0lGODdhIAAgAIAAAAAAAP///ywAAAAAIAAgAIAAAAD///8CT4SPqcvtD6OctNqLcwogcK91nEhq
                3gim2Umm4+W2IBzX0fvl8jTr9SeZiU5E4a1XLHZ4yaal6XwFoSwMVUVzhoZSaQW6ZXjD5LL5jE6r
                DQUAOw==
                ICON
        print $icon->save_stream;

Typical saving code will be:

   die "$@" unless $x-> save( 'filename.duf');

Upon a single-frame invocation save returns 1 upon success an 0 on failure. Save requests also can be performed with package syntax:

   die "$@" unless Prima::Image-> save( 'filename.duf',
       images => [ $x]);

Saving to a stream requires explicit "codecID" to be supplied. When an image is loaded with "loadExtras", this field is always present on the image object, and is an integer that selects image encoding format.

   my @png_id =
      map  { $_-> {codecID} }
      grep { $_-> {fileShortType} =~ /^png$/i }
      @{ Prima::Image-> codecs };
   die "No png codec installed" unless @png_id;
   open FILE, "> a.png" or die "Cannot save:$!";
   binmode FILE;
   $image-> save( \*FILE, codecID => $png_id[0])
      or die "Cannot save:$@";

In multiframe invocation save returns number of successfully saved frames. File is erased though, if error occurred, even after some successfully written frames.

    die "$@" if scalar(@images) > Prima::Image-> save( $f,
       images => \@images);

All information, that is found in object hash reference 'extras', is assumed to be saved as an extra information. It is a codec's own business how it reacts on invalid and/or unacceptable information - but typical behavior is that keys that were not recognized by the codec just get ignored, and invalid values raise an error.

       $x-> {extras}-> {comments} = 'Created by Prima';
       $x-> save( $f);

Extras field 'codecID', the same one that is defined after load requests, selects explicitly a codec for an image to handle. If the codec selected is incapable of saving an error is returned. Selecting a codec is only possible with the object-driven syntax, and this information is never extracted from objects but passed to 'images' array instead.

       $x-> {extras}-> {codecID} = 1;
       $x-> save( $f);

Actual correspondence between codecs and their indices is described latter.

NB - if codecID is not given, codec is selected by the file extension.

Codecs usually are incapable of saving images in all formats, so Prima either converts an image to an appropriate format or signals an error. This behavior is governed by profile key 'autoConvert', which is 1 by default. 'autoConvert' can be present in image 'extras' structures. With autoConvert set it is guaranteed that image will be saved, but original image information may be lost. With autoConvert unset, no information will be lost, but Prima may signal an error. Therefore general-purpose save routines should be planned carefully. As an example the "Prima::Dialog::ImageDialog::SaveImageDialog" code might be useful.

When the conversion takes place, Image property 'conversion' is used for selection of an error distribution algorithm, if down-sampling is required.

This functionality is under design, but the common outlines are already set. Profile key 'append' ( 0 by default ) triggers this behavior - if it is set, then an append attempt is made.

Prima provides single function, Prima::Image-> codecs, which returns an anonymous array of hashes, where every hash entry corresponds to a registered codec. 'codecID' parameter on load and save requests is actually an index in this array. Indexes for a codecs registered once never change, so it is safe to manipulate these numbers within single program run.

Codec information that is contained in these hashes is divided into following parameters:

Unique integer value for a codec, same as index of the codec entry in results of "Prima::Image->codecs";
codec full name, string
codec vendor, string
usually underlying library versions, integers
array of strings, with file extensions that are typical to a codec. example: ['tif', 'tiff']
Description of a type of a file, that codec is designed to work with. String.
Short description of a type of a file, that codec is designed to work with. ( short means 3-4 characters ). String.
Array of strings, with some features description that a codec supports - usually codecs implement only a part of file format specification, so it is always interesting to know, what part it is.
Specify a perl module, usually inside Prima/Image directory into Prima distribution, and a package inside the module. The package contains some specific functions for work with codec-specific parameters. Current implementation defines only ::save_dialog() function, that returns a dialog that allows to change these parameters. See "Prima::Dialog::ImageDialog::SaveImageDialog" for details. Strings, undefined if empty.
1 if a codec can load images, 0 if not
1 if a codec can load images from streams, 0 otherwise
1 if a codec can handle multiframe load requests and load frames with index more than zero. 0 if not.
1 if a codec can save images, 0 if not.
1 if a codec can save images to streams, 0 otherwise
Set if a codec can save more that one frame
Set if a codec can append frames to an existing file
Array of integers - each is a combination of im:: flags, an image type, which a codec is capable of saving. First type in list is a default one; if image type that to be saved is not in that list, the image will be converted to this default type.
Hash, where keys are those that are accepted by Prima::Image-> load, and values are default values for these keys.
Array of strings, each of those is a name of extra information entry in 'extras' hash.
Hash, where keys are those that are accepted by Prima::Image-> save, and values are default values for these keys.
array of strings, with file extensions that are typical to a codec. example: ['image/xbm', 'image/x-bitmap']

This section describes parameters accepted and data returned by "Prima::Image::load"

Loading parameters

Affects how to treat alpha channel bits, if any.

If set, mixes the alpha channel with background color in case if loading to an image, or premultiplies color bits (either data or palette) with alpha, if loading to icon. Note that saving back the object will result in different image, but the object is ready to be displayed immediately.

If unset, color and eventual alpha bits, if loaded to an icon, will not be affected in any way. Note that saving back the object will result in the same image, but the object is not ready to be displayed immediately. See also: "premultiply_alpha" in Prima::Image.

When loading more than one image, this string is used to create instances of image containers. By default the calling class is used (i.e. "Prima::Image" or "Prima::Icon").
Specifies "onDataReady" event granularity in microseconds, if image codec is capable of triggering this event.

Default: 100

If set, "Prima::Icon::autoMasking" is set to "am::None" prior to the file loading.

Default: false. Only actual for "Prima::Icon" loading.

index INT
When loading from a multiframe file, selects the frame index to load.

Default: 0

map [INT]
When loading from a multiframe file, selects set of frame indexes to load.

Default: undef

If set, all available extra information will be stored in "{extras}" hash on the loaded object.

Default: false

loadAll BOOL
When loading from a multiframe file, selects that all frames are to be loaded

Default: false

When set, neither image data is not loaded, nor image dimensions are changed (newly created images have size of 1x1). Instead, "{extras}" contains "width" and "height" integers.

Default: false

Affects the action when image is incomplete, truncated, etc. If set, signals an error. Otherwise no error is signaled and whatever data could be recovered from the image are returned, and "truncated" flag is set.

Default: false

Array of hashes passed down to each frame in multiframe loads. Each frame load request will be passed an individual hash, a result of hash join of all profiles passed to "Image::load" and the nth hash in the array.
Affects how the number of frames in a file is reported in "frames" flag. If set, always scans the file for exact number. Otherwise it is up to the codec to do that.

Default: false

See also: frames.

Load output

codecID INT
Indicates the internal codec ID used to load the image. Can be used for "Image::save".
If set to a positive integer, indicates number of frames in a file. Otherwise signals that there are frames, but codec needs an expensive scan to calculate the frames (and "wantFrames" set).
When "noImageData" is in action, contains image height.
When "noIncomplete" is in action, is set if image was truncated. The value is the error string.
When "noImageData" is in action, contains image width.

Saving parameters

Affects the action when image cannot be stored in file format in its existing pixel format. If set, the system tries to convert image into a pixel format understood by the selected codec. Fails otherwise.

Default: true

codecID INT
Overrides codec selection based on filename extension.

Default: undef

BMP, the bitmap codec is not depended on external libraries and is always available.

Original bit depth, may differ from "Image::bpp".

Not valid as a saving parameter.

Bitmap compressing method.

Not valid as a saving parameter.

If loading from cursor file, contains pointer hotspot coordinates
Minimal number of colors needed to display the image
Set when loading OS/2 bitmap
Image resolution in PPM

X11, the X Consortium data file codec may depend on external libraries, but is implement internally if these are not found, and is thus always available.

Contains pointer hotspot coordinates, if any

Set of xpm-specific extension strings. Cannot be used for saving.
Contains comments to different sections
Contains pointer hotspot coordinates
Array or transparent colors. Cannot be used for saving.

Load parameters

If set to "auto" or "wipe", tries to detect whether there is are any exif tags hinting that the image has to be rotated and/or mirrored. If found, applies the transformation accordingly.

When set to "wipe", in addition to that, removes the exif tags so that subsequent image save won't result in transformed images with exifs tags still present.

This parameter requires "loadExtras" flag set, because exif tags are stored in extra JPEG data.

Load output and save input

Array of raw binary strings found in extra JPEG data.
Any comment text found in file.
If set, produces a progressively encoded JPEG file.

Default: 0

Only used for saving.

quality INT
JPEG quality, 1-100.

Default: 75

Only used for saving.

Load input

When PNG file contains alpha channel, and "alpha" is set to "blend", this color is used to blend the background. If set to "clInvalid", default PNG library background color is used.

Default: clInvalid

Not applicable for "Prima::Icon".

Override gamma value applied to the loaded image

Default: 0.45455

Current gamma value for the operating system, if specified.

Default: 2.2

Load output and save input

Default PNG library background color

Default: clInvalid, which means PNG library default

Signals whether the new frame to be blended over the existing animation, or replace it.
Delay time between frames
When set, means that the first image is a "default" frame, a special backward-compatibility image that is supposed to be excluded from the animation sequence, to be displayed only when all animation frames cannot be loaded for whatever reason.
Signals whether the frame, before being replaced, is to be erased by the background color, previous frame, or none.
Gamma value found in file.

Default: 0.45455

If set, image contains alpha channel
Embedded ICC color profiles in raw format

Default: "unspecified" and "".

If set, PNG file is interlaced

Default: 0

Frame horizontal offset from the screen
How many times the animation sequence should run, or 0 for forever.
If set, file contains a MNG datastream

Default: 0

Positive offset from the left edge of the screen to offset_x and the positive offset from the left edge of the screen to offset_y

Default: 0

Offset units

Default: pixel

See PNG docs.

Default: none

Image resolution

Default: 0

Image resolution units

Default: meter

Image scale factors

Default: 1

Image scale factor units

Default: unknown

Free-text comments found in the file

Default: "{}"

Frame vertical offset from the screen
When a paletted image contains transparent colors, returns array of palette indexes ("transparency_table") in 0-255 range, where each number is an alpha value.

Default value: empty array

One transparent color value for 24-bit PNG images.

Default value: clInvalid (i.e. none)

One transparent color value, as palette index for 8- or less- bit PNG images.

Default value: -1 (i.e. none)

Not applicable for load.

Load input

Automatically invert "PHOTOMETRIC_MINISWHITE" images

Default: 1

If set, converts 1-bit grayscale with ratio 2:1 into 2-bit grayscale (algorithm also known as faxpect).

Default: 0

Load output

TIFF "PHOTOMETRIC_XXX" constant. One of:

  MinIsWhite
  MinIsBlack
  Palette
  YCbCr
  RGB
  LogL
  LogLUV
  Separated
  MASK
  CIELAB
  DEPTH
  Unknown
    
Bits used to represent a single sample, 1-64
Number of samples per pixel, 1-4. F.ex. most images have 1 sample. Planar TIFFs may split low and high bytes in 2 samples. RGB has 3 samples, YCbCr and RGBA has 4.
"separate" images split individual samples or components (f.ex. R and G and B) into individual planes. "contiguous" mix sample bytes one after another.
Pixel sample format, one of:

  unsigned integer
  signed integer
  floating point
  untyped data
  complex signed int
  complex floating point
    
If set, TIFF is tiled
When "Fax" option set set to "true", and indeed the image was converted from 1 to 2 bits, this parameter will be set to signal this.
Compression algorithm used for reading TIFF. One of:

  NONE
  CCITTRLE
  CCITTFAX3
  CCITTFAX4
  LZW
  OJPEG
  JPEG
  NEXT
  CCITTRLEW
  PACKBITS
  THUNDERSCAN
  IT8CTPAD
  IT8LW
  IT8MP
  IT8BL
  PIXARFILM
  PIXARLOG
  DEFLATE
  ADOBE_DEFLATE
  DCS
  JBIG
  SGILOG
  SGILOG24
    

Save input

Same values as in "CompressionType". Different names are used to avoid implicit but impossible compression selection, because tibtiff can decompress many types, but compress only a few.

Load output and save input

The following keys have no specific meanings for Prima, but are both recognized for loading and saving:

  Artist
  Copyright
  DateTime
  DocumentName
  HostComputer
  ImageDescription
  Make
  Model
  PageName
  PageNumber
  PageNumber2
    
Default: 1
Default: none
Default: Prima
Default: 0
Default: 1200

For GIF animation see Prima::Image::Animate.

The following load output and save input keys are recognized:

GIF comment text
Delay in hundredth of a second between frames

Default: 1

Animation frame disposal method

  DISPOSE_NOT_SPECIFIED    = 0; # Leave frame, let new frame draw on top
  DISPOSE_KEEP             = 1; # Leave frame, let new frame draw on top
  DISPOSE_CLEAR            = 2; # Clear the frame's area, revealing bg
  DISPOSE_RESTORE_PREVIOUS = 3; # Restore the previous (composited) frame
    

Default: 0

If set, GIF is interlaced

Default: 0

Frame offset in pixels

Default: 0

How many times the GIF animation loops. 0 means indefinite.

Default: 1

GIF screen background color

Default: 0

Default: 256
Default: -1, i.e. use image width and height
Default: 0,0,0,255,255,255
Index of GIF transparent color

Default: 0

User input flag

Default: 0

Load input

Integer constant encoded as ARGB, hints the background to be used
Signals whether the new frame to be blended over the existing animation, or replace it.
Delay time between frames
Signals whether the frame, before being replaced, is to be erased by the background color or not.
If set, image contains alpha channel
Frame horizontal offset from the screen
How many times the animation sequence should run, or 0 for forever.
Frame vertical offset from the screen

Save input

WebP requires all images to have same dimensions. Also, saving the webp loading result might fail because loaded frames might only contains parts to be superimposed on each other, while saving requires always full frames. To convert webp loaded frames to something that can be saved later more-or-less identically, use "Prima::Image::webp::animation_to_frames" converter:

   use Prima qw(Image::webp);
   my @i = Prima::Icon->load('source.webp', loadAll => 1, loadExtras => 1) or die $@;
   @i = Prima::Image::webp::animation_to_frames(@i);
   die $@ if @i != Prima::Icon->save('target.webp', images => \@i);
Integer constant encoded as ARGB, hints the background to be used
compression lossless (default)|lossy|mixed
Setting between 0 and 100, 0 means off.
Min distance between key frames. Default is 9 for lossless compression, and 3 for lossy
Max distance between key frames. Default is 17 for lossless compression, and 5 for lossy
How many times the animation sequence should run, or 0 for forever.
Compression method vs size, 0 (fast) to 6 (slow)
Minimize output size (off by default)
quality INTEGER
Quality factor (0:small..100:big)
Use multi-threading if available (off by default)

Load output

Number of depth images available for the frame
Original image size before tranformations (crop, rotation, etc) are applied
Set if this is the primary image
Set if alpha is premultiplied
Array of hashes with keys type, content_type, and content.
Set it is thumbnail of the INDEXth toplevel frame

Save input

0-100
HEIC,AV1,AVC
0 gets to be a primary by default, but an be set explicitly
True if alpha is premultiplied
Array of hashes with keys type, content_type, and content.
Sets this images as a thumbnail of the INDEXth toplevel frame

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

Prima, Prima::Image, Prima::codecs

2025-07-04 perl v5.40.2

Search for    or go to Top of page |  Section 3 |  Main Index

Powered by GSP Visit the GSP FreeBSD Man Page Interface.
Output converted with ManDoc.