=head1 NAME

Imager::Cookbook - recipes working with Imager


Various simple and not so simple ways to do things with Imager.

=head1 FILES

This is described in detail in L<Imager::Files>.

=head2 Reading an image from a file

  my $image = Imager->new;

  $image->read(file=>$filename) or die $image->errstr;


  my $image = Imager->new(file => $filename)
    or die Imager->errstr;

See L<Imager::Files>.

=head2 Writing an image to a file

  $image->write(file=>$filename) or die $image->errstr;

=head2 Write an animated GIF

  # build an array of images to use in the gif
  my  @images;
  # synthesize the images or read them from files, it doesn't matter

  # write the gif
  Imager->write_multi({ file=>$filename, type=>'gif' }, @images)
    or die Imager->errstr;

See L<Imager::Files/"Writing an animated GIF"> for a more detailed

=head2 Reading multiple images from one file

Some formats, like GIF and TIFF support multiple images per file.  Use
the L<read_multi()|Imager::Files> method to read them:

  my @images = Imager->read_multi(file=>$filename)
    or die Imager->errstr;

=head2 Converting from one file format to another

This is as simple as reading the original file and writing the new
file, for single images:

  my $image = Imager->new;
  # Imager auto-detects the input file type
  $image->read(file => $input_filename)
    or die $image->errstr;
  # Imager derives the output file format from the filename
  $image->write(file => $output_filename)
    or die $image->errstr;

  # or you can supply a type parameter:
  $image->write(file => $output_filename, type => 'gif')
    or die $image->errstr;

The main issue that can occur with this is if the input file has
transparency and the output file format doesn't support that.  This
can be a problem when converting from GIF files to JPEG files for

By default, if the output format doesn't support transparency, Imager
will compose the image onto a black background.  You can override that
by supplying an C<i_background> option to C<write()> or

  $image->write(file => "foo.jpg", i_background => "#808080")
    or die $image->errstr;

Some formats support multiple files, so if you want to convert from
say TIFF to JPEG, you'll need multiple output files:

  my @images = Imager->read_multi(file => 'input.tif')
    or die Imager->errstr;
  my $index = 1;
  for my $image (@images) {
    $image->write(file => sprintf('output%02d.jpg', $index++))
      or die $image->errstr;

=head2 Transparent PNG

To save to a transparent PNG (or GIF or TIFF) you need to start with
an image with transparency.

To make a transparent image, create an image object with 2 or 4

  # RGB with alpha channel
  my $rgba = Imager->new(xsize => $width, ysize => $height, channels => 4);

  # Gray with alpha channel
  my $graya = Imager->new(xsize => $width, ysize => $height, channels => 2);

By default, the created image will be transparent.

Otherwise, if you have an existing image file with transparency,
simply read it, and the transparency will be preserved.


=head2 Creating an image

To create a simple RGB image, supply the image width and height to the
new() method:

  my $rgb = Imager->new(xsize=>$width, ysize=>$height);

If you also want an alpha channel:

  my $rgb_alpha = Imager->new(xsize=>$width, ysize=>$height, channels=>4);

To make a gray-scale image:

  my $gray = Imager->new(xsize=>$width, ysize=>$height, channels=>1);

and a gray-scale image with an alpha channel:

  my $gray_alpha = Imager->new(xsize=>$width, ysize=>$height, channels=>2);

When a new image is created this way all samples are set to zero -
black for 1 or 3 channel images, transparent black for 2 or 4 channel

You can also create paletted images and images with more than 8-bits
per channel, see L<Imager::ImageTypes> for more details.

=head2 Setting the background of a new image

To set the background of a new image to a solid color, use the box()
method with no limits, and C<< filled=>1 >>:

  $image->box(filled=>1, color=>$color);

As always, a color can be specified as an L<Imager::Color> object:

  my $white = Imager::Color->new(255, 255, 255);
  $image->box(filled=>1, color=>$white);

or you supply any single scalar that Imager::Color's new() method
accepts as a color description:

  $image->box(filled=>1, color=>'white');
  $image->box(filled=>1, color=>'#FF0000');
  $image->box(filled=>1, color=>[ 255, 255, 255 ]);

You can also fill the image with a fill object:

  use Imager::Fill;
  # create the fill object
  my $fill = Imager::Fill->new(hatch=>'check1x1')

  # let Imager create one automatically
  $image->box(fill=>{ hatch=>'check1x1' });

See L<Imager::Fill> for information on Imager's fill objects.

=head2 Create an image from raw RGB data

If your data matches the layout supported by Imager's
L<Imager::Files/RAW> file support closely enough, you can simply read
the data as if it was a raw file:

  my $img = Imager->new(
     filetype => "raw",
     xsize => $width,
     ysize => $height,
     # stored as RGBXRGBX where X is dropped
     raw_interleave => 0,
     raw_datachannels => 4,
     raw_storechannels => 3,

If the channel order doesn't match you can further use the
L<Imager::Transformations/combine()> method to extract the channels in
the correct order, eg. if the source data above was in C<BGRX> format
instead of C<RGBX>:

  my $img2 = $im->combine(src      => [ $img, $img, $img ],
                          channels => [ 2,    1,    0    ]);

Adapted from L<perl monks|https://perlmonks.org/?node_id=11139983>.

For more control you can use the setsamples() method to set pixels
directly to the image:

  # for AGBR data in $data
  my $img = Imager->new(xsize => $width, ysize => $height, channels => 4);
  my $rowbytes = $width * 4;
  my @chans = [ 3, 1, 2, 0 ];
  for my $row (0 .. $height) {
      y => $row,
      data => $data,
      offset => $rowbytes * $row,
      channels => \@chans,
      type => '8bit',


As with any CGI script it's up to you to validate data and set limits
on any parameters supplied to Imager.

For example, if you allow the caller to set the size of an output
image you should limit the size to prevent the client from specifying
an image size that will consume all available memory.

This is beside any other controls you need over access to data.

See L<CGI> for a module useful for processing CGI submitted data.

=head2 Returning an image from a CGI script

This is similar to writing to a file, but you also need to supply the
information needed by the web browser to identify the file format:

  my $img = ....; # create the image and generate the contents
  ++$|; # make sure the content type isn't buffered
  print "Content-Type: image/png\n\n";
  binmode STDOUT;
  $img->write(fd=>fileno(STDOUT), type=>'png')
    or die $img->errstr;

You need to set the Content-Type header depending on the file format
you send to the web browser.

If you want to supply a content-length header, write the image to a
scalar as a buffer:

  my $img = ....; # create the image and generate the contents
  my $data;
  $img->write(type=>'png', data=>\$data)
    or die $img->errstr;
  print "Content-Type: image/png\n";
  print "Content-Length: ",length($data),"\n\n";
  binmode STDOUT;
  print $data;

See C<samples/samp-scale.cgi> and C<samples/samp-image.cgi> for a
couple of simple examples of producing an image from CGI.

=head2 Inserting a CGI image in a page

There's occasionally confusion on how to display an image generated by
Imager in a page generated by a CGI.

Your web browser handles this process as two requests, one for the
HTML page, and another for the image itself.

Each request needs to perform validation since an attacker can control
the values supplied to both requests.

How you make the data available to the image generation code depends
on your application.

See C<samples/samp-form.cgi> and C<samples/samp-image.cgi> in the
Imager distribution for one approach.  The POD in C<samp-form.cgi>
also discusses some of the issues involved.

=head2 Parsing an image posted via CGI

C<WARNING>: file format attacks have become a common attack vector,
make sure you have up to date image file format libraries, otherwise
trying to parse uploaded files, whether with Imager or some other
tool, may result in a remote attacker being able to run their own code
on your system.

If your HTML form uses the correct magic, it can upload files to your
CGI script, in particular, you need to use C< method="post" > and
C<enctype="multipart/form-data"> in the C<form> tag, and use
C<type="file"> in the C<input>, for example:

  <form action="/cgi-bin/yourprogram" method="post" 
    <input type="file" name="myimage" />
    <input type="submit value="Upload Image" />

To process the form:


=item 1.

first check that the user supplied a file

=item 2.

get the file handle

=item 3.

have Imager read the image


  # returns the client's name for the file, don't open this locally
  my $cgi = CGI->new;
  # 1. check the user supplied a file
  my $filename = $cgi->param('myimage');
  if ($filename) {
    # 2. get the file handle
    my $fh = $cgi->upload('myimage');
    if ($fh) {
      binmode $fh;
      # 3. have Imager read the image
      my $img = Imager->new;
      if ($img->read(fh=>$fh)) {
        # we can now process the image
    # else, you probably have an incorrect form or input tag
  # else, the user didn't select a file

See C<samples/samp-scale.cgi> and C<samples/samp-tags.cgi> in the
Imager distribution for example code.

You may also want to set limits on the size of the image read, using
Imager's C<set_file_limits> method, documented in
L<Imager::Files/set_file_limits()>.  For example:

  # limit to 10 million bytes of memory usage
  Imager->set_file_limits(bytes => 10_000_000);

  # limit to 1024 x 1024
  Imager->set_file_limits(width => 1024, height => 1024);

=head1 DRAWING

=head2 Adding a border to an image

First make a new image with space for the border:

  my $border_width = ...;
  my $border_height = ...;
  my $out = Imager->new(xsize => $source->getwidth() + 2 * $border_width,
                        ysize => $source->getheight() + 2 * $border_height,
                        bits => $source->bits,
                        channels => $source->getchannels);

Then paste the source image into the new image:

  $out->paste(left => $border_width,
              top => $border_height,
              img => $source);

Whether you draw the border before or after pasting the original image
depends on whether you want the border to overlap the image, for
example a semi-transparent border drawn after pasting the source image
could overlap the edge without hiding it.

If you want a solid border you could just fill the image before
pasting the source for simplicity:

  $out->box(filled=>1, color=>'red');
  $out->paste(left => $border_width,
              top => $border_height,
              img => $source);

=head1 TEXT

=head2 Drawing text

=head2 Aligning text

=head2 Measuring text

=head2 Word wrapping text

=head2 Shearing (slanting) or Rotating text

This requires that you have Imager installed with FreeType 2.x support
installed, and that the font be created using the FreeType 2.x driver,
for example:

  my $font = Imager::Font->new(file=>$fontfile, type=>'ft2');

First you need a transformation matrix, for shearing that could be:

  my $angle_in_radians = ...;
  my $tan_angle = sin($angle_rads) / cos($angle_rads);
  # shear horizontally, supply this as y instead to do it vertically
  my $matrix = Imager::Matrix2d->shear(x=>$tan_angle);

For rotation that would be:

  my $matrix = Imager::Matrix2d->rotate(radians => $angle_in_radians);


  my $matrix = Imager::Matrix2d->rotate(degrees => $angle_in_degrees);

Feed that to the font object:

  $font->transform(matrix => $matrix);

and draw the text as normal:

  $image->string(string => $text,
                 x => $where_x,
                 y => $where_y,
                 color => $color,
                 font => $font);

See samples/slant_text.pl for a comprehensive example, including
calculating the transformed bounding box to create an image to fit the
transformed text into.


=head2 Shearing an image

=head2 Convert to gray-scale

To convert an RGB image to a gray-scale image, use the convert method:

  my $grey = $image->convert(preset => 'gray');

convert() returns a new image.

See: L<Imager::Transformations/"Color transformations">


=head2 Image format

When Imager reads a file it does a magic number check to determine the
file type, so C<foo.png> could actually be a GIF image, and Imager
will read it anyway.

You can check the actual format of the image by looking at the
C<i_format> tag.

  my $format = $image->tags(name=>'i_format');

=head2 Image spatial resolution

Most image file formats store information about the physical size of
the pixels, though in some cases that information isn't useful.

Imager stores this information in the tags C<i_xres> and C<i_yres>,
and this is always stored in dots per inch.

Some formats, including TIFF and JPEG allow you to change the units
spatial resolution information is stored in, if you set the tag that
changes this the Imager will convert C<i_xres> and C<i_yres> to those
units when it writes the file.

For example to set the resolution to 300 dpi:

  $image->settag(name => 'i_xres', value => 300);
  $image->settag(name => 'i_yres', value => 300);

If you want the file format to store the resolution in some other
unit, for example you can write a TIFF file that stores the resolution
in pixels per centimeter, you would do:

  # 150 pixels/cm
  $image->settag(name => 'i_xres', value => 150 * 2.54);
  $image->settag(name => 'i_yres', value => 150 * 2.54);
  $image->settag(name => 'tiff_resolutionunit', value => 3);

Keywords: DPI


=head2 Replacing a color with transparency
X<replacing colors>

To replace a color with transparency you can use the
L<Imager::Filters/difference()> method.

  # make a work image the same size as our input
  my $work = Imager->new(xsize => $in->getwidth, ysize => $in->getheight,
                         channels => $in->getchannels);
  # and fill it with the color we want transparent
  $work->box(filled => 1, color => $color);

  # get an image with that color replaced with transparent black
  my $out = $work->difference(other => $in);


=head2 Drop Shadows
X<drop shadow>X<effects, drop shadow>

This can be used for a glow effect as well.

First create a new image, either with an alpha channel (if you want
transparency behind the shadow) or without, if you want a background

  my $out = Imager->new
     xsize => $shadow_size * 2 + $src->getwidth,
     ysize => $shadow_size * 2 + $src->getheight,
     channels => 4,
  # fill it with your background color, if you want one
  # $out->box(filled => 1, color => $back_color);

Make a work image to render the shadow on:

  my $shadow_work = Imager->new
    xsize => $back->getwidth,
    ysize => $back->getheight,
    channels => 1,

Extract the alpha channel from the source image, first the alpha version:

  my $alpha = $src->convert(preset => "alpha");

and draw that on the work shadow:

    src => $slpha,
    left => $shadow_size,
    top => $shadow_size,

otherwise just draw a box for the non-alpha source:

    filled => 1,
    color => [ 255 ],
    xmin => $shadow_size,
    ymin => $shadow_size,
    xmax => $shadow_size + $src->getwidth() - 1,
    ymax => $shadow_size + $src->getheight() - 1,

Blur the work shadow:

  $shadow_work->filter(type => "gaussian", stddev => $shadow_size);

Convert it to an RGB image with alpha:

  $shadow_work = $shadow_work->convert
      matrix => [ [ 0, $red / 255 ],
                   [ 0, $green / 255 ],
                   [ 0, $blue / 255 ],
                   [ 1 ] ]

Draw that on the output image:

  $out->rubthrough(src => $shadow_work);

Draw our original image on the output image, perhaps with an offset:

    src => $src,
    tx => $shadow_size + $x_offset,
    ty => $shadow_size + $y_offset,

See F<samples/drop_shadow.pl> for an example of this recipe.

=head1 AUTHOR

Tony Cook <tony@develop-help.com>

=head1 SEE ALSO

L<Imager>, L<Imager::Files>, L<Imager::Draw>.