Registered processors allow you to modify data, e.g. resizing an image.

You can register as many processors as you like.

Let's say we have a Dragonfly app

app = Dragonfly[:images]

and an image object (actually a Job object)...

image = app.fetch('some/uid')

...OR a Dragonfly model accessor...

image = @album.cover_image

We can process it using any processing methods that have been registered with the processor.

Lazy evaluation

new_image = image.process(:some_method)

doesn't actually do anything until you call something on the returned Job object, like url, data, etc.

Bang method


modifies the image object itself, rather than returning a new object.

ImageMagick Processor

See ImageMagick.

Custom Processors

To register a single custom processor:

app.processor.add :watermark do |temp_object, *args|
  # use, temp_object.path, temp_object.file, etc.
  SomeLibrary.add_watermark(, 'some/watermark/file.png')  # return a String, Pathname, File or Tempfile


new_image = image.process(:watermark)

You can create a class like the ImageMagick one above, in which case all public methods will be counted as processing methods. Each method takes the temp_object as its argument, plus any other args.

class MyProcessor

  def coolify(temp_object, opts={})
    SomeLib.coolify(, opts)

  def uglify(temp_object, ugliness)
    `uglify -i #{temp_object.path} -u #{ugliness}`


  def my_helper_method    # do stuff




new_image = image.coolify(:some => :args)

new_image = image.uglify(:loads)