Resizing and cropping

Creating thumbnails (by resizing images) and cropping are most frequent works about images. This guide explains ways to deal with sizes of images.

Above all, to get the current size of the image check width and height properties:

>>> from urllib.request import urlopen
>>> from wand.image import Image
>>> f = urlopen('http://pbs.twimg.com/profile_images/712673855341367296/WY6aLbBV_normal.jpg')
>>> with Image(file=f) as img:
...     width = img.width
...     height = img.height
...
>>> f.close()
>>> width
48
>>> height
48

If you want the pair of (width, height), check size property also.

Note

These three properties are all readonly.

Resize images

It scales an image into a desired size even if the desired size is larger than the original size. ImageMagick provides so many algorithms for resizing. The constant FILTER_TYPES contains names of filtering algorithms.

See also

ImageMagick Resize Filters
Demonstrates the results of resampling three images using the various resize filters and blur settings available in ImageMagick, and the file size of the resulting thumbnail images.

Image.resize() method takes width and height of a desired size, optional filter ('undefined' by default which means IM will try to guess best one to use) and optional blur (default is 1). It returns nothing but resizes itself in-place.

>>> img.size
(500, 600)
>>> img.resize(50, 60)
>>> img.size
(50, 60)

Sample images

Although Image.resize() provides many filter options, it’s relatively slow. If speed is important for the job, you’d better use Image.sample() instead. It works in similar way to Image.resize() except it doesn’t provide filter and blur options:

>>> img.size
(500, 600)
>>> img.sample(50, 60)
>>> img.size
(50, 60)

Crop images

To extract a sub-rectangle from an image, use the crop() method. It crops the image in-place. Its parameters are left, top, right, bottom in order.

>>> img.size
(200, 300)
>>> img.crop(10, 20, 50, 100)
>>> img.size
(40, 80)

It can also take keyword arguments width and height. These parameters replace right and bottom.

>>> img.size
(200, 300)
>>> img.crop(10, 20, width=40, height=80)
>>> img.size
(40, 80)

There is an another way to crop images: slicing operator. You can crop an image by [left:right, top:bottom] with maintaining the original:

>>> img.size
(300, 300)
>>> with img[10:50, 20:100] as cropped:
...     print(cropped.size)
...
(40, 80)
>>> img.size
(300, 300)

Specifying gravity along with width and height keyword arguments allows a simplified cropping alternative.

>>> img.size
(300, 300)
>>> img.crop(width=40, height=80, gravity='center')
>>> img.size
(40, 80)

Transform images

Use this function to crop and resize and image at the same time, using ImageMagick geometry strings. Cropping is performed first, followed by resizing.

For example, if you want to crop your image to 300x300 pixels and then scale it by 2x for a final size of 600x600 pixels, you can call:

img.transform('300x300', '200%')

Other example calls:

# crop top left corner
img.transform('50%')

# scale height to 100px and preserve aspect ratio
img.transform(resize='x100')

# if larger than 640x480, fit within box, preserving aspect ratio
img.transform(resize='640x480>')

# crop a 320x320 square starting at 160x160 from the top left
img.transform(crop='320+160+160')

See also

ImageMagick Geometry Specifications
Cropping and resizing geometry for the transform method are specified according to ImageMagick’s geometry string format. The ImageMagick documentation provides more information about geometry strings.

Seam carving (also known as content-aware resizing)

New in version 0.3.0.

Seam carving is an algorithm for image resizing that functions by establishing a number of seams (paths of least importance) in an image and automatically removes seams to reduce image size or inserts seams to extend it.

In short: you can magickally resize images without distortion! See the following examples:

Original Resized
seam.jpg seam-resize.jpg
Cropped Seam carving
seam-crop.jpg seam-liquid.jpg

You can easily rescale images with seam carving using Wand: use Image.liquid_rescale() method:

>>> image = Image(filename='seam.jpg')
>>> image.size
(320, 234)
>>> with image.clone() as resize:
...     resize.resize(234, 234)
...     resize.save(filename='seam-resize.jpg')
...     resize.size
...
(234, 234)
>>> with image[:234, :] as crop:
...     crop.save(filename='seam-crop.jpg')
...     crop.size
...
(234, 234)
>>> with image.clone() as liquid:
...     liquid.liquid_rescale(234, 234)
...     liquid.save(filename='seam-liquid.jpg')
...     liquid.size
...
(234, 234)

Note

It may raise MissingDelegateError if your ImageMagick is configured --without-lqr option. In this case you should recompile ImageMagick.

See also

Seam carving — Wikipedia
The article which explains what seam carving is on Wikipedia.

Note

The image seam.jpg used in the above example is taken by D. Sharon Pruitt and licensed under CC-BY-2.0. It can be found the original photography from Flickr.