django-thumbnails 0.5.0

Design:

• Uses Django Storage API
• Uses flexible meta data store. Supports DB and Redis backend.
• Supports creating thumbnails in different formats, for example from JPG to WEBP to reduce file size

Supported image formats:

• JPG/JPEG
• GIF
• PNG
• WEBP

Installation

• Add thumbnails to INSTALLED_APPS in settings.py.
• Run python manage.py migrate to create database metadata backend.

Usage

settings.py:

THUMBNAILS = {
    'METADATA': {
        'BACKEND': 'thumbnails.backends.metadata.DatabaseBackend',
    },
    'STORAGE': {
        'BACKEND': 'django.core.files.storage.FileSystemStorage',
        # You can also use Amazon S3 or any other Django storage backends
    }
    'SIZES': {
        'small': {
            'PROCESSORS': [
                {'PATH': 'thumbnails.processors.resize', 'width': 10, 'height': 10},
                {'PATH': 'thumbnails.processors.crop', 'width': 80, 'height': 80}
            ],
            'POST_PROCESSORS': [
                {
                    'PATH': 'thumbnails.post_processors.optimize',
                    'png_command': 'optipng -force -o7 "%(filename)s"',
                    'jpg_command': 'jpegoptim -f --strip-all "%(filename)s"',
                },
            ],
        },
        'large': {
            'PROCESSORS': [
                {'PATH': 'thumbnails.processors.resize', 'width': 20, 'height': 20},
                {'PATH': 'thumbnails.processors.flip', 'direction': 'horizontal'}
            ],
        }
    }
}

If you prefer to use Redis as your metadata storage backend (for performance reasons):

THUMBNAILS = {
    'METADATA': {
        'PREFIX': 'thumbs',
        'BACKEND': 'thumbnails.backends.metadata.RedisBackend',
        'db': 2,
        'port': 6379,
        'host': 'localhost',
    },
}

Image Processors

django-thumbnails comes with a few builtin image processors:

# To use the following processors, put the arguments of processors in SIZES definition
thumbnails.processors.resize(width, height, method) ## `method` can be `stretch`, `fit` or `fill`
thumbnails.processors.rotate(degrees)
thumbnails.processors.flip(direction)
thumbnails.processors.crop(width, height, center)

Processors are applied sequentially in the same order of definition.

Storage Backend

New in version 0.5.0 is per field, customizable storage backend. If you want specific fields to use a different storage backend, you can specify it directly when declaring the field. e.g:

class Food(models.Model):
    image = ImageField(storage=FileSystemStorage(), upload_to='food')

Storage that is specified on field will be used instead of storage that is specified in the settings.

In python:

from thumbnails.fields import ImageField

class Food(models.Model):
    image = ImageField()


food = Food.objects.latest('id')
food.image.thumbnails.all()
food.image.thumbnails.small  # Generates "small" sized thumbnail
food.image.thumbnails.large  # Generates "large" sized thumbnail
food.image.thumbnails.small.url  # Returns "small" sized thumbnail URL

And here's how you'd use it in Django's template:

{{ food.image.thumbnails.small.url }}  # Returns "small" sized thumbnail URL

Use resize_source_to to resize your image while saving it:

from thumbnails.fields import ImageField

class Food(models.Model):
    image = ImageField(resize_source_to="medium")

Assuming medium is the size that you define in the settings. By passing medium your saved image will be resized into medium's size

Use pregenerated_sizes to save your thumbnails into storage backend while saving it:

from thumbnails.fields import ImageField

class Food(models.Model):
    image = ImageField(pregenerated_sizes=["small", "large", "medium")

When deleting an image, you can opt to retain thumbnails by doing this:

banner.image.delete(with_thumbnails=False)

Performance

If you need to fetch multiple thumbnails at once, use the provided fetch_thumbnails function for better performance. fetch_thumbnails uses Redis pipeline to retrieve thumbnail metadata in one go, avoiding multiple round trips to Redis.

from thumbnails.field import fetch_thumbnails

food_a = Food.objects.get(id=1)
food_b = Food.objects.get(id=2)

fetch_thumbnails([food_a.image, food_b.image], ['small', 'large'])

This way, when we get thumbnails like thumbnail1.size_small or even thumbnail1.all() we won't query to redis anymore. This feature is currently only available for Redis metadata Backend.

Management Commands

If you changed your size definition and want to regenerate the thumbnails, use:

python manage.py delete_thumbnails --model=app.Model --size=thumbnail_size_to_delete

Running Tests

To run tests:

`which django-admin.py` test thumbnails --settings=thumbnails.tests.settings --pythonpath=.

Changelog

Version 0.5.0 (2021-05-1)

• You can now pass in storage kwarg into ImageField so you can specify different storage backends for different fields. Thanks @marsha97!
• Calling image.delete(with_thumbnails=True) will delete original image along with all thumbnails. Thanks @marsha97!

Version 0.4.0 (2021-01-08)

• Support for Django >= 3.0. Thanks @christianciu!
• Added pregenerated_sizes to ImageField to allow thumbnails to be pregenerated on upload. Thanks @marsha97!
• Thumbnails can be generated in different formats (e.g: JPG source image to WEBP thumbnail). Thanks @yosephbernandus!

Version 0.3.2

• Fixed another bug in fetch_thumbnails() bug. Thanks @marsha97!

Version 0.3.1

• Fixed fetch_thumbnails() bug. Thanks @marsha97!

Version 0.3.0

• Added fetch_thumbnails() command to fetch multiple thumbnail metadata from Redis. Thanks @marsha97!

Version 0.2.2

• Fixed RedisBackend.get_thumbnail() bug that may cause excessive trips to Redis. Thanks @marsha97!

Version 0.2.1

• Add support for Django 1.11, 2.0 and 2.1

Version 0.2.0

• Improves performance on fields that have a large number of thumbnails
• Add support for Django 1.8, 1.9 and 1.10

Version 0.1.3

• Fixes deprecation warning in Django 1.8

Version 0.1.2

• Fixes deprecation warning in Django 1.8

Version 0.1.1

• Use shortuuid instead of uuid4() to be more space efficient

Version 0.1.0

• First public release

As of February 2015, this library is suitable for production use and has been used for more than a year in Stamps, an Indonesian based CRM/loyalty system.