DEV Community

Cover image for How to apply filters to images with Python
Juan Benitez
Juan Benitez

Posted on • Updated on

How to apply filters to images with Python

A few weeks ago, we received a new ticket: “Users want to be able to apply filters to their pictures”, yes, something like Instagram does. We immediately thought about the Lightroom API, but after some research, we came to the conclusion that it doesn’t quite meet our needs. We were searching for something like a library, where we can just load the image, apply the filters and then save the new image with the filters applied, since we couldn’t find it we decided to build one.

FImage

FImage is a Python module to apply and create multiple filters to images, it exposes an API that you can use for applying the different color transformations to the images. It works by converting the image to an RGB matrix and applying different math formulas to it. We used NumPy for all the matrix operations since it is faster and optimized, and Pillow for handling the loading and saving of the images.

How to use it?

First, we need to install it, for this you need to be using Python 3.6 or greater to be able to use FImage.

pip install fimage
Enter fullscreen mode Exit fullscreen mode

And for these examples, I’m gonna use this picture to apply it filters:

my_picture.jpg

Applying a simple filter

Create a file app.py with:

from fimage import FImage
from fimage.filters import Sepia


def main():
    # replace 'my_picture.jpg' with the path to your image
    image = FImage('my_picture.jpg')

    # apply the Sepia filter to the image
    image.apply(Sepia(90))

    # save the image with the applied filter
    image.save('my_picture_sepia.jpg')


if __name__ == "__main__":
    main()

Enter fullscreen mode Exit fullscreen mode

Now, just run it :

python app.py
Enter fullscreen mode Exit fullscreen mode

And this is how the new image my_picture_sepia.jpg looks like after the filter was applied.
my_picture_sepia.jpg

Applying multiple filters

FImage offers more filters besides the Sepia one, even you can combine multiples filters to give a better look to your picture.

Modify the file app.py to import more filters from FImage

from fimage import FImage
from fimage.filters import Contrast, Brightness, Saturation


def main():
    image = FImage('my_picture.jpg')

    # apply the mutiple filters to the image
    image.apply(
        Saturation(20),
        Contrast(25),
        Brightness(15)
    )

    # save the image with the applied filter
    image.save('my_picture_mixed.jpg')


if __name__ == "__main__":
    main()
Enter fullscreen mode Exit fullscreen mode

We run it by

python app.py
Enter fullscreen mode Exit fullscreen mode

And our new my_picture_mixed.jpg looks like
my_picture_mixed.jpg

The order in which the filters are passed to the apply function matters, this is because the filters are applied sequentially, so the next filter will be applied over the resultant image from the previous one.

Presets

Presets are just the combinations of multiple filters with already defined adjustment values.

Let’s change our app.py one more time to use the Presets

from fimage import FImage
from fimage.presets import SinCity


def main():
    # replace 'my_picture.jpg' with the path to your image
    image = FImage('my_picture.jpg')

    # apply the SinCity preset to the image
    image.apply(SinCity())

    # save the image with the applied preset
    image.save('my_picture_sincity.jpg')


if __name__ == "__main__":
    main()
Enter fullscreen mode Exit fullscreen mode

After we run it, we get our new my_picture_sincity.jpg
my_picture_sincity.jpg

Custom Presets

If you like the look your picture got after testing different filters and want to store this combination for applying it to more pictures, you can create your own Preset by just extending the Preset Class and specifying these filters and their adjust values in it.

In our app.py let’s do

from fimage import FImage
from fimage.presets import Preset
from fimage.filters import Contrast, Brightness, Saturation


# Create my custom preset and specify the filters to apply
class MyOwnPreset(Preset):
    transformations = [
        Contrast(30),
        Saturation(50),
        Brightness(10),
    ]


def main():
    # replace 'my_picture.jpg' with the path to your image
    image = FImage('my_picture.jpg')

    # apply MyOwnPreset to the image
    image.apply(MyOwnPreset())

    # save the image with the applied preset
    image.save('my_picture_custom.jpg')


if __name__ == "__main__":
    main()
Enter fullscreen mode Exit fullscreen mode

The new my_picture_custom.jpg

my_picture_custom.jpg

This is basic usage of FImage, we are still developing it, and it would be really great any feedback or contribution you have.

GitHub Repo

You can follow me on Twitter and GitHub to be up to date with all my projects and content.

Top comments (7)

Collapse
 
jbristow profile image
Jon Bristow

What advantages does your library provide over imagemagick?

How does it compare speed-wise?

Collapse
 
juanbenitezdev profile image
Juan Benitez

FImage is a pure Python library and it allows you to create and even modify some of the filters ( I'm not sure if you can do this with ImageMagick).

About the speed, I will need to do some tests, in average FImage lasts around one second applying a filter, this depends on the size of the image as well.

Collapse
 
jbristow profile image
Jon Bristow

If you're using numPy, you're already using c libraries and not Pure Python. Equivalent to using one of the python based shims built around libmagic.

Collapse
 
mccurcio profile image
Matt Curcio

Awesome, Thanks

Collapse
 
cricketsamya profile image
Sameer

Awesome article!

Collapse
 
grayson_gaming_64e043e8ef profile image
Grayson Gaming

hey, sorry I'm late as hell to this. But I'm making a software project that uses fimage and I cannot get it to install using pip, I've tried lots of different methods, anyone know how to get it installed to my python library?

Collapse
 
rencutie profile image
Marwan

hello
how can I convert an image edit by one of these filters to a cv2 format ?
thank you