Back to home

PrerequisitesGetting startedCreating the notebookLoading in the imageCreating the destination image and adding the icon layersCreating the text layerSaving and displaying the destination imageSummaryResources and further reading
Composing Images and Text with Python main image

Composing Images and Text with Python

This post will use the Pillow image library to add text and icons layers to an image programmatically.


  1. An image that is 1000x600 pixels.
  2. Some demo icons 100x100 pixels.
  3. Familiarity with [Pipenv]. See here for my post on Pipenv.
  4. Familiarity with JupyterLab. See here for my post on JupyterLab.
  5. Open Sans Bold font installed to fonts/OpenSans-Bold.ttf.

Getting started

Let's create the hello-img-layers directory and install Pillow.

1 2 3 4 5 6 7 8 9 10 # Make the `hello-img-layers` directory $ mkdir hello-img-layers $ cd hello-img-layers # Create a folder to place your icons $ mkdir icons # Init the virtual environment $ pipenv --three $ pipenv install pillow $ pipenv install --dev jupyterlab

At this stage, add your 1000x600 image to the root directory (hello-img-layers) as base_img.png (at least that is what I am using) and add your 100x100 icons to hello-img-layers/icons/.

Now we can start up the notebook server.

1 2 3 # Startup the notebook server $ pipenv run jupyter-lab # ... Server is now running on http://localhost:8888/lab

The server will now be up and running.

Creating the notebook

Once on http://localhost:8888/lab, select to create a new Python 3 notebook from the launcher.

Ensure that this notebook is saved in hello-img-layers/docs/<your-file-name>.

We will create four cells to handle four parts of this mini project:

  1. Load the image in.
  2. Creating the destination image and adding the icon layers.
  3. Creating the text layer.
  4. Saving and displaying the destination image.

Loading in the image

This section simply loads the image in var base_img (assuming you are following the directory structure where the notebook is in the docs folder).

1 2 3 from PIL import Image # Concatenating an image base_img ='../base_img.png')

Creating the destination image and adding the icon layers

We create the destination layer here based on the base images width and height and paste that base image back in.

Then we make use of the glob library to fetch all the paths to our icons and paste them on top of the base image.

1 2 3 4 5 6 7 8 9 10 11 12 13 # Add in icons using a glob import glob from os.path import join, dirname, abspath dst_img ='RGBA', (base_img.width, base_img.height)) dst_img.paste(base_img, (0, 0)) icons_dir = join(dirname(abspath("__file__")), '../icons') icons = glob.glob(f"{icons_dir}/*") for i, icon_path in enumerate(icons): icon = # @see dst_img.paste(icon, (60 + (i * icon.width) + (i * 20), base_img.height - 100 - icon.height), icon)

Creating the text layer

This layer will load our Open Sans font and draw that onto the image.

The max_text_width=25 that I selected here was through trial and error. There may be a more reusable way to handle this for different sized images.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 import os # Add your own font in font = ImageFont.truetype(os.path.join('fonts/OpenSans-Bold.ttf'), 58) import textwrap text = "Sample Text that is too long but let us write a bunch anyway" print(dst_img.width) max_text_width = 25 wrapped_text = textwrap.wrap(text, width=max_text_width) # setup up a variable to invoke our text method draw = ImageDraw.Draw(dst_img) for i, line in enumerate(wrapped_text): # draw.text((0, height - (i * 20)), line, (0, 0, 0), font=font) draw.text((60, 100 + (i * 80)),line,(255,255,255),font=font)

Saving and displaying the destination image

Finally, we will save and display our base image.

1 2 3 4'../dest_img.png') # Display the image inline display('../dest_img.png'))



Today's post demonstrated how to use the Pillow package to add text and icons layers to an image programmatically.

This will be the technique I use for helping to build out my blog post main images (including this one!).

Resources and further reading

Dennis O'Keeffe

  • Melbourne, Australia

Hi, I am a professional Software Engineer. Formerly of Culture Amp, UsabilityHub, Present Company and NightGuru.
I am currently working on, Den Dribbles and LandPad .

Related articles


Get fresh posts + news direct to your inbox.

No spam. We only send you relevant content.