We all have a treasure trove of old photos passed down from our parents, grandparents, and so on. Often, these photos are in black and white, and sometimes, we wish to bring them to life in color. There are numerous websites online where you can achieve this with just a few clicks right in your browser. However, the problem with these sites lies in their many restrictions on quantity and quality. Additionally, these services often require registration and subscriptions, which can be quite frustrating. In this post, I'd like to share how you can colorize any number of photos using the fantastic tool, DeOldify, without the hassle of limitations, registrations, or subscriptions. DeOldify has an amazing introduction, and you may use it, but for non-prepared people, it may seem a bit difficult. Here, I'd like to demonstrate a simpler way of using DeOldify. Preparing the Environment To get started, you'll need a computer running Linux or macOS. If you're using Windows, you can still proceed by utilizing the Windows Subsystem for Linux (WSL). There is a list of required packages for Ubuntu: sudo apt update
sudo apt install python3-pip wget git ffmpeg libsm6 libxext6 You may use the list as a reference if you have another Linux distribution. For macOS (unfortunately, this is an approximate list as I don't have a clear macOS for testing): brew install python wget git ffmpeg libsm libxext The next step is creating a separate folder named Sandbox to do everything in. You can choose any other name or even choose not to create anything and skip this step altogether. # Optional
mkdir Sandbox
cd Sandbox Then clone DeOldify: git clone https://github.com/jantic/DeOldify.git DeOldify doesn't have any releases or tags, which means that we work with the master branch. I believe that everything should work with the latest master at any time. But just in case, here is a revision number with which I tested everything: be725ca6c5f411c47550df951546537d7202c9bc. If you want, you may check it out: # Optional
cd DeOldify
git checkout be725ca6c5f411c47550df951546537d7202c9bc
cd .. After that, install all Python dependencies: pip3 install --user -r DeOldify/requirements.txt
pip3 install --user -r DeOldify/requirements-colab.txt
pip3 install --user -r DeOldify/requirements-dev.txt Next, create a directory for the models, and download the model: mkdir -p DeOldify/models
wget https://data.deepai.org/deoldify/ColorizeArtistic_gen.pth -O DeOldify/models/ColorizeArtistic_gen.pth That's it! The environment is ready; the hardest part is behind us. Now, let's write a bit of Python code. Colorize Everything The most interesting and useful function for us is ModelImageVisualizer.plot_transformed_image(). All we need to do is write code around it. Here is my implementation: #!/usr/bin/python3

from deoldify import device
from deoldify.device_id import DeviceId

# choices:  CPU, GPU0...GPU7
device.set(device=DeviceId.GPU0)

from deoldify.visualize import *
import warnings

import sys
import os

warnings.filterwarnings("ignore", category=UserWarning, message=".*?Your .*? set is empty.*?")

# Play with this constant!
render_factor = 35

input_dir = sys.argv[1]
output_dir = sys.argv[2]

if not os.path.isdir(input_dir):
    print("input directory is not a directory or not exist")
    sys.exit(1)

if os.path.exists(output_dir):
    print("out directory is already exist")
    sys.exit(1)

os.makedirs(output_dir)

root_dir = Path(os.environ['PYTHONPATH'])
colorizer = get_image_colorizer(root_folder=root_dir, artistic=True)
for filename in os.listdir(input_dir):
    f = os.path.join(input_dir, filename)
    if os.path.isfile(f) and (f.endswith(".png") or f.endswith(".jpg") or f.endswith(".jpeg")):
        image_path = colorizer.plot_transformed_image(
            path=f,
            results_dir=Path(output_dir),
            render_factor=render_factor,
            compare=True,
            watermarked=False
        )
        print("{} ready".format(image_path)) Feel free to copy, paste, and modify it however you want. The script reads all .png, .jpg, and .jpeg images from the directory passed as the first script argument and colorizes them, saving the results into the directory passed as the second script argument. Note that the output directory must not exist; the script creates it itself. I saved the script in the Sandbox/runner.py file, but you may use a different location and name if you prefer. Let's try it! I placed a black-and-white photo into Sandbox/photos/in, and the output directory is Sandbox/photos/out: # from Sandbox directory
PYTHONPATH=./DeOldify/ python3 runner.py photos/in/ photos/out/ The environment variable PYTHONPATH must refer to the DeOldify directory. Here is the result: We all have a treasure trove of old photos passed down from our parents, grandparents, and so on. Often, these photos are in black and white, and sometimes, we wish to bring them to life in color. There are numerous websites online where you can achieve this with just a few clicks right in your browser. However, the problem with these sites lies in their many restrictions on quantity and quality. Additionally, these services often require registration and subscriptions, which can be quite frustrating. In this post, I'd like to share how you can colorize any number of photos using the fantastic tool, DeOldify , without the hassle of limitations, registrations, or subscriptions. DeOldify DeOldify has an amazing introduction , and you may use it, but for non-prepared people, it may seem a bit difficult. Here, I'd like to demonstrate a simpler way of using DeOldify . DeOldify amazing introduction DeOldify Preparing the Environment To get started, you'll need a computer running Linux or macOS . If you're using Windows , you can still proceed by utilizing the Windows Subsystem for Linux (WSL) . Linux macOS Windows Windows Subsystem for Linux (WSL) There is a list of required packages for Ubuntu : Ubuntu sudo apt update
sudo apt install python3-pip wget git ffmpeg libsm6 libxext6 sudo apt update
sudo apt install python3-pip wget git ffmpeg libsm6 libxext6 You may use the list as a reference if you have another Linux distribution. Linux For macOS (unfortunately, this is an approximate list as I don't have a clear macOS for testing): macOS macOS brew install python wget git ffmpeg libsm libxext brew install python wget git ffmpeg libsm libxext The next step is creating a separate folder named Sandbox to do everything in. You can choose any other name or even choose not to create anything and skip this step altogether. Sandbox Sandbox # Optional
mkdir Sandbox
cd Sandbox # Optional
mkdir Sandbox
cd Sandbox Then clone DeOldify : DeOldify git clone https://github.com/jantic/DeOldify.git git clone https://github.com/jantic/DeOldify.git DeOldify doesn't have any releases or tags, which means that we work with the master branch. I believe that everything should work with the latest master at any time. But just in case, here is a revision number with which I tested everything: DeOldify master be725ca6c5f411c47550df951546537d7202c9bc . If you want, you may check it out: be725ca6c5f411c47550df951546537d7202c9bc # Optional
cd DeOldify
git checkout be725ca6c5f411c47550df951546537d7202c9bc
cd .. # Optional
cd DeOldify
git checkout be725ca6c5f411c47550df951546537d7202c9bc
cd .. After that, install all Python dependencies: pip3 install --user -r DeOldify/requirements.txt
pip3 install --user -r DeOldify/requirements-colab.txt
pip3 install --user -r DeOldify/requirements-dev.txt pip3 install --user -r DeOldify/requirements.txt
pip3 install --user -r DeOldify/requirements-colab.txt
pip3 install --user -r DeOldify/requirements-dev.txt Next, create a directory for the models, and download the model: mkdir -p DeOldify/models
wget https://data.deepai.org/deoldify/ColorizeArtistic_gen.pth -O DeOldify/models/ColorizeArtistic_gen.pth mkdir -p DeOldify/models
wget https://data.deepai.org/deoldify/ColorizeArtistic_gen.pth -O DeOldify/models/ColorizeArtistic_gen.pth That's it! The environment is ready; the hardest part is behind us. Now, let's write a bit of Python code. Colorize Everything The most interesting and useful function for us is ModelImageVisualizer.plot_transformed_image() . ModelImageVisualizer.plot_transformed_image() All we need to do is write code around it. Here is my implementation: #!/usr/bin/python3

from deoldify import device
from deoldify.device_id import DeviceId

# choices:  CPU, GPU0...GPU7
device.set(device=DeviceId.GPU0)

from deoldify.visualize import *
import warnings

import sys
import os

warnings.filterwarnings("ignore", category=UserWarning, message=".*?Your .*? set is empty.*?")

# Play with this constant!
render_factor = 35

input_dir = sys.argv[1]
output_dir = sys.argv[2]

if not os.path.isdir(input_dir):
    print("input directory is not a directory or not exist")
    sys.exit(1)

if os.path.exists(output_dir):
    print("out directory is already exist")
    sys.exit(1)

os.makedirs(output_dir)

root_dir = Path(os.environ['PYTHONPATH'])
colorizer = get_image_colorizer(root_folder=root_dir, artistic=True)
for filename in os.listdir(input_dir):
    f = os.path.join(input_dir, filename)
    if os.path.isfile(f) and (f.endswith(".png") or f.endswith(".jpg") or f.endswith(".jpeg")):
        image_path = colorizer.plot_transformed_image(
            path=f,
            results_dir=Path(output_dir),
            render_factor=render_factor,
            compare=True,
            watermarked=False
        )
        print("{} ready".format(image_path)) #!/usr/bin/python3

from deoldify import device
from deoldify.device_id import DeviceId

# choices:  CPU, GPU0...GPU7
device.set(device=DeviceId.GPU0)

from deoldify.visualize import *
import warnings

import sys
import os

warnings.filterwarnings("ignore", category=UserWarning, message=".*?Your .*? set is empty.*?")

# Play with this constant!
render_factor = 35

input_dir = sys.argv[1]
output_dir = sys.argv[2]

if not os.path.isdir(input_dir):
    print("input directory is not a directory or not exist")
    sys.exit(1)

if os.path.exists(output_dir):
    print("out directory is already exist")
    sys.exit(1)

os.makedirs(output_dir)

root_dir = Path(os.environ['PYTHONPATH'])
colorizer = get_image_colorizer(root_folder=root_dir, artistic=True)
for filename in os.listdir(input_dir):
    f = os.path.join(input_dir, filename)
    if os.path.isfile(f) and (f.endswith(".png") or f.endswith(".jpg") or f.endswith(".jpeg")):
        image_path = colorizer.plot_transformed_image(
            path=f,
            results_dir=Path(output_dir),
            render_factor=render_factor,
            compare=True,
            watermarked=False
        )
        print("{} ready".format(image_path)) Feel free to copy, paste, and modify it however you want. The script reads all .png , .jpg , and .jpeg images from the directory passed as the first script argument and colorizes them, saving the results into the directory passed as the second script argument. Note that the output directory must not exist; the script creates it itself. I saved the script in the Sandbox/runner.py file, but you may use a different location and name if you prefer. .png .jpg .jpeg Sandbox/runner.py Let's try it! I placed a black-and-white photo into Sandbox/photos/in , and the output directory is Sandbox/photos/out : Sandbox/photos/in Sandbox/photos/out # from Sandbox directory
PYTHONPATH=./DeOldify/ python3 runner.py photos/in/ photos/out/ # from Sandbox directory
PYTHONPATH=./DeOldify/ python3 runner.py photos/in/ photos/out/ The environment variable PYTHONPATH must refer to the DeOldify directory. The environment variable PYTHONPATH must refer to the DeOldify directory. PYTHONPATH DeOldify Here is the result:

Walkthroughs, tutorials, guides, and tips. This story will teach you how to do something new or how to do something better.

How to Colorize a Black and White Photo

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Why Being Likable At Work Matters

Windows Sticky Keys Exploit: The War Veteran That Never Dies

Is ‘bias for action’ making product managers lazier?

Zero People Charged With Online Pirating, Swedish Prosecutor's Office Reports

What No One Told Me About Being a Product Manager at an Early Stage Startup

0 to $1 Million in 5 Months for Our Cryptocurrency Startup

Why Being Likable At Work Matters

Windows Sticky Keys Exploit: The War Veteran That Never Dies

Is ‘bias for action’ making product managers lazier?

Zero People Charged With Online Pirating, Swedish Prosecutor's Office Reports

What No One Told Me About Being a Product Manager at an Early Stage Startup

0 to $1 Million in 5 Months for Our Cryptocurrency Startup

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps