Facemation

Turn your daily selfies into a good-looking timelapse.

This script automatically scales, rotates, crops, and captions all frames so your eyes are aligned in each photo, and compiles these frames into a timelapse.

Usage instructions

If you need help with installing or using Facemation, please feel free to start a discussion or contact me directly.

How to install

Windows 10 / Windows 11

1. Download FFmpeg. (If you have 7-Zip installed, download the .7z archive instead.)
2. Unzip the downloaded FFmpeg archive into a new directory.
3. Download the latest version of Facemation for Windows.
4. Unzip the downloaded Facemation archive into another new directory.
5. Enter the directory where you unzipped FFmpeg, enter the bin directory, and copy the file ffmpeg.exe to the directory where you unzipped Facemation.
   You should now have facemation.exe and ffmpeg.exe in the same directory.

Linux

1. Install FFmpeg. On Ubuntu/Debian, you can simply do apt install ffmpeg.
2. Download the latest version of Facemation for Linux.
3. Unzip the downloaded archive into a new directory.

macOS

Unfortunately, I don't have macOS, which means that I cannot create an executable for macOS systems. Your best bet is probably to run the Python scripts directly by following the development instructions below, but even then I cannot guarantee it will work. If you have suggestions for how I can solve this, please let me know by opening an issue, starting a discussion, or contacting me directly.

How to use

1. Enter the directory where you installed Facemation.
2. Put the images you want Facemation to process in the input directory.
3. Rename files if necessary so that they are in the right order. Images are processed in natural sort order.
4. Execute facemation by double-clicking it.
5. Check the created video in output/facemation.mp4.

All intermediate results are heavily cached, so subsequent runs are much faster.

How to configure

You can change how Facemation behaves by editing the config.py file. Below are some examples of how you can configure Facemation. Check config_default.py for a list of all options.

Disable FFmpeg

If you do not have FFmpeg, you can disable it. Facemation will still work, but will skip the final step of creating a video.

config = {
    "ffmpeg": {
        "enabled": False,
    }
}

Show the date in each frame

This code assumes that each filename is something like IMG_20230104_174807.jpg.

from datetime import datetime

config = {
    "caption": {
        "enabled": True,
        "generator": (lambda filename: str(datetime.strptime(filename, "IMG_%Y%m%d_%H%M%S.jpg").date())),
    },
}

Show the number of days since an important event in each frame

This code assumes that each filename is something like IMG_20230104_174807.jpg.

from datetime import datetime

important_date = datetime(year=2023, month=1, day=1).date()

config = {
    "caption": {
        "enabled": True,
        "generator":
            (lambda filename: str((datetime.strptime(filename, "IMG_%Y%m%d_%H%M%S.jpg").date() - important_date).days)),
    },
}

Adding music

Put your music file in the directory that contains config.py. Then, update your configuration as below; replace music.mp3 with the name of your music file.

config = {
    "ffmpeg": {
        "custom_inputs": ["-i", "music.mp3"],
        "custom_output_options": ["-map", "0:v", "-map", "1:a", "-shortest"],
    },
}

Development instructions

If you are a developer and want to help with or change Facemation, these instructions are for you.

Requirements

All systems

• shape_predictor_5_face_landmarks.dat (extract and store in src/main/python/resources/)
• Roboto-Regular.ttf (extract and store in src/main/python/resources/)

Linux

• Python 3.10
  The commands in this README invoke Python as python. Use python3 instead if you have not linked python to python3.
• venv
  You can check if you have venv installed by running python -m venv; if you see usage information, you have venv. To install venv on Debian/Ubuntu, run apt install python3-venv.
• CMake (required to build dlib)
  On Debian/Ubuntu, install with apt install cmake.
• C++ compiler (required to build dlib)
  On Debian/Ubuntu, install with apt install g++.

Windows 10/11

• Always use PowerShell.
• Python 3.10
• CMake (required to build dlib)
• C++ compiler (required to build dlib)
  You will need either Visual Studio (an editor) or Visual Studio Tools (a library). You can find both on the Visual Studio downloads page.

Run script for development

Setup

1. Check that you satisfy the development requirements.
2. Create a venv:

   python -m venv venv/

3. Activate the venv:
   • Linux

     source venv/bin/activate

   • Windows PowerShell

     ./venv/Scripts/activate

4. Install dependencies:

   python -m pip install --upgrade pip wheel
   python -m pip install -r requirements.txt

5. (Optional) Create config_dev.py to override both config_default.py and config.py.

   cp src/main/resources/config_empty.py config_dev.py

   Note that config_dev.py is always searched for in the current working directory.

Usage

1. Activate the venv:
   • Linux

     source venv/bin/activate

   • Windows PowerShell

     ./venv/Scripts/activate

2. Run script:

   python src/main/python/facemation.py

Build executable for distribution

Requirements

• All development requirements listed above
• (Linux only) Requirements for staticx
• (Windows only) Always use PowerShell
• (Windows only) Windows SDK
  Copy all DLLs in C:/Program Files (x86)/Windows Kits/10/Redist/[version]/ucrt/x64 to src/python/resources/. Note that the [version] in the path differs per system. I don't know what the implications of this are.

Usage

1. Check that you satisfy the distribution requirements.
2. Check the version number in the version file.
3. Check that config_empty.py is up-to-date with config_default.py.
4. Build executable into dist/ and create .zip distribution:
   • Linux

     ./build_linux.sh

   • Windows PowerShell

     ./build_windows.ps1

5. Run executable:

   dist/facemation

Acknowledgements

In chronological order of contribution:

• Thanks to Luc Everse for finding a bunch of bugs in v1.0.0!

If I should add, remove, or change anything here, just open an issue or email me!