lang22 / face-morphing

Face Morphing

This is a tool which creates a morphing effect. It takes two facial images as input and returns a video morphing from the first image to the second.

---

Usage

Clone this repository and run setup.py and requirements.txt. Also install below mentioned dependencies. Then pass the following command-line arguments to __init__.py in morphing folder.

python3 __init__.py face/body path_to_first_image path_to_second_image duration_of_video frame_rate path_where_video_is_to_be_saved

• Note that presently only face is allowed as first argument.
• Also instead of the path to image file, Python File Object of that image can also be passed in as an argument.

Examples

Using the below command you can generate a video depicting morphing.

python3 __init__.py face ben.jpg morgan.jpg 5 25 morphing-example.mp4

How It Works

1. Find point-to-point correspondences between the two images using Dlib's Facial Landmark Detection.
2. Find the Delaunay Triangulation for the average of these points.
3. Using these corresponding triangles in both initial and final images, perform Warping and Alpha Blending and obtain intermediate images to be used in creating videos.
4. Use ffmpeg to return a video from above created frames.

Requirements

These are the requirements apart from what is mentioned in requirements.txt :

• FFMPEG
• X11 (Xquartz on macOS)
• Boost and Boost-Python
• Dlib

For macOS : https://www.learnopencv.com/install-dlib-on-macos/
For Linux : https://www.learnopencv.com/install-dlib-on-ubuntu/
For Windows : https://www.learnopencv.com/install-opencv-3-and-dlib-on-windows-python-only/

Citations

Most of the help for this project came from this post http://www.learnopencv.com/face-morph-using-opencv-cpp-python/