With this application, you will be able to:

• Choose a list of photos featuring the faces you want to morph.
• Automatically insert a triangulation with 68 recognized points for each face.
• Adjust the triangulation to fit each face accurately.
• Create a morphing video that transitions smoothly from one face to another

Main features:

• Multi-language
• Configurable multi-resolution zoom
• Dark mode option
• New version notification
• User manual

High-level algorithm:

• Beginning with a triangulation that remains consistent across all images
• However, ensure that the vertices of the triangles are adjusted to fit the unique features of each face
• The morphing effect is created by transitioning from the triangulation of one image to that of the next while simultaneously fading the images of each triangle

Delaunay triangulation:

• It utilizes a professional library for incremental Delaunay triangulation, which I translated into Java from a C++ implementation I found online based on an algorithm developed by researchers in the field.
• Added feature: option to undo the last vertex insertion, allowing users to have a tentative vertex on the fly until they formalize it in the triangulation (demo).
• Added bonus feature: define the initial domain as a rectangle instead of a triangle, which is not trivial, especially considering my background!
• Added function that allows for traversing the triangles in a triangulation using the Visitor design pattern.

Functions that are worth noting:

• Basic algebra: Transformation of the coordinates of a point inside a triangle to the coordinates inside another triangle, knowing the vertices of both triangles.
• Basic mathematics: Use the Visitor design pattern to traverse the pixels in a triangle.
• Option to create the video sequence using multi-threading.

With the release of version v1.1, creating triangulation has become much easier for users by integrating a face detection library powered by AI. Users can access the library via a REST API to a cloud server.

With this library, users can begin with triangulating the 68 detected vertices for each face and continue adding vertices if necessary.