by Davide Modolo
The presentation aims to explain what is Image Morphing and what is it used for. It also explains how our notebook works.
Since functions we use require two images of the same size, we wrote a library that edit one or both of them with a minimum (to none) information loss.
We started explaining the easier approach to image morphing: the Fading Effect. It requires two Pillow Images
Function to compute the affine transformation steps. It requires two images as numpy array.
Function to pick by hand three points in each image in order to perform the morph. It requires two images as numpy array.
Look for 3 high-similarity points and morph on them. It requires two images as numpy array.
Using DLIB library and shape_predictor_68_face_landmarks.dat to recognise faces in images and morph. It requires two images as numpy array.
It requires shape_predictor_68_face_landmarks.dat file taken from here
Faces are taken from This Person Does Not Exist
Since DLIB can easly bring errors, this environment ensures everything works.