CS194-26 Project 3 Vincent Zhu
Part 1: Image Morphing
I followed the established pattern of:
- defining keypoints and correspondences between two images
- establishing a set of midpoints through linear interpolation
- creating a Delaunay triangulation on said midpoints
- solving for a series of affine transformations between corresponding triangles on each image and the midpoints
- looping over each triangle and using inverse warping in conjunction with cross dissolving to find the pixel intensity value for the resulting midway (morphed) image.
Part 2: Mean Image
Average face of "Danes" dataset: (parsing ASF code taken from Piazza, thank you Sean Farhat!)
some examples from the dataset morphed into the average shape:
my face morphed into the average geometry, and vice versa:
Caricatures: the average image warped into me, done with a warp fraction of 1.3
Bells and whistles:
I morphed myself into the average French female face. The last three pictures, in order, are just shape, just appearance, and both shape and appearance respectively.