15-862 Computational Photography

Assignment 2: Image morphing and modeling

- Charudatta Phatak

cphatak@andrew.cmu.edu

Section 1: Morph Video

The delaunay triangulation was used for creating mesh between the correspondence points between the two images. These are some of the frames from my image morph into cpratt. Here is a link to the gif image of the morph. Somehow the gif image was stacked the reverse way. So the morph goes backward from cpratt to myself. Morph Video.

Figure 1 Frames from the morph video

The morph could have been more improved if more correspondence points were chosen to include the neck, shirt region etc.

Section 2: Mean Face

The mean face of the students of class 15463 was computed. The mean face is shown in figure 2. This does not look really good because of small amount of image data set.

Figure 2 Average Face

My face was then warped into the geometry of the average face.

Figure 3 My face warped into average geometry

Average face was warped into my face geometry.

Figure 4 Avg. Face into my geometry

Section 3: Bells and Whistles

An average woman’s face and an average man’s face were computed using the image data set from the class. Using these data sets, some caricatures were produced. Also extrapolating the warping factors produced some more caricatures.

Figure 5 More feminine face

Figure 6 More masculine face

The results for this would have been more better if the average woman’s face and the man’s face were sampled with more data!

Figure 7 Caricature...Bottle Face!

Figure 8 Caricature. Squinty Eyed?!