One of my last projects in my digital imaging class through Duke’s visual studies department was titled, “Blurring Identities” and involved layering my face along with the faces of four of my classmates. The idea was to mimic the eigenface facial recognition technology.
Each member of the class was photographed against the same background under the same light conditions. I was given a set of five faces—my own, an Asian female, a black female and two Caucasian males (one of whom wears glasses).
I created three different images—one containing equal parts of all five faces, one made up of 50 percent of my face and 12.5 percent of the other four and a third mixture that was supposed to be the last point where you could still see my face as the dominant.
Naturally, the explanation makes more sense if you can see the finished products:
Each face was placed in its own layer in Adobe Photoshop, and the transparency was set to 20 percent. To start, I lined up all the eyes as closely as possible. At the end, I used an easer brush to clean up the excess hair.
For this image, the transparency for the layer with my face was set to 50 percent. For the other four layers, the transparency was set to 12.5 percent. I left a bit more of my hair this time and used a feathering effect to draw more attention to my face.
With so many faces competing for domanince, I found that I couldn't go lower than 44 percent and still be the principal face in the image. The other four faces were set to a transparency of 14 percent. I also got rid of the feathering effect from the previous image.