Oh, thank you, AGRBear. I enjoy history and reading the posts on this forum, but rarely post because I don't feel I know enough. It's nice to hear that maybe of bit of this geekspeak could be of use to historians.
In order to do a "fade- out" and "fade-in", don't you need two faces that are at the same angles?
You would for film and television, because of the format of the screens.
But in the NOVA documentary, the resolution or graininess of the images didn't matter (except for the identification), so the images were simply re-sized and rotated. The dissolve was zoomed in with the edges of both photos off the screen, so the audience doesn't see the rotation.
However, there is a distortion problem when rotating an image on the computer. Photographic images are made up of polygonal grains in a random pattern. Computer images are a grid system of square pixels. When you rotate an image, the software interpolates the position of the pixel to a new position, and that's what causes the distortion, which increases with every iteration. Every unsuccessful rotation must be followed by an Undo, or the image will be distorted. The distortion may not be visible until after a number of iterations, depending on the background and training of the viewer.
If all that is needed is to overlay the images, you don't need a computer. All that's needed is an art projector (sometimes called an opaque projector), or an overhead projector and a transparency. Move the projector until the image sizes agree, and rotate and align the transparency. You won't have the emotional impact of graphic dissolves and it won't be computer-generated, but the information is the same.
In order to rotate a face you'd, also, need the addition of 3D type of software? If so, would this take more than a enlarged "flatten" photo?
Creating a 3D model of a geometry as complicated as the human face is much more difficult than rotating a 2D image. We did have the software tools to do this in 1995, but re-creating a face from photographs is time-consuming and costly. In film, television, and games, the fastest and therefore cheapest way is to project a grid on an actor and photograph that to create the 3D geometry.
But you are right about the tilting of the head in the AA and FS photographs; AA's chin is slightly down. There will be a degree of foreshortening because of perspective, which is basically a mathematical description of photography. I don't know if the amount of foreshortening is enough to affect the apparent match shown in the dissolve.
