Matting film to make a star appear as twins

by John Cunningham 1998

Ever wondered how they make a scene where the same star plays two different characters, like Hayley Mills as both Sharon and Susan in THE PARENT TRAP (1961) or Bette Davis as both Margaret and Edith in DEAD RINGER (1964)? I did. So I asked John Cunningham and this was his answer:

This is done with a technique known most simply as matting (or matteing). It can be done in-camera (and frequently was done this way during the days of the silents), but most often was done in the lab (and nowadays is done with digital compositing in with computers).

Traditionally, during the days of the "classics" it was done as follows for a scene such as those in THE PARENT TRAP: (I'll simplify here and try to avoid technical jargon) --

The scene would be shot two times with the actor playing each role and the camera (usually) in the exact same position. A stand-in would play the "other" role for the actor who was "doubling." The scene would be laid out so that there would be some part of the background along which a "divide" could be easily made-- such as a straight edge of a wall, or a straight tree, etc. (although the divides didn't necessarily have to be straight, it did make things easier).

The dialogue was carefully timed by the script clerk so that the doubled characters would speak on the proper cues. Sometimes during the shooting of the second "twin" the taped dialog from the first "twin" could be played back on the set to help ensure proper timing of the dialogues and reactions. These reactions can later be helped along by good editing, of course.

In the compositing lab, there would then be two strips of film, each of which would contain one "twin." These would be combined ("composited") onto one a new single strip of film by means of the matte process. The equipment used for this is called an optical printer. Simply put, it is a device containing a projector on one end which projects directly into a camera on the other end. (This is an oversimplification, but I'm trying not to get too technical here.) Thus, a new strip of film can be made from previously shot material.

One of the previously photographed "twin" shots would be loaded into the projector side of the printer along with a "matte" which is simply a tailor made piece of film, part of which is clear and part of which is black. The black part of the matte will cover up (or "matte out") that portion of the scene with the stand-in, (i.e.. keep that portion of the frame from projecting into the camera head of the printer) and the clear part will allow the part of the shot we want to pass through to the new strip of film. This setup would then be run, thus exposing only half of our new strip of film (one "twin").

Then, the new film (still undeveloped) would be rewound, and the other "twin" footage would be added to the other side of the new filmstrip in much the same way. (A reverse matte would be created to matte out the stand-in in the second portion, and let the second "twin" portion of the previously shot footage pass through to the new strip of film).

Thus, our new strip of film will contain both "twins." It is a composite of the two different "sides" of the two previously shot scenes.

It was possible to move the camera during such a shot, but it made the matte process much harder because "traveling mattes" had to be used which would track the divide in the picture as the camera moved. Therefore, usually you will notice that shots showing both "twins" are usually static shots with no camera movement.

What really makes these shots work in the final film is, of course, the editing and sound. The over-the-shoulder shots (which show one "twin's" face with the back of the other "twin" in the foreground) were done normally, using a double for the "twin" whose back is showing. Thus, the complicated compositing process is kept to a minimum.

Needless to say the process also got complicated when one "twin" had to interact with another twin (like, for example, hand them something or shake hands). This called for much ingenuity on the part of the special effects departments, but a solution could usually be found.

For example, in David O. Selznick's THE PRISONER OF ZENDA (1937) with Ronald Coleman playing both Rudolph Rassendyll and King Rudolf V, I recall a scene in which the "twins" must shake hands (or something similar--this is from memory). They handled that one by actually compositing only the head and shoulders of one "twin" into the previously shot footage of the other twin shaking hands with a stand-in.

You really can't even see it on TV, but the effect is barely discernable on a projected print. (I happen to be lucky enough to have that film on 16mm.) Well, you might be able to detect the matte on a good TV if you looked very closely.

If you taped the series HOLLYWOOD off of AMC in 1997 (the series narrated by Ed Asner about RKO studios), there is one episode in which special effects wizard Linwood Dunn (who pioneered the modern use of matte effects) explains some of this using film clips from CITIZEN KANE (1941) and BRINGING UP BABY (1938). The scenes from BRINGING UP BABY with the leopard would have been done with the same technique as was used when compositing "twins"-- so you might find that helpful to look at. Some of these shots had camera movement and, thus, used the traveling matte process.

© 1998 John Cunningham

Return to the Tech Talk Index.