At the risk of being absurdly reductionist: there are two kinds of special effects. The first is the Invisible effect—the kind that you aren’t meant to notice.
Back in the days of classic film, these sorts of effects included matte paintings used in establishing shots, or rear projection effects. Of course, keen eyed observers probably did notice these effects from time to time, and modern eyes are even more attuned to spot them, but the point was that you weren’t really supposed to remark on these things—they were executed by the production team simply as a means of filling out the scenery and frame within which the important stuff happened. Nowadays there are even more sophisticated CGI techniques to perform similar functions. When film fans (like those of us congregated here) argue over the relative merits of CGI vs. old school practical effects, we’re generally arguing over the second kind of special effect, which I’ll get to in the next paragraph. The fact is, the vast majority of CGI effects and digital compositing pass you by without you even noticing, because they’re Invisible. Here’s a link to a surprising video montage of the extraordinary ubiquity of digital effects in the most ordinary of situations:
http://petapixel.com/2011/02/15/amazing-effects-from-popular-tv-shows/
Which brings us to the second category of special effects, which is the one most people think of when you say “special effects,” and that’s the Spectacle. Instead of techniques used to help dress the set and set the stage for the real drama, these are effects that are the drama in and of themselves.
A couple of examples help clarify the distinction: consider the 1922 Nosferatu. This film is full of trick effects—the vampire’s carriage shown in negative, or sped up motion; the vampire stuttering his way out of his coffin as if the camera can’t hope to capture him moving normally; the vampire vaporized by daylight… these are Spectacles, and rank among some of the most memorable scenes in this film, if not among the most memorable scenes in all of film. But F.W. Murnau is using trick effects for prosaic reasons as well—such as his reliance on day-for-night shooting because his film stock wasn’t up to the task of actually exposing an image in the real nighttime. You’re meant to marvel and gasp at the monster’s unholy emergence from the ship’s hold—you’re not meant to notice that this actually happens in broad daylight.
Or, let’s compare the 1954 Godzilla with the 2014 reboot. The Japanese original was for all intents and purposes directed by two different people—with Ishiro Honda directing the somber human drama, while Eiji Tsuburaya directed Godzilla destroying Tokyo. As good as Honda’s part of the film is, most people come for Tsuburaya’s Spectacle.
By contrast, the 2014 film keeps the human drama center stage throughout, and shoves the monster action into the background—the film repeatedly cuts away from the spectacle, or shows it only in incomplete views glimpsed imperfectly. The 1954 film makes Tsuburaya’s scenes the center of attention; the 2014 version refuses to let the special effects take over. The result is a profound difference in tone: the 2014 Godzilla insists that the scenes of monsters fighting may be catastrophic and epic, but is in no way special, and does not deserve the same attention as that given to the people who try to cope with the catastrophe.
There is however a curious edge case, a special effect that takes center stage and briefly does the “look at me!” dance that we associated with the Spectacle, but which is also firmly intended as set dressing to support the real drama as something separate. It’s the Schroedinger’s Cat of special effects—the Invisible Spectacle.
We find this hybrid thriving in movies where an actor plays his or her own double. You know the kind—pictures about evil twins or dopplegangers, which give their stars a chance to show off their acting talents by playing two different roles at the same time.
The key here is that the star is playing a doppleganger, such that the whole point of the exercise is to show off how they can differentiate two different roles while actually looking pretty much exactly the same in both parts. There are times when actors are cast in multiple roles in a film without a doppleganger aspect—consider the many faces of Alec Guinness in Kind Hearts and Coronets. He plays eight different roles in that film, but there’s no effort made to get any of them onscreen together.
Now consider Alec Guinness in The Scapegoat, a minor British thriller from 1959 based on a book by Daphne du Maurier. Here, Guinness plays a dissolute and depressed wanderer who accidentally stumbles into the path of his improbably unrelated exact double, a troubled roué with money problems. One Guinness swaps places with the other, and the bulk of the film is watching him attempt to impersonate a man he doesn’t know.
As such, the backbone of the drama comes from Guinness’ subtle performance and the delicious tension that arises from our growing awareness of the traps he keeps blithely setting for himself. In other words, an efficiently inexpensive production (from Michael Balcon, and director Robert Hamer). But early on, the film indulges in what must have been some pricey effects, expertly staged, to get “both” Guinnesses (Guinnessi?) on screen at the same time.
Prior to the digital compositing techniques mentioned above, to get the same actor in two places in the same frame meant the use of one of the following effects: 1) Back projection; 2) Traveling matte and optical printing; or 3) double exposure. Each approach had a downside.
With back projection, you could shoot one part of the scene, and then project that behind the actor while s/he performed the second part. But the obvious discrepancy between the grain, contrast, and focus of the projected image versus the “live” performer right next to it would make the effect easy to spot.
A similar problem plagues the traveling matte—even the very best optically printed shots involve a change in grain structure and contrast that call them out to the eye.
And the problem here is that, while the audience might be willing to forgive those degradations in quality for a back projected street behind a prop car so the actors can play their scene without actually having to drive, the split screen effect can’t get the same level of tolerance. The whole point of making a movie in which, say, Alec Guinness plays two people is to show off how convincing he is in both roles. The moment we’re aware of the artifice of it, and remember that he’s just one guy and this is just a movie, the whole thing collapses. And so the effect has to be truly Invisible to work.
Which leaves us with double exposure—the best way to make sure both Alecs have the same filmic qualities on both sides of the screen. But the technique involves masking one part of the screen for the first pass with Alec 1, and then rewinding the film and re-exposing it with the exact opposite mask and Alec 2. This is really tricky. If the mask isn’t perfect then a small portion of the screen where the divide exists will either end up underexposed (a tiny shimmering black line between the two sides) or double exposed (a brighter shimmering line between the two sides). The margin of error is the width of a single grain of film.
Since getting it right to that level of tolerance is virtually impossible in a pre-digital age, the join has to be hidden somehow—by accounting for the dividing line as an element in the frame so that any imprecision is concealed. A conventional approach would be to use an existing vertical line in the set (such as a door frame, or a supporting column) as the boundary. This serves to hide any gap or overlap in the mask, so that both sides of the frame appear cleanly without a visible join. But… it also means that the two characters get divided by some arbitrary vertical line in the set behind them which they can’t cross, and audiences start to notice that weird arbitrary restriction on their behavior just as much as they would notice the dividing line itself.
Which is why I am so impressed by the split screen effects in The Scapegoat—there aren’t many, but the ones there are, are done with a jaw-dropping precision by cinematographer Paul Beeson. He has clearly realized that a 1959 audience has seen enough of these split screens by now they know the score and are probably looking for the join. And sure enough, in every one of the split screen shots there are a number of prominent vertical lines stabbing through the frame, any one of them a good candidate for serving as the mask line.
But… in every scene, the Alecs reach across those vertical lines. In one shot, Alec 1 reaches a candle across the table to Alec 2, who recoils from the flame. They don’t touch, but both men seem to occupy an impossible space. We know on an intellectual level that they could not possibly have sat at that table at the same time. That discrepancy in time has been concealed from view, and in stitching two moments of time together in the same frame we get an impossible space. How do their hands get so close? There must be a join somewhere, but it must be a complicated jigsaw puzzle pieced shape—and all the vertical lines here are there as distractions, to keep us from recognizing the actual join. The rest of the movie has no need for so many straight lines—it’s a quirk of the set design that appears only in the split screen scenes early on, as a way of (let’s admit it) showing off.
And in showing off the effect, by demonstrating it to be invisible, The Scapegoat manages to present the rare Invisible Spectacle.