Three Endings

by Bruce Kawin

from Film Quarterly Fall 2011, Vol. 65, No. 1

When he saw the clouds moving around the mountain like a wheel and had the cameraman take several shots of them, Abel Gance must have been ecstatic. He may have been waiting for the shot for months, or it may have been a matter of good fortune that he was presented with a real-world image that related to the thematic core of his La Roue (The Wheel, 1922) and was fit to be part of the summing up of the whole long movie: a natural image of the wheel, defined at the end (quoting the intertitles) as an “eternal tragic dance” in which “everything turned … even the clouds,” and shown in an icon in the corner of many of the intertitles as a torture wheel on which a prisoner’s body is broken. Symbolically, the image of clouds around the mountain is related, however positively and transfiguratively, to the fateful, crushing wheel that is evoked throughout the movie; indexically, it is a film image taken in real time; and iconically, it is a recognizable picture of clouds and a mountain—those clouds, shot when they were present, intercut with intertitles and with shots of young people dancing in a circle up the mountain. In the terms posed by C. S. Peirce and examined in relation to the cinema by Peter Wollen, the shot of the clouds around the mountain is a complete cinematic sign, with iconic, indexical, and symbolic elements that are clearly expressed and plainly interdependent. And it is not an effects shot, which is one reason it compels and rewards belief. Although silent, it is a shot richly expressive of the semiotic range of the cinema.

No lucky weather was involved in catching the image of the “Toruk Makto” flying off into the sunset near the end of Avatar (2009): there was nothing to catch. There was no gigantic dragon in flight, and the sky itself wasn’t there. There are clouds in the shot, but we cannot tell whether they began as photographed clouds or were generated or reworked on a computer. (Would Cameron have been content with a real, untweaked sunset?) The star and the clouds, the sky itself, are not ours but are those of the fictional moon Pandora. And the Toruk Makto is a creature given form and movement by the computer. Complete with sound and color and in 3D, the shot is a representative digital composite. It is, of course, an effects shot. Iconically, it resembles the event it portrays; symbolically, it evokes Earthly images of a bat or bird flying across the moon, and the sunset and the creature’s journey away from the camera imply resolution; indexically, it is nowhere.


La Roue. © 2008 Film Preservation Associates Inc. DVD: Flicker Alley.

The hero who sometimes rides this airborne reptile was never on it and was created partly by motion-capture work that yielded a photorealistic figure, one who looked as if he had been photographed although he never existed physically in that form. The photorealistic shots of the extraterrestrials and their environment, with everything sharply shot or as if sharply shot, are easy to accept on their own terms, as parts of the story and its coherent world. Audiences have no trouble suspending disbelief in the world, characters, and events of Avatar because the effects are both photorealistic and seamless. As the Toruk Makto flies and we see the planet that Pandora orbits and the other moons—all that and the sunset create a skyscape that is utter science fiction and that appears to be present. Effects have always attempted to convince the audience that an event that could not be photographed in reality was somehow happening; all that is new about Avatar in this context is that it does what today’s audience considers a better job—that is to say, a more credible, convincing job, while the effects shots in older films look dated. (The definitive characteristic of the digital, electronic image is that it lends itself to manipulation as the film image does not. A film image can be modified—on an optical printer, for example—but that is far more difficult and “against the grain” than revision of a digital image.) Those who saw the Red Sea part in The Ten Commandments (1956) may have vaguely noticed the black matte line that separates the figures on the shore from the sea; they may even have recognized it as a matte line or as something that showed up in many spectacles and science-fiction films. But their primary attention would have been on the water, on the other side of the matte line, where the sea did appear to part, hold itself up, and then rush together. Today’s audience sees the matte line and may be pulled out of the illusion by it. But the sunset shot in Avatar lives up to current expectations, and most audiences are conscious less of a digital composite than of an extraterrestrial sky.

There are clouds in the sky, but they are not in that sky if they were ever anywhere at all, and in spite of the willing suspension of disbelief, we know it. We also know that the image was imagined and then made, not captured in a fortunate moment, and in this respect the shot is representative of another aspect of digital cinema: that its essential, originating space can be found on a monitor more often than in a viewfinder.

And everything turned. Even the clouds, those white shadows of the mountains which danced the eternal tragic dance of the Wheel around the peaks.


Avatar. © 2009 Twentieth Century Fox Film Corporation and Dune Entertainment III LLC. DVD: 20th Century Fox Home Entertainment.

It is a sky in which we believe only for the purposes of fiction. The shot in La Roue offers a skyscape in which we can and do believe not only as part of the fiction, but also as a real mountain and sky that existed in front of the camera long enough to be photographed. The Avatar sunset looks like a sunset and could even have been photographed, though probably not in 3D (at the end of the shot there is a small lens flare that might be an artifact from that shoot or could have been added to give a sense of photographic realism). But we do not take it for a real sunset. The fact that both Avatar and La Roue are fictions is not the point. The La Roue shot presents itself to be believed, while the Avatar shot invites and depends on the willing suspension of disbelief. Both shots look photorealistic, but only one was photographed. The digital image, whether it presents a fantasy world or an everyday one, can become as convincing as photography and persuade us of the existence of its world, even on the as-if basis of fiction. But, if only because we cannot be sure to what extent it has been computer-manipulated, it will always to some degree lack the often exciting groundedness of filmed reality.

Even if a digital shot is entirely plausible, it may still strike the audience differently from a filmed shot. Part of the reason for this is that we know the digital shot or some of its details could be phony. There is an authority of reality behind the La Roue shot, taken on the real Mont Blanc—as in the “hypothetical experiment” cited by Siegfried Kracauer in Theory of Film (New York: Oxford University Press, 1960): “Blaise Cendrars … imagines two film scenes which are completely identical except for the fact that one has been shot on Mont Blanc (the highest mountain of Europe) while the other was staged in a studio. His contention is that the former has a quality not found in the latter” (35). Of course the two Mont Blanc shots cannot have different qualities from each other because they are both hypothetical. But the shots we imagine can be distinguished by the assumptions we bring to them. If we know that the real Mont Blanc was used for the first imaginary shot, we can read into it that it is more grounded and trustworthy, and we can consider or perceive the second more as an excellent artifice, even while we are considering both shots to be visually identical. But this is a matter of reading in and of imagining a valid, even obvious but invisible difference. In any case, if we now look back at images from classical cinema, whether shot on location or in a studio, we know at least that they cannot be CGI. A digital effects shot, by contrast, can be wonderfully plausible at the same time as it abandons the authority of reality. The folding of the street in Inception (2010) can be such a persuasive if fantastic experience because it looks just as it might if the event had happened in front of a camera. The only indication that it was an effect came from its physical impossibility, from our knowing it couldn’t happen except in the dream terms of that movie. This may increase our willing suspension of disbelief—after all, we just saw the street fold over—but what we see is still fake, lacking in reality. In a digital film, our knowledge that anything can have been modified may keep many of us, especially those who first knew cinema before the advent of CGI and digital-image processing, from willingly suspending disbelief in the fiction.


Although I enjoy being carried along and convinced by great digital images, I will miss shots like the one from La Roue. I will also miss battle scenes with people rather than pixels in the extreme long shots. Developments in movie technology increasingly mean that my willing suspension of disbelief will no longer involve a complementary trust in the reality that faced the camera (which is, for example, one aspect of enjoying Buster Keaton’s stunts). Digital cinema’s relationship to reality is always hypothetical to some degree, uncertain—even when, in the finest high-definition work, there is as much visual detail about a physical, photographed subject as one could want. It is no secret that a Blu-ray disc can give a better image than a 16mm print and that a 4K digital master can look as good as a 35mm print, yet every one of the numerous pixels that constitute the digital image’s stunning detail may have been altered.

Gance’s 35mm image is rich, and I fall in readily with its representation of reality. I accept it as a true record of a few seconds in the 1920s and as a fictitious element within the story of La Roue. I do not see any evidence of a miniature mountain or a cloud chamber. Even if it were an image of a different mountain from where the action was set, I would still recognize a real mountain with clouds around it, knowing at the very least that the history of movie technology did not permit a fake mountain to substitute plausibly for a real one.

A shot like the one near the end of La Roue still could be taken, though no doubt many digital filmmakers might actually prefer what they can achieve on a computer. Such an image might look very similar to Gance’s, it might even be nearly as compelling, but it would not have the same automatic claim on belief (which may be just a potential for belief, grounds for it). To make too much of this development is perhaps to indulge in nostalgia for the indexicality of the photographed image. But I doubt that I am the only viewer who will continue to suffer a crisis of disbelief when it comes to images that I know can have been modified in any way and still look unmodified.

The time of the pre-digital image is ending, thanks to a major shift in technology of which both filmmakers and audiences are entirely aware. Even if it had its share of effects shots and fakery, of make-up and lighting, and generated worlds that were often paradoxically both real and artificial, pre-digital cinema could shoot sheer reality if it wanted to and would find an audience willing to embrace the shot. The audience that wants digital photorealism now will find it most readily not in a narrative film but in a documentary, such as Werner Herzog’s Cave of Forgotten Dreams (2010), where the material is authentic and the digital 3D enhances a beautiful kind of realism whose gradual passing I lament.

Bruce Kawin is Professor of English and Film at the University of Colorado at Boulder. His book of poems, Love If We Can Stand It, will be published by Thames River Press.