Emotional Response to Computer Generated Special Effects:Realism Revisited
The art of visual effects in motion pictures is an art of illusion. For over 100 years, film audiences have experienced cinematic illusions, some more believable than others. When a film grosses millions of dollars during its first week of national release, it is likely that it has had a large pre-release budget, that it has opened in a large number of theaters, and is entertaining, perhaps boasting the latest in computer-generated special effects. Before the 1990s, motion picture special effects were created by photographic process, choreographed before the camera during the production phase of the film. Today, the computer-generated special effect flourishes in modern motion picture production, particularly in the horror and science fiction film genres, as an alternative to filmed special effects.
Computer-generated special effects have become more technically mature, resulting in their greater use by filmmakers, and film spectators have given them a positive reception (see Morse, 1995). It has been argued that as the technology improves, its emotional impact on the viewer will increase, resulting in a greater emotional connection to the motion picture (Weiss, Imrich, & Wilson, 1993), along with increased believability of the filmed image (Anzovin, 1993; Rayl, 1990). Some investigators have even proposed that the film viewer may soon be unable to distinguish between filmed and computer-generated images (Anzovin, 1993, Rayl, 1990).
Although audience reaction to filmic special effects has been studied (Hill, 1998, Hoffner, 1995, Johnston, 1995, Zillman & Gibson, 1996), little is known about audience response to computer-generated special effects. Some obvious questions arise: Do viewers perceive computer-generated special effects to be as realistic as filmed effects? Do viewers respond to computer-generated special effects with the same emotional intensity as filmed special effects? Is the degree
of realism of special effects related to the viewer’s emotional response? In other words, to what extent does the realism of special effects drive the emotional intensity of the viewer’s response? For a given set of images, can viewers distinguish between filmed images and computer-generated images, and can they distinguish between unstaged filmed images and staged film images?
In the provocative, but largely unscientific literature of film studies the issue of the importance of realism in motion pictures has been hotly debated. The French film journalist and theorist Andre Bazin (1973) eloquently argued that realism brings the viewer into a closer relationship with the world of the film, that it brings the viewer into a relationship more like the relationship the viewer enjoys with reality itself. In her writings on photography, Susan Sontag lays out the difficulty: She states "Photographs furnish evidence. Something we hear about, but doubt, seems proven when we are shown a photograph of it. In one version of its utility, the camera happened. The picture may distort; but there is always a presumption that something exists, or did exist, which is like what is in the picture.” Extending these notions a bit, one might expect that if viewers perceive a filmed effect as more realistic than a computer generated effect, they might also be expected to find it more emotionally intense. I decided to test this notion in the form of the following hypotheses:
(H1) The degree of realism perceived by the viewer will be greater with noncomputer-generated special effects (live film footage) than with computer-generated special effects.
(H2) Emotional response to exposure to graphic violence will be greater with noncomputer- generated special effects (live film footage) than with computer-generated special effects.
(H3) There is a positive relationship between the viewer’s perceived realism and the viewer’s experienced emotional intensity when watching film footage of graphic violence.
Participants were 65 undergraduate students in communication at Georgia State University (female=34; male=31), ranging from 19 to 51 years of age, with a mean age of 23 years. The Perception Analyzer, a new device designed to measure response in film viewers, was used. The Perception Analyzer consists of a computer linked to wireless control modules to be managed by participants in a study. Each participant was given a wireless control module with a dial set at the
midway point and instructed that 0 (far left) was lowest and 100 (far right) was highest. Participants were given five warm-up questions with which to practice using the device, and proficiency was confirmed by the administrators.
Two videotapes of film clips containing graphic violence were prepared, each tape containing eight 30 second clips in three catagories, computer-generated special effects (C); filmic special effects (F; live, staged); and, documentary footage (D; live, unstaged), for a total of 24 film clips per tape. Both tapes contained the same film clips, but in a different order to minimize order effect. The first videotape (Videotape 1) was used to obtain the participants’ ratings of emotional intensity to the clips, and the second (Videotape 2) to record the extent to which the participants perceived each clip as real. The participants viewed Videotape 1, and during the viewing , reported on emotional intensity of each film clip by turning the dial. The participants did the same with Videotape 2, but this time, reported their perception of the realism of the film clips. After viewing each film clip, the participants’ dials were returned to midway position. The readings from the Perception Analyzer modules were collectively recorded on computer for analysis.
The mean scores for each type of footage, computer-generated special effects (C ), filmic special effects (F; live, staged), and, documentary footage (D; live, unstaged), support the first hypothesis, that viewers perceived the film clips with noncomputer-generated special effects (F and D) as more real than film clips with computer-generated special effects (C) (see Table 1). The mean scores for each type of footage support the second hypothesis, that film clips with noncomputer-generated special effects (F and D) would receive higher ratings of emotional intensity than film clips with computer-generated special effects (C). Preliminary analyses of the data indicate support for the third hypothesis: There is a significant positive relationship, in the
expected direction, between degree of perceived realism and emotional intensity. The greater the perceived realism, the greater the emotional intensity rating.
Looking more closely at the data one sees that the realism means for the "D" (unstaged film) category are higher than those for the "F" (staged film) category, and that the intensity means are the reverse (Table 2). The indications are that:
1) Participants were generally able to distinguish the unstaged film from the staged film, and to distinguish both from computer generated footage. They rated the unstaged as more realistic than the staged, and they rated both the staged and unstaged film as more realistic than the computer generated sequences.
2) In terms of intensity, viewers found the staged footage more intense than the unstaged even though they seemed to have known it was staged. And both staged and unstaged film sequences were judged to be more intense than the computer generated sequences.
3) There are a couple of anomalies in the first scenes in both the "D" and "F" groupings in Table 1, in that for these scenes the staged scene is rated as more realistic than the unstaged. These can perhaps be explained by the difference in scene construction. The unstaged scene, even though it was unstaged, was presented in a series of shots utilizing different camera angles and image sizes edited together; and the staged scene, even though staged, was presented in what appears to be one long take. This may be a clue as to the information participants used to sort the unstaged and staged sequences.
4) Overall, there is a positive correlation between intensity and realism, but this correlation is statistically significant in only 7 of the 26 sequences.
I have presented evidence that filmic special effects and documentary footage convey a greater degree of realism to the viewer than do computer-generated special effects, and that film
viewers are more likely to experience emotional intensity through filmic special effects and documentary footage than through computer-generated special effects. There is a legitimate question of which is the cause and which is the effect here. That is, do viewers react with more emotional intensity to scenes they see as more realistic, or do they attribute greater realism to scenes that provide them with more emotional intensity? One can't be sure. Note that while the realism and intensity seem to be positively correlated, many viewers in our sample rated the staged film scenes as more intense than the documentary unstaged scenes. We suspect that there are other variables at work here. What I may have inadvertently measured here is the effect of the filmmaker's presenting images moment to moment in a way that will maximize the impact on the viewer. The filmmaker presumably knows how to do this by having learned from example and by trial and error. Why should we be surprised that the makers of theatrical motion pictures are able to evoke a strong emotional reaction from viewers? This is, after all, what the makers of theatrical motion pictures are good at. If mere documentary footage consistently evoked such strong emotional responses, documentary films would fill the theaters. But they do not. Well crafted theatrical spectacles with an engaging storyline make up the list of mega-hits. Yet no matter how far fetched the premise or how far removed from earthly reality, no matter how bizarre or fanciful the subject matter, the makers of theatrical motion pictures strive to create a compelling appearance of reality. And as our data indicates the computer-generated effects just aren't quite there--yet.
What does seem clear from this research is that the viewer’s perception of the degree of realism of a scene is positively linked with her/his experience of emotional intensity. And because emotional intensity is an important, if not the primary, goal of graphically violent special effects, attention to realism seems to be key to enhancing any special effects, be they filmic, documentary, or computer-generated. Perhaps Andre Bazin was correct -- the realistic image may indeed bring us closer to that relationship that we enjoy with the real world.
Incorporating computer generated images into cinema also introduces a new spin on a long standing debate in film history, arguing the extent to which a motion picture camera can be viewed as a chronicler of events. If one subscribes to the philosophy that a degree of objectivity is inevitably lost at the point when one decides in which direction a camera should be aimed, then the use of computer generated imagery is little more than a step away from some ideal aspiration for objectivity and perhaps the ultimate demonstration that such an ideal can never be achieved. At the same time, if the technology can be used to create images which audiences cannot distinguish from live filmed images the question becomes, "to what degree does such a technique not only compromise the directors' art, but radically alter the viewers relationship with the film?”
S. (1991). Synthetic stars.
Compute, 13, 100 Omni,
Table 1. Mean Intensity and Realism Ratings of Film Clips and Correlations.
*p<.05 Number of participants=65
Table 2. Average Mean Intensity and Realism Ratings of Film Clips