February 03, 2008

Cuts in movies and their impact on memory

When we watch a movie, we're usually not conscious of the cuts made by the editor. The camera angle may change dozens of times during a scene, and we follow along as if the flashing from one viewpoint to another wasn't at all unusual. You might think this is just because we've been accustomed to watching TV and movies, but researchers have found that even people who've never seen a motion picture have no difficulty following along with the cuts and different camera angles in a video.But little research has actually been done on the impact of changing camera angles in a movie on our perception and memory of a scene. While cutting abruptly between camera angles seems unnatural, moving a camera from place to place while filming can be quite realistic: after all, people walk around all the time; their own viewpoint is constantly changing. One study did find that people have better memories for a static scene filmed with a moving camera, compared to two still shots taken from the beginning and end- points of the camera's motion.

But what about dynamic scenes? If the people in a scene are themselves moving, will an abrupt cut to a new camera angle disorient the viewer? Filmmakers have found anecdotally that a 180-degree shift in a cut can be extremely disorienting -- that's why when watching a football or basketball game we usually see the action from just one side of the field or court. But do smaller cuts have a similar impact?

A team led by Bärbel Garsoffky showed computer-generated ten-second movies of a half-court basketball game to 12 volunteers. In some of the movies, the camera maintained a steady position either at the side of the court or midcourt, looking straight at the hoop, like this:


Garsoffky1.gif

In some movies, the camera angle abruptly changed form sidecourt to midcourt (or vice versa) four seconds into the film. In others, the camera moved smoothly between the two positions in a two-second-long pan. After watching each movie, viewers saw 24 still images. Twelve of the images represented actual court configurations from the movie they had just watched, while twelve images depicted the same players, but in positions they had never occupied during the movie. Viewers indicated whether each still shot represented a part of the game they had just watched.

Some of the still shots used the camera angle the viewer had originally seen them from, but others were from different camera angles: 45°, 90°, or 135° offset. Regardless of the camera angle in the test, viewers were equally accurate at remembering whether they had seen that still shot. But the camera motion during the original movie did matter:


Garsoffky2.gif

There was no significant difference in the results for a static camera versus a moving camera, but viewers were significantly less accurate when they saw an abrupt cut in the movie. This decrease in accuracy was almost entirely found at the point in the movie immediately following the cut, suggesting quite strongly that the cut itself momentarily disoriented viewers. So although the perceptual system can handle cuts in a movie presentation, those cuts do have some cost.

I do wonder if the costs would be as evident in a longer scene. One reason movie editors like to make a lot of cuts is because it maintains visual interest. Perhaps at some point viewers would lose interest in a scene without cuts, and their memory for such a scene would actually be worse than a scene with cuts.

Article by Dave Munger
Via scienceblogs.com
Garsoffky, B., Huff, M., Schwan, S. (2007). Changing viewpoints during dynamic events. Perception, 36(3), 366-374. DOI: 10.1068/p5645

No comments: