28 2. FIVE STORIES TO A MODEL OF VIDEO STRUCTURE
and … what? Perhaps remember those words or patterns, then go to a second performance and pick
up a stone for each occurrence. With several graduate assistants each dropping stones for a dierent
attribute, perhaps a structural analysis would emerge.
e post-Homeric scholar has access to a manipulable representation of the story—words
printed on paper or held in digital les. Words and their positions can be counted or graphed; a
segment near the beginning can be copied (or cut out) and set beside a segment near the end for
comparison; patterns in one text can be compared with those in another. e digital post-Homeric
scholar can re-represent the text in binary form and accomplish word counts and frequencies or
statistical measures of similarity in minutes
e lm scholar has followed a similar journey. e digital environment for moving images
is now suciently robust to allow digital translation of the enormous amounts of data in a video
segment. We can now return time to the representation palette. We can now bring book-like ca-
pabilities to lm scholarship. We can, with some thought, humility, and insight step beyond the
book-like engagement with lm and put dierent forms of evidence onto the same platform.
2.3.2 UNITS OF MEANING
e digital environment enables the lm theorist, the lm producer, and the lm student to make
use of a more rigorous and more widely applicable vocabulary of analysis. e diculties in repre-
senting moving image documents lie, in part, in the literary metaphor. ere is not in the moving
image document a sequence of minimal units of meaning subject to easy demarcation. Where the
physical verbal print document is a single ordered set of discrete and denable units (i.e., letters or
phonemes), the physical moving image document is comprised of multiple strands of data—light
values at dierent points on a plane surface, often together with multiple levels of sound.
Word-based descriptors of units of meaning as well as stylistic categories are of little help.
Terms such as shot, close up (CU) medium shot ( MS), long shot (LS), and documentary, which
are common in lm theory and in cataloging rules, lack precise denitions, are “endlessly bifur-
cated” (Bonitzer, 1977) and, thus, severely limit the powers for discrimination and analysis. Is a
CU anything that occupies more than 50% of the image area? 60%? 80%? Genre terms such as
“documentary” say very little about what to expect of a work’s content or structural characteristics.
How might we use the time line concept common to producers to construct precise and gen-
eralizable processes for abstracting moving image documents? How might we enhance the recogni-
tion and understanding … attendant upon structural consideration, so as to enable discrimination
within and among works on similar topics? While the utility of structural representations has long
been recognized by producers and artists, so have the limitations of structure charts developed for
personal use without replicable and transportable units and modes of representation. at is, charts
of structural relations have been used frequently by lm and video producers, but they have been
based on the production requirements and the working habits of the production crew. e elements
are not standardized and often deal with minutiae beyond the representation needs of most users.
Still, their long-term use in the eld suggests that they yield a robust model for representation. e
three-dimensional model uses a timeline element in the form of one of the three axes.
Timeline in Frame Increments
Figure 2.5: ree-dimensional, time-varying, pixel-based model of lm text.
Visualizing a timeline with each of the frames stacked along it presented a stream of data, as
illustrated in Figure 2.5. Modeling the moving image document as a bundle or stream of stimulus
values means that one can apply procedures for the measurement of signal variability.
e signal bundle model assumes a virtual, continuous signal or cluster of signals—strands in
a bundle—in depth. Each moving image document frame is a perpendicular slice across the signal
2.3 STORY THREE: KEY FRAMES