63
many series of ‘memory pictures’ described by Einstein. (Houltn On Scientic inking p. 380)
is in turn recalls Pryluck and Arnheim on the imbedded meaning of moving image documents,
in part, by the sequencing of images "ambiguous precisely as a consequence of their specicity.’"
Pryluck (1976, ) and Arnheim (1974, p. 246)
We look at our model as something of a topographic map showing the terrain of a lmic
document—here meaning a small number of frames, an entire production, a collection of produc-
tions, and all the derivative lms such as trailers, reviews, and fan re-edits. Such a map may serve
as a guide to a particular lm, as a picture of the lay of the land of similar lmic documents, or as
an orienteering or exploring map suggesting possible paths and features but no particular target
or goal.
We share Mamber’s notion of providing spatial and visual access points—indeed O’Connor
and Augst produced a digital interface to e Birds not dissimilar from Mamber’s. Both interfaces
have the appearance of inuence from Bellour’s analysis of key frames in the Bodega Bay sequence
from the lm. e key frames were "eyeballed" so the work was tedious and, more importantly,
subject to aws of human interpretation. Bellour and Mamber both rely on “the shot” as the unit
of measure and the unit of meaning. is is not a fundamental problem so long as one is dealing
only with classic Hollywood narrative lms; however, once one steps out of that realm changes in
the data stream can occur without well-dened markers of change. Even within Hitchcock’s e
Birds the entry of the sea gull that strikes the heroine is not marked for change at the frame when
the gull actually appears or the frame in which it actually exits. Rorvig noted that NASA was once
wrestling with the problem of how to analyze long-take data streams of space exploration with as
many as two dozen cameras running continuously how many eyes would it take to look at hours
of almost the same scene watching for the rst sign of a crack in a hull or (as he joked) the three
frames with little green aliens waving (Rorvig, Personal Communication, 1999).
ere is a fundamental problem with the “shot”—the denition is “endlessly bifurcated” in
Bonitzer’s terms. ere is no universally applicable standard of just how long a shot is, what denes
its beginning and end, or how much should be contained within a shot. is is why we look at the
data stream and its changes over time at the pixel level. is is where analysis of viewer reactions to
stimuli and providing tools for nding and understanding overlap.
3.4 PROVOCATIONS
At the time of O’Connor’s initial research in the early 1980s there were no print catalogs of doc-
umentary or educational lms that had even one frame from each and every lm in the catalog.
Today, images derived from videos abound in web-based interfaces to collections of lms. Here we
have just two samples to raise the issue of "just where to jump." e issue is nowhere near so vexing
3.4 PROVOCATIONS