Earlier Work
The two film experiments were based in previously published hypermedia research. Stretchtext is, like hypertext, a concept invented by Nelson
. These ideas were later incorporated in Guide
, the first commercially available hypertext system, Stretchtext features are also incorporated in several adaptive hypertexts systems
.
All these examples use text as the main mode of representation. In addition, there are important forerunners in the area of hyperfilm:
- Kon-Tiki Interactive
- Automatic video summaries
- Hyper-Hitchkock
more...
Kon-Tiki
While many stretchtext systems were modeled on outliner systems in word processing applications, footnotes in books were the inspiration for Gunnar Liestøl's interactive Kon-Tiki projects
. In the Kon-Tiki Interactive
, six documentaries are available from a central menu. They may be watched as they are, but each of them includes three "footnotes," additional sequences that may be inserted smoothly into the story at specific points.
Automatic Abstraction
Works in automatic abstraction or summarizing of film material are also related to what is discussed here, such as the work of Michel Crampes, Jean Paul Veuillez, and Sylvie Ranwez
, or Frank Shipman, Andreas Girgensohn, and Lynn Wilcox
. Much of this work has an emphasis on automatic summaries of longer film material. My focus is different, as I concentrate more on how these different versions are presented to the user.
Hyper-Hitchcock
Shipman, Girgensohn, and Wilcox
have also addressed this issue in their work with the Hyper-Hitchcock system. Working on raw film material, the Hitchcock system sorts the good shots from the bad, using a wide range of image analysis techniques. From these good clips, it computes a suitable length of each clip, and generates three edits of different lengths from the material. The different versions can then be interlinked with hyperlinks. Within the Hitchcock project, the name for this stretchtext-like structure is detail-on-demand. An interesting detail in this system is that the same clips often will not be chosen for two different versions, although the overall sequence is the same. This is very different from my approach.
(collapse)
Two observations may be made from the study of earlier work.
(1) There are two major different interface styles. One is Nelson's approach, where the user selects the level of detail for the text as a whole, the other is the outliner approach, where the user may select long or short version at every point where alternatives exist.
more...
It is no surprise that the outliner style is far more widespread. Whether or not it is simpler for the user, it is at least easier to create.
For stretchtext to work, it is imperative that the user's location in the text sequence remains intact when the text stretches or shrinks. Nelson
seems to have imagined the text as a succession of screens with only a few lines of text on each screen. If a screen is almost full of text in the short version, however, a stretch will flow the text out of the borders of the screen. The line the user is reading might become invisible causing the user to loose orientation.
In an interface where the user selects the exact point to be stretched, he or she also at the same time shows the system where the current focus is, and loss of orientation is less of a problem.
This problem does not arise in hyperfilm, as the user's position in the text always will be the current point in time of the film. In hyperfilm systems such as the Adaptive abstraction system
or Hyper-Hitchcock
, the user may select a general length of the text, and the system will compute a summary, an approach related to Nelson's original vision.
(collapse)
(2) In hypertext literature, stretchtext is often described as a replacement mechanism. In a set overall sequence, the user may replace one node with another that is about the same, only longer (or shorter). (This principle is used for interesting narrative effects in the Fluid Reader system
).
A different concept is that of the footnotes used by Liestøl in Kon-Tiki
. The analogy is a footnote in print, where a visual symbol is inserted in the running text to indicate that some additional detail about what is discussed in the paragraph is found somewhere else in the text, if the reader wants to read it. In the video equivalent of this, all users see the shortest version of the film (the basic version). At certain points, a footnote marker appears for a period of time, allowing the user to let the system insert a short additional sequence at that point, before the basic sequence continues.
With these two observations in mind, my two experimental films tested Nelson's general altitude selection versus the outliner style of specific point selection, and replacement versus footnotes.