Background Research on Lev Manovich’s Soft Cinema

At the end of my MA, one of the avenues I wanted to explore for further work was to use algorithm to relinquish control on the editing of a piece of video art.

Last month I took part in onedotzero cascade and my team decided to explore narrative in a visual way (I’ll blog more about cascade and this project in a subsequent post). Because Lev Manovich’s Soft Cinema was relevant to our project, I did a bit of background research on it to present to my team.

http://en.wikipedia.org/wiki/Database_cinema

Manovich opposes narative and database in his theoretical writings. However, he cites Peter Greenaway as an example of database cinema (soft cinema). Yet, Greenaway does not do away with narrative, rather he experiments with non-linear forms of narrative, in the tradition of modernist literature.

http://www.softcinema.net/?reload

Soft Cinema as a software edits movies in real time by selecting multimedia elements from a database based on rules defined by the authors.

http://www.softcinema.net/form.htm

SOFT CINEMA explores 4 ideas:

1. “Algorithmic Cinema.”
Using a script and a system of rules defined by the authors, the software controls the screen layout, the number of windows and their content. The authors can choose to exercise minimal control leaving most choices to the software; alternatively they can specify exactly what the viewer will see in a particular moment in time. Regardless, since the actual editing is performed in real time by the program, the movies can run infinitely without ever exactly repeating the same edits.

2. “Macro-cinema.” If a computer user employs windows of different proportions and sizes, why not adopt the similar aesthetics for cinema?

3. “Multimedia cinema.” In Soft Cinema, video is used as only one type of representation among others: 2D animation, motion graphics, 3D scenes, diagrams, maps, etc.

4. “Database Cinema.” The media elements are selected from a large database to construct a potentially unlimited number of different narrative films, or different versions of the same film. We also approach database as a new representational form in its own right. Accordingly, we investigate different ways to visualise Soft Cinema databases.

I’m not too interested in ‘2 – Macros cinema’ because I like the traditional aesthetics of one frame.

I’m interested in ‘3 – multimedia cinema’ because it offers the possibility to mix video clips with sound and/or music from a different source, or superimpose a still image on a video, but not in the way Manovich does it: ‘While some music videos and artist videos already mix some of these different types of imagery in one work, Soft Cinema assigns each type of imagery to a separate window in order to dramatize the new status of “normal” video, photographic and film image today – no longer the dominant but just one source of visual information about reality among many others.’ Apparently, Soft cinema keeps the different media in separate windows, therefore not creating a real multimedia final product. I would rather find a way to mix the contents together like it is done in some video clips.

What I’m really interested in is ‘1 – algorithmic cinema’ (4 – database cinema cannot in my opinion really be considered as a separate concept, because the database is only the collection of multimedia raw material the algorithm will choose from. The database itself does nothing but hold a collection of content.)

The Soft cinema website explains how ‘algorithmic editing of media materials’ works:
‘Each video clip used in Soft Cinema is assigned certain keywords that describe both the “content” of a clip (geographical location, presence of people in the scene, etc.), and to its “formal” properties (i.e., dominant color, dominant line orientation, contrast, camera movement). Some of the keywords are automatically generated by an image-processing software (written in VideoScript), while others are input by hand. The program (written in LINGO) assembles the video track by selecting clips one after another using a system of rules (i.e. an algorithm). Different systems of rules are possible. For instance, one system selects clips closest in color, or type of motion to a previous one; another matches the previous clip in content and partially in color, replacing only every other clip to create a kind of parallel montage sequence, and on and on.

The current version of Soft Cinema software allows the author to define such a particular system of rules, which it then uses to compile a sequence of video clips that best satisfy these rules. However, it is also possible to create other versions of the software that would give the author tighter control over the sequencing. For instance, one version may involve a video track completely edited by the author beforehand. Some shots could be designated as “replaceable” while others would remain unmodified (to keep narrative continuity.) Another version may contain a variable set by the author, which tells the program the probability of any shot being replaced. In summary, instead of posing complete randomness against the complete control of a human author, Soft Cinema investigates a different paradigm: using a computer as an “association machine” that complements / reacts to images selected by the user with other images.’