directed by: John Cassavetes


IMDB link:

Submitted by Charley Leary on 2006-10-29

Charley Leary's comment:
Number of figures in frame is measured.
Note: there are a few mirror shots (there I counted the real body plus his reflection), and in the scene in MoMA's sculpture garden, counted live bodies plus statues.

Number of shots:
MSL 1.7 2.7 4.1 4.7 5.5 6.1 8.3 9.3
StDev 4.2 3.6 12.5 13.3 12.8 5.5 10.7 12
Min 1.2 0.6 0.7 0.7 1.1 1.1 1.7 1.8
Max 11.3 21.9 107.7 91.6 81.4 32.5 41.1 65.7
CV 1.07 0.94 1.51 1.29 1.38 0.74 0.91 0.97

Step: Vertical resolution: Height:
Degree of the trendline: Moving average : Color code?

Users' comments:

Author: Yuri Tsivian Date: 2006-10-30

Interesting. In 1992 I did a similar count manually for Bauer's slow-cut 1914 Silent Witnesses and Kuleshov's fast-cut Engineer Prait's Project (1918). I published the results in my essay "Cutting and Framing in Bauer's and Kuleshov's Films," Kintop 1, 1992. A brief chapter in that essay entitled "Average Shot Density" (pp. 108-109) explains what I wanted to establish by counting characters in each frame. What I was after was to prove then was a correlation between closer framing (e.d. shot scale) and faster cutting. I proceeded from a simple logical assumption that films in which one-shots (= shots with a single character in the frame) outnumber two-three-etc-shots will be likely to be cut faster than films in which two-shots or group-shots prevail over one-shots. I even went as far as suggesting something like a quantitative borderline between tableau-based and cutting-based styles in early films: if Average Shot Density (ASD, i.e. the sum total of bodies that take part in all shots divided by the number of shots) is less than one this will mean that shot-reverse-shot is prevalent here. The counting yielded the following results: the slow Bauer movie had an ASL of 50 seconds and ASD of 3.2 characters per shot whereas the fast-cut Kuleshov movie had an ASL of 6 seconds and ASD of 1.6 characters per shot.

I thought I could rest my case there, but Charley reopened it today adding Cassavetes' Shadows to my old evidence. For interest's sake, I did a series of simple multiplying/adding/dividing operations with Charley's data (assuming, by rule of thumb, that his 6+ category amount to, say, 10 bodies per shot, which can be grossly off the mark, but this is just an excercise on my part). The ASD I thus obtained for Shadows is of 2.46 bodies per shot. Going by my old logic, its ASL here should be slower that it actually is. But then, it's the US, it has a larger cast of characters, its editing is jazzy, plus there are so many factors which we don't yet know. Patience, man, not everything at once.

What one can also tell from Charley's data is that there is a tangible if not too stable correlation between the number of bodies in the shot and an ASL that the shots whcih belong to this category show. Take a look. No need to explain why, but this correlation dovetails with the Scale/ASL correlation observed in some films submited earlier on. We are on the right trail. Some more time and patience, and more editing secrets will be cracked.

Author: Charley Leary Date: 2006-10-30

I need to read Yuri's essay before making a decent reply, but a few initial notes...

First, regarding the logical correlation between shot scale and ASD, as Yuri notes one can assume the less people, the smaller the scale, and thus - as per the trend in the shot scale measurements in the database thus far - one can expect faster ASL for smaller scale, and SHADOWS, as per above cinemetrics, does conform to this trend. However...I would like to note that many two- and three-shots (even a few four-shots) are, in SHADOWS, shot in close-up, or medium close-up, with faces cut-off by the frame; i.e. in this film, two, not to mention three, is definitely a crowd for the frame.

The film has some exterior scenes (indeed has more than some other Cassavetes films) and I am wondering if I should focus/measure just the interiors (primarily shot in a studio [acting workshop] or apartments) - particularly to allow a better comparison alongside Bauer and Kuleshov? The exteriors included a number of the 6+, with a character disappearing into a crowded NYC street...or the opportunity for enough space to include 6+...

One other question. Part of my objective was to demonstrate how figures dominate the frame in this film. There are many shots in which figures move in and out, new characters enter and exit. On the other hand, a few camera movements pan from one group (usually two) to another (often of same number) -- so, do I count this as 2 or 4, and if 2, is this combination of camera movement and staging constitute something akin to a cut -- or, if not editing per se, at least some definable unit? I am thinking along the lines of Bordwell, in THE WAY HOLLYWOOD TELLS IT, how a rack-focus from something in the foreground to background effects the sensation of a cut.

Author: Matt Hauske Date: 2006-10-31

Along the lines of Charley's question regarding combination of camera movement and staging amounting to something like a cut, I noticed a similar shot in THE LADY FROM SHANGHAI when I watched it without measuring it. Grisby sits on a chair in the foreground with O'Hara in the background; the camera follows Grisby as he gets up and walks past O'Hara to a wall safe; at the wall safe, only Grisby is visible, but O'Hara's voice is heard offscreen. I had to rewind this scene to verify that it was all one shot because after the fact of the camera movement, I was not sure. In other words, as a casual viewer, the combination of camera movement and character restaging did have an effect on me that might begin to be categorized as something like a cut (see how noncommittal I'm being?).

Certainly if one were to analyze this scene closely, in say a paper or conference, one would divide the shot into three sections and have three different images to illustrate it (A, B, and C). But how does this play into the measurement of a film?

Author: Charley Leary Date: 2006-11-01

Also there is the wipe-by cut, discussed by Bordwell in the aforementioned book, with something, or someone, passing by the foreground (blocking the camera, in effect) used to disguise/mask a cut (see his example with JAWS). But a wipe I think can offer the semblance of a cut in itself, and Cassavetes' early films (SHADOWS, TOO LATE BLUES, FACES) seem to use this effect a lot, particularly FACES, evoking a transition from one image to another, but not masking an actual cut.

Counting the figures in SHADOWS, I was thinking of another alternate measurement in a scene I submitted for A KILLING OF A CHINESE BOOKIE, trying to measure how the scene is broken up by an actor's "beats" (change of mood, attention, movement/gesture, ?bits of business? etc.) - and the term "beat" would invite a rhythmic notation (although I understand that this might not have been the best translation of Stanislavsky). Anyway, as you can see from my comment on that submission, I certainly remain noncommittal and need to be patient.

I am looking for a way to examine the dynamics within a particular shot, as one frequent criticism of Cassavetes' films (at the time of their release) was, simply, that they were boring and tiring, criticism that went hand in hand with his reputation for long takes and smaller scale with little establishing shots. However, thus far, actual measurements with Cinemetrics has revealed that Cassavetes' films are not as slow (in terms of ASL that is) as they might be reputed to be.

By the way, measuring the ASD in SHADOWS made me notice how similiar the film is to Cassavetes' later HUSBANDS: both feature three guys on the town, each of whom has a duty to pick up a female companion, thus the count runs from 3 to 6, with 2 and 4 thrown in. HUSBANDS stages the breakdown among the three friends most effectively, as the story changes from one of 3 guys to 3 stories, ending with one left out of the group...

One more note, to Matt: the editing of such a pick-up scene in SHADOWS has been compared with the editing of the "breakfast montage" scene of CITIZEN KANE in an article by Adrian Martin, "John Cassavetes: Inventor of Forms" SENSES OF CINEMA no. 16, 2001 (skim to section on SHADOWS).

Author: Yuri Tsivian Date: 2006-11-01

Wipe cuts are Kurosawa's signiture transitions. Deleuse writes about them and Mitsuhiro Shinoda, too.

Add your own comment:
Your name:

Enter code from above picture: