directed by: Yasujiro Ozu


IMDB link:

Submitted by Matt Hauske on 2006-11-12

Matt Hauske's comment:
This is one of Ozu's last silent films. I'm measuring this in conjunction with FLOATING WEEDS, Ozu's sound remake from 1959, for a project in Michael Raine's class. I may end up re-measuring this film -- particularly towards the end I made errors in the measuring, particularly between image shots and dialogue shots. This is due mainly to fatigue on my part, but also to Ozu's tricky cutting. He definitely plays with the rhythms of classical style, cutting (or, more often, not cutting) when you least expect it.

Number of shots:
MSL 3.9 3
StDev 3.8 1.1
Min 0.8 1
Max 36.6 7.4
CV 0.74 0.36

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

Users' comments:

Author: Yuri Tsivian Date: 2006-11-14

Given the number of shots it makes sense to change step to 1 to get a full range picture.

The graph is consistent with that of Ozu's Dragnet Girl with its barely perceptible trendline slope on degree 1 and its minimal undulations at degree 6. I feel I should repeat here what I wrote under Charles O?Brien?s dynamic profile for Mutter Krausens. That this fast-cut movie (ASL of 4.6) also has a low (3.5) cutting swing does not surprise me. This seems to be the case when a filmmaker sees fast cutting as his/her conscious stylistic choice. And we know that Ozu studied the cutting times of American films, as did that German observer of 1926 whome Kristin Thompson quotes in her book (see Overview, click on Thompson). When fast cutting becomes an artistic goal in itself, rather that being subservient to story needs or acting low tolerance for longer takes is what we should expect. Reputedly, Ozu used a stopwatch on the set in his attempt to make Dragnet Girl as "American-fast" as he could. Take a look at Dragnet Girl's trendline, it barely moves when you change its degree, and its cutting swing is even lower that Jutzi's Mutter Krausens, only of 3 sec. Montage or bust!

Author: Matt Hauske Date: 2006-11-14

I haven't looked carefully at my results for this film yet, but I do have one quick note on the data: My results contradict those obtained by David Bordwell and listed in his book OZU AND THE POETICS OF CINEMA. In his Appendix on page 377, in which he lists the number of shots and average shot lengths of all Ozu's films (along with other pieces of data), Bordwell lists the number of shots, including titles, as 982 (I have 1084), and the ASL as 5.2 seconds (I have 4.6). Did Bordwell see a different version of the film? I used the one from the Criterion Collection, which is included in a set with Ozu's 1959 color/sound remake, which I plan on measuring tomorrow. It will be interesting to see how my data stack up against Bordwell's. This is not a competition, of course. But unfortunately, the technology of Cinemetrics is merciless!

(I should take this opportunity to admit that my measurement was not 100% accurate, either, but I know any extra clicks I may have made could not total over 100...98, at most!)

Author: Yuri Tsivian Date: 2006-11-14

Actually, estimating ASL as a mean value, as David did for the Ozu movies, is a more reliable method, for he only had to count shots which is easier hence more errorproof, than clicking on cuts, and know the length of the film hich is often given in filmographies. My cinemetrics timing of Dragnet Girl is only 0.2 seconds off Bordwell's, which is a fair match knowing how fast this film is, and how little experience I had with the tool when I timed it last year. In Matt's case it looks like two different versions were used, Michael Raine should know.

Author: Matt Hauske Date: 2006-11-21

According to Michael in a conversation I had with him last week, he didn't think any new versions of this film have been found since the publication of Bordwell's book, so the film version he saw and the DVD version I saw should be the same. In addition, the length of the film listed in Bordwell's book is the same as that listed on the DVD (86 minutes). About 2 minutes of credits at the beginning account for the discrepancy in the time recorded during this measurement--could that account for some of the ASL discrepancy? Does Bordwell count the credits when taking ASL values? Even if he did, I don't think that would fully account for the discrepancy.

Author: Yuri Tsivian Date: 2006-11-22

David's method is to exclude the credits. The discrepancy may be due to more precise tool we are using now in comparison to what we were using in the pre-computer pre-DVD era. I am tempted to re-time my old measurements of Bauer and Kuleshov, too (as a second though, why don't I tell Alyson Hrynek who is writing a dissertation on Kuleshov with me to do this?). This is the way of all thought: we correct and revise, and we move ahead.

Author: Matt Hauske Date: 2006-12-08

The longest shot is at the very end, when Kihachi and Otaka are riding off together on the train. Otaka serves him sake and food, and he reciprocates. Similarly, in the 1959 remake, the longest shot takes place at the end, and features Sumiko trying to light Komajuro's cigarette.

Add your own comment:
Your name:

Enter code from above picture: