DVD Talk
Reviews & Columns
Reviews
DVD
TV on DVD
Blu-ray
International DVDs
Theatrical
Reviews by Studio
Video Games

Features
Collector Series DVDs
Easter Egg Database
Interviews
DVD Talk Radio
Feature Articles

Columns
Anime Talk
DVD Savant
HD Talk
Horror DVDs
Silent DVD

discussion forum
DVD Talk Forum

Resources
DVD Price Search
Customer Service #'s
RCE Info
Links

Columns



DVD SAVANT

Metropolis
and the
Frame Rate Issue


A note from Savant correspondent Filip Jensen prompts a brief discussion.

by Glenn Erickson

With all the buzz among film and disc fans about the new, nearly fully-recovered restoration of Fritz Lang's Metropolis, I'm also receiving notes asking about the new restoration's frame rate. You may recall my concern nine years ago when the 2001 Kino DVD of Metropolis was presented at 24 fps. I had just seen the German restoration experts exhibit their restoration at The L.A. County Museum of Art. They had chosen a slower film speed of (my guess at the time) 20 frames per second. At that frame rate the movement of the actors seemed more natural, and the film seemed appropriately "heavier". Rapid montages no longer seemed to flit by in a blur; Brigitte Helm's erotic dance didn't elicit laughs. Metropolis took on a new power.

Kino's 2002 DVD seemed a return to the old sped-up norm for Metropolis. I assumed that the projection had been standardized at 24 because Kino wanted to distribute the film theatrically, and normal theater projectors couldn't change their frame rate. Then Kino's experts explained that they were simply doing what was necessary to show the film with its original score, which was composed to match the film running at a speed of 25 or 26 frames per second. Metropolis was projected that fast at its premiere back in January of 1927. (A photo of the specially-built Berlin premiere venue is above).

For ten years now I've been writing about Metropolis using information gleaned from a number of sources, and not a year goes by when I'm puzzled to learn that something I was taught / read / assumed from common sense, just isn't true. Hearing about Ufa's financial implosion and the exit of Fritz Lang's producer/patron Erich Pommer, I'd always assumed that Metropolis was projected fast at its premiere because the company felt that the film was far too lengthy and was nervous about its investment. Accelerating the frame rate made a 3+ hour film finish more than 30 minutes sooner. The print I saw projected at the museum played a full 147 minutes. The same print on Kino's DVD has a duration of only 123 minutes. It makes a big difference.

I was so impressed with how well Metropolis played at 20 fps, that I got into one of my few web arguments over the issue, with readers, Kino people and a few other experts telling me that 24 fps was a legitimate frame rate and the one used at the 1927 premiere. My argument hung on the fact that the expert Murnau-Stiftung restorers themselves had chosen a slower frame rate for their authorized museum screenings. Even more to the point, I stressed the not-wholly-subjective observation that some of the film's action is ridiculous at 24fps. Maria bounces like a puppet on a too-fast cathedral bell, and our hero Freder zips through the streets like The Roadrunner. At sound speed, I also felt that Fritz Lang's quick-cut montages were just too quick -- an admittedly subjective judgment.

Now nine years later the newest restoration is here. It appears to be transferred at 24 fps as well, and clocks in at approximately 140 minutes. If I haven't complained about the frame rate this time, it's because I've seen Kino's excellent 2002 disc so many times that the speed now seems normal to me. The film needs to be running at that frame rate speed for Gottfried Huppertz' handsome original score to fit properly. Kino's 24 fps discs are historically correct.

I've received several more emails from readers hoping that a slower transfer will surface. For now the answer is No. Perhaps the German restorers have prepared one for internal use but we haven't heard of anything of this kind yet. I know that computer programs exist to convert 25 fps video transfers to 24 fps, and I wonder if such a program could be used to slow down Metropolis, even if the result had to be silent. The film has come out in so many versions that we have to realize that there are probably viewers out there who are going to prefer the older incoherent shorter copies, just for perversity's sake. Around my house, for instance, my kids grew up listening to the Giorgio Moroder version, disco vocals and all. Nostalgia-wise, that's the real Metropolis for them.

Last week I received this note from Filip Jensen of Copenhagen:


Dear Glenn, I read your older review of Metropolis and the frame rate issue. However, I have the right solution that should also convince the people at Kino to finally release this masterpiece at the correct frame rate. I cannot see any argument against it.

The answer to the question of correct frame rate lies in the movie itself. In several places we see a clock ticking, and even though the number of hours have changed from 12 to 10 to make working hours longer, Fritz Lang did not change the original clock behind it, thus it is very easy to find out at what speed the film was shot.

Using a recording of the 2010 restored version shown on Arte in February (at 25 fps as it is PAL), I focused on the scene with the clock in the beginning. I made a cut that shows the arm striking three seconds, and these three seconds make a total of exactly 57 frames. It does not take much math to figure out that the correct frames per second are 19, as 19 x 3 = 57!

So, there you have it after all these years. I hope this convinces the people at Kino to finally release a version where the film is shown at the original recorded frame rate. It is quite a different experience. Kind regards, Filip Jensen. Copenhagen, Denmark

I'm impressed by Mr. Jensen's ingenuity. I also hasten to agree with his first conclusion: to my standard of proof he's shown that the shot of the clock was almost certainly filmed at 19 or 20 frames per second. But that doesn't prove that all of Metropolis was filmed at 19 fps. Some scenes were purposely filmed at an even slower rate to yield accelerated motion, such as certain shots of Freder sprinting through the streets of the city. We're told that silent cameramen frequently varied the frame rate during filming to achieve various effects.  1

More importantly, Jensen's measurement doesn't take into account the fact that until 1928 or so, silent film projection speed was not standardized either. The fact that Fritz Lang filmed most of Metropolis at 20 fps doesn't necessarily mean that he intended it to be projected at that speed. Even if films went out with notes specifying a particular speed, exhibitors (and projectionists) could arbitrarily choose whatever frame rate they wanted. At sound speed, nobody notices that the clock's second hand is advancing too quickly. My 2002 mini-conniption fit over the Metropolis frame rate was basically an argument between (a) how it was projected at its premiere -- which has been proven to be at sound speed or even faster -- and (b) my second-guessing that Fritz Lang had a slower frame rate in mind and was thwarted by an executive decision at UFA or Parufamet.

As it turns out, my so-called "learned opinion" may be even less learned than I thought. I wrote film researcher Aitam Bar-Sagi asking for comment on Mr. Jensen's observation. I've been corresponding with Aitam for ten years now. Instead of a quick opinion, what I got back puts into doubt a lot of what I've believed about Metropolis. Judging by what Aitam has to say, some rewriting of the history of the film is in order:

Hi Glenn. Metropolis was planned from the start to be shown at around 26 fps. As fast as that might seem, that was the speed. It was written into the original script by Gottfried Huppertz in creative meetings before filming actually began. And one must remember that it was Lang who hired Huppertz, not UFA and not Parufamet. If Lang wanted the film to be projected at a slower speed, he would have instructed Huppertz to write the music accordingly, as is the case in the earlier Die Nibelungen where the speed was around 20 fps. In the latest restored version that premiered a few days ago in Berlin, Part One runs at 20 fps and Part 2 at 22 fps. Lang eventually premiered the film as he wanted, and that is also the way the film was shown in Austria and The Netherlands.

Parufamet's involvement was to restrict the showing of Metropolis to one single theater, insuring its failure. When the film was rereleased in August 1927, it was shortened by a quarter and had some inter-titles changed. It was sped up even more to 28 fps, and most of Gottfried Huppertz' score was replaced with stock pieces of music.

I tested the 'seconds' in the 10-hour clock a long time ago. They are not of equal length: the dials were moved by hand and so mean nothing. We also have no idea if they are supposed to represent real seconds at all. -- Aitam

I haven't seen the source of Aitam's information on all this but I know that he isn't prone to making wild claims. He's considered an expert on Fritz Lang and composer Gottfried Huppertz and is credited in the 'thanks' section at the end of the new restoration. Jensen and Bar-Sagi disagree about the regularity of the "seconds" ticked off on the on-screen Metropolis clocks, and I'd rather avoid an argument on that issue. What bothers me is the information that the fast projection rate was fixed before production began. Personally, I just can't see Lang approving shots of Maria bouncing up and down on the cathedral bell rope, and the shots of the False Maria gyrating her hips are indeed comical at the faster speed. I no longer have difficulties reading the film's fast montages, but I've seen Metropolis at least a dozen times now. We're attuned to "reading" rapid montages these days, but I have a hard time believing that Lang would want those quick montage cuts to be quite so quick. Yes, I know that Abel Gance and some of the Russians were into one-frame flutter cuts, etc, but Lang's style was totally different. After the heavy, slow establishing passages of Metropolis, the fantastic lab scene and the False Maria's erotic club dance are bravura passages of pure cinematic wonderment. The use of the musical term "symphonic" to describe the film's movements is not an artistic affectation.

In other words, I'm not refuting Aitam's authoritative information, but neither will I discount Filip's direct observations. The big clocks in Metropolis could very well be stage props manipulated by hand, and I have a feeling that Aitam has read a first person account that confirms this. But their motion looks awfully regular to me ... and Germany is a country that prides itself on precision instruments. For Filip Jensen's part, he reminds me that he's not talking about the big 10-hour clocks, but a smaller clock shown early in the movie. (pictured) So maybe Filip and Aitam are talking about different clocks.

I think Aitam should publish a myth-busting Concordance for Metropolis, indexing his impressive research with references. This is one complicated movie!

For the rest of us, scratching our heads, the best and most concise explanation I've seen of the vagaries of silent movie frame rates is at this McGraw-Hill Film History: An Introduction page. Just scroll down a bit to "Silent Films: How Fast is Too Fast?"

Filip Jensen's 19 or 20 fps sounds like my personal ideal projection speed for Metropolis, after my experience at that museum screening back in 2001. That screening required an intermission, if only to prevent organist Robert Israel from having a heart attack. But the bottom line is that the movie works fine at 24, too: KINO's discs are correct. Last week's North American re-premiere at the TCM Fest (April 25) was another one-of-a-kind event. The Alloy Orchestra's musical accompaniment fit the film like a glove, and the audience was mesmerized. When Metropolis comes to repertory theaters I'll probably be going one more time. For Los Angelenos, it will be playing starting May 14 at the Royal in Santa Monica.

If I do attend, I'll write up my overall impression of the restored cut for DVD Savant in a new review.  2  By that time I should have squirmed out from under all these technical imponderables and sorted out my overall thoughts about this revelatory experience.

May 1, 2010


Footnotes:

1. Many sound pictures don't always shoot at 24 fps either. It is common in 1930s and 1940s films to see street action, cars and people just "doing things" filmed at 20 or 22 fps to keep the action from dragging. Westerns almost always added a 'speed edge' to horse action, to the point that a horse filmed galloping at a true frame rate can seem unnaturally slow.
Return

2. Yes, I know: just how many times is DVD Savant going to write about Metropolis? Until I get it right, I suppose. The same thing applies to Major Dundee -- the beauty of this page is the freedom it affords me. I only hope a few readers tag along for the ride.
Return





A frame blow-up of a newly-recovered scene from Metropolis. The 16mm print was probably struck in Buenos Aires in the 1940s or 1950s, so that the dangerous 35mm nitrate source could be disposed of.

The printer used must have been formatted for Academy-aperture work, as shown by the cropping on two sides. The vertical stripe matted away on the left is where the optical soundtrack goes. The fat frame line bar on the top corresponds to the fatter Academy frame line, to maintain the aspect ratio at 1:37.



DVD Savant Text © Copyright 2010 Glenn Erickson

See more exclusive reviews on the Savant Main Page.
Reviews on the Savant main site have additional credits information and are often updated and annotated with reader input and graphics.
Also, don't forget the 2010 Savant Wish List.
T'was Ever Thus.

Return to Top of Page

Advertise With Us

Review Staff | About DVD Talk | Newsletter Subscribe | Join DVD Talk Forum
Copyright © MH Sub I, LLC dba Internet Brands. | Privacy Policy

Subscribe to DVDTalk's Newsletters

Email Address

DVD Talk Newsletter (Sample)
DVD Savant Newsletter (Sample)