Videographic Deformations: Equalized Pulse


This is the third and final (and, to me, most interesting) excerpt from my essay draft on “Videographic Criticism as a Digital Humanities Method.” The first laid out my approach to deformative criticism via the format of PechaKuchas; the second explored videographic 10/40/70 analyses. I highly recommend watching some of the musical videos discussed near the end of the post.

A videographic 10/40/70 relies upon the single shot as the core unit of a film, a key tendency common to much academic work on moving image media. My third and final type of videographic deformation also highlights the shot, but from a distinctly different approach. One of the most prominent forms of quantitative and computational analysis within film studies is statistical stylistics, especially as shared on the crowd-sourced Cinemetrics website. While there are numerous metrics on the site, the most common and well known is ASL, or average shot length, computed by dividing the time of a full film by its number of discrete shots. The resulting number indicates a film’s overall editing pace, charting a spectrum from quickly-cut movies (such as Batman Begins at 2.37 seconds or Beverly Hills Chihuahua at 2.72) to longer-take films (such as An American In Paris at 21 seconds or Belle du Jour at 24).[1] The most typical range is between 3 and 8 seconds per shot, with much variability between historical eras, genres, national traditions, and specific filmmakers.

An ASL is in itself a kind of deformation, a reduction of a film to a single numeric representation. Cinemetrics does allow more detailed quantification and visualization of a film’s editing patters—for instance, this is a more granular and graphic elaboration of Mulholland Drive’s ASL of 6.5:[2]

MD Cinemetrics visualization

But these numbers, tables, and graphics make the film more distant and remote, leaving me uncertain what we can we learn from such quantification. According to Yuri Tsivian, Cinemetrics’s founder, the insights are quite limited: “ASL is useful if the only thing we need to know is how long this or that average shot is as compared to ASL figures obtained for other films, but it says nothing about each film’s internal dynamics.”[3] Certainly comparison is the most useful feature of ASL, as it allows quantitative analysis amongst a large historical corpus, a general approach that has proven quite productive in digital humanities across a range of fields. But I wonder about Tsivian’s quick dismissal that ASL “says nothing about each film’s internal dynamics.” Doesn’t a film with a 2.5 second cutting rate feel and function differently than one with a 15 second ASL? Certainly, and it doesn’t take a quantification to notice those differences. But perhaps such a quantification might guide a more thorough understanding of editing rates by extending the deformation?

Videographic methods allow us to impose a film’s ASL back onto itself. I have created a videographic experiment called an “equalized pulse”: instead of treating ASL as a calculated average abstracted from the film, I force a film to conform to its own average by speeding up or slowing down each shot to last precisely as long as its average shot length.[4] This process forces one filmic element that is variable within nearly every film, shot lengths, to adhere to a constant duration that emerges quantitatively from the original film; but it offsets this equalizing deformation with another one, making the speed of each shot, which is typically constant, highly variable. Thus in a film with an ASL of 4 seconds, the equalized pulse extends a 1-second shot to 25% speed, while an 8-second shot runs at 200% speed. If you equalized an entire film to its average pulse, it would have the same running time and the same number of shots, but most would be slowed down or sped up to conform to an identical length. Every shot exerts the same temporal weight, but each feels distinct in its tempo and pace. The result is, unsurprisingly, very strange—but I believe productively so.

What does Mulholland Drive look and feel like when equalized to a pulse of its 6.5 second ASL? Can we learn something about the “film’s internal dynamics” more than its numeric representations on Cinemetrics? Take the film’s opening scene following the credits, with Rita’s car accident on the titular street; in the original, it lasts 4:07 with 49 shots ranging in length between .3 and 27 seconds.

The deformed version with an equalized pulse of every shot lasting precisely 6.5 seconds runs 5:18, as the original sequence is cut comparatively faster (ASL of 5.04 seconds) than the film as a whole. The effect is quite uncanny, with super slow-motion action sequences bookended by sped up shots with less onscreen action; the car accident is particularly unsettling, turning a 9-shot, 6-second sequence into a grueling and abstract 58-second ordeal that oddly exaggerates the effect of experiencing a moment of trauma in slow motion. As a whole, the video does convey the sense that a pulse of 6.5 seconds feels quite deliberate and drawn out, although the variability of action obscures the consistency of the editing pulse.

Another scene from Mulholland Drive offers quite different effects, despite the same algorithmic deformation to its same equalized pulse. The memorable scene in Winkies Diner, where two unnamed men discuss and confront a dream, has always been the film’s pivotal scene for me, signaling its affective impact that transcends any rational comprehension or interpretation. When equalized to a 6.5 second pulse, the scene’s uncanniness is ratcheted up, downplaying the dialogue rhythm for a more even distribution between the two men. The slow motion close-ups with distorted voice highlight the dreamlike quality, and the overall slower pace increases the sense of foreboding that already pervades the scene. By the time the horrific bum is revealed at the scene’s end, I find myself completely enthralled by the editing pulse and pulled into the affective horror that the scene always produces, suggesting that its impact is not dependent on Lynch’s designed editing rhythms. I have not extended this equalized pulse to the entire film, but clearly each scene and sequence will feel quite different, even with a uniform shot length throughout.

Mulholland Drive is a film packed with abundant strangeness, even before its deformation; how does an equalized pulse impact a more conventional example? Even though Mildred Pierce features the unusual combination of noir crime and family melodrama, it is still a far more straightforward film in keeping with its 1940s era. Its ASL of 10.09 is much slower than films of today, but is fairly typical of its time. Equalizing the pulse of a crucial scene in the family melodrama, with Veda driving a wedge between Mildred and Monty who finally end their dysfunctional relationship, highlights various character interactions.

When Mildred gives Veda a car, it speeds through her thanking her mother but lingers on her exchange with Monty, underscoring the closeness between the stepfather and daughter—in the original, the emphasis is reversed in terms of timing, but equalizing the shots actually better represents Veda’s attitudes. The deformation lingers over shots without dialogue, letting us closely examine facial expressions and material objects, but speeds through lengthy dialogue shots, like an impatient viewer fast-forwarding through the mushy emotional scenes. The final lines exchanged between Mildred and Monty are unreasonably drawn out, milking their mutual contempt for all that it is worth. The scene is still legible, especially emotionally, but it redirects our attention in unpredictable ways—arguably a key goal of an effective deformance.

What about the other end of the pacing spectrum, equalizing the pulse of an action film like Raiders of the Lost Ark. The film has an ASL of 4.4 seconds, longer than most contemporary action movies but still quite brisk, especially for director Steven Spielberg. I deformed the iconic opening sequence, but used the sequence’s faster ASL of 3.66 rather than the whole film’s pacing, as that allows for a direct comparison of the original and equalized versions.[5]

The effect is definitely striking, as the deformed version races through the build-up toward action and peril, while lingering painfully on darts flying through the air, near-miss leaps, and other moments of derring-do. In the slowed down shots, you notice odd details you never would see in the regular film, like the discoloration of Indy’s teeth, and sense a very different momentum. When placed side by side with the original, it highlights how much of the sequence is weighted toward the approach and build-up rather than the action, while the deformed version lingers on moments that regularly flit by.

Raiders Equalized Timeline

The editing timeline visualizes these differences, but in a way that is analytically obscure; the videographic form allows us to feel and experience the analysis in ways that computational visualization cannot. What stands out most to me in this watching and listening to this deformation is the role of music, as John Williams’s score still manages to hit its key themes and punctuate the action, despite its variable tempo and rhythms.

This experiment in equalizing a film’s pulse points most interestingly toward different types and functions of rhythm and tempo. In a conventionally edited film, variation of shot length is a main source of rhythmic play, both in creating emotional engagement and guiding our attention. Eliminating that variation by equalization creates other forms of rhythm and tempo, as we notice the relative screen time given to various characters, anticipate the upcoming edits in a steady pulse, and engage with the interplay between image and sound. These equalized deformations highlight how much the analysis of editing and ASL privileges the visual track over the audio—we are not quantifying audio edits or transitions in such metrics, as sounds bridge across shots, slowing or speeding up like an accordion.

Experimenting with these equalized pulse videos piqued my curiosity in how visual editing functions in conjunction with music, especially for instances where the musical track is more dominant, as with film musicals or music videos. These explorations into musical sequences proved to be the most exciting examples of equalized pulse, as they highlight the transformation of rhythm and tempo: the musical track stretches and squashes to create unpredictable rhythms and jettisons its standard tempo, allowing the steady beat of the changing visuals to define the speed.

For instance, “Can’t Buy Me Love” from The Beatles film A Hard Day’s Night becomes a collage of fast and slow motion when equalized to its sequence ASL of 4.9 seconds, making an already playful and experimental sequence even more unpredictable. Musical sequences combined with dance add another layer of rhythmic play, as with the transformation of Singin’ in the Rain’s “Broadway Melody” into a deformed and almost uncanny work when equalized to its ASL of 14.9 seconds.

Musical numbers typically are edited at a slower pace than their films as a whole, providing more attention to performance and dance without being pulled away by edits. A rare exception is one of the fastest cut films listed on the Cinemetrics site, and certainly the fastest cut musical I know of: Moulin Rouge, with an ASL of 1.9 seconds.

The “Roxanne” number, with an even brisker ASL of 1.05 seconds, is the only equalized pulse video I’ve yet made where the visual tempo becomes noticeably dominant, offering a steady beat of images and sounds whose speed deformations go by so quickly as to often escape notice.

These equalized pulse versions of musical numbers are the most engaging and affective examples of videographic deformations I have made, functioning as compelling cultural objects both on their own and as provocatively deformative paratexts. They also demand further analysis and study, opening up a line of examination concerning the relative uses of edits, music, and dance to create rhythm and tempo. As such, these videographic deformations are not scholarship on their own, but they do function as research, pointing the way to greater scholarly explorations. Whether that subsequent scholarship is presented in written, videographic, or multimodal forms is still to be determined, but I hope that this discussion has shown how videographic criticism is more than just a form of dissemination. Transforming a bound cultural object like a film into a digital archive of sounds and images enables a mode of critical engagement that is impossible to achieve by other methods; as such, videographic criticism functions as a digital humanities research method that is poised to develop the field of film and media studies in unpredictable new ways.

Some bonus equalized pulse videos to consider:


[1] Unless otherwise noted, all ASL data are taken from Barry Salt’s dataset on Cinemetrics; even though the site includes many more films with crowdsourced information, I have found they lack consistency and methodological clarity of Salt’s list, which is easier to compare among films with.

[2] Salt’s list doesn’t include this film, so I used the ASL and this graphic from Nikki Esselaar’s submission.

[3] Yuri Tsivian, “Taking Cinemetrics into the Digital Age,” Cinemetrics.

[4] The process to do this is fairly straightforward in Adobe Premiere: first cut the source video into clips per the original edits. Then select all of the clips and use the Clip Speed / Duration tool. Unlink the Speed and Duration variables, and enter the number of seconds and frames in Duration corresponding to the ASL. Relink Speed and Duration, and be sure to check the Maintain Audio Pitch and Ripple Edit buttons. The only troubles come when a clip is stretched or sped up more than 1000%, as then the audio needs to be manually processed with more complex intervening steps.

[5] The opening 12:47 of the film consists of 209 shots, resulting in a 3.66 ASL.

2 Responses to “Videographic Deformations: Equalized Pulse”

  1. 1 Kaitlin Fyfe

    I agree with Yuri’s hesitation about ASLs as an indicator of the internal dynamics of a film. I think, once you move away from ASL as a comparative tool, it loses a lot of utility. More importantly, at least from my perspective as a teacher of editing, I think it’s a bit too far removed from the typical editing thought process and procedure. When you’re cutting, you’re certainly aware of the pace at which you’re going, and thinking about a given shot’s length in relation to the length of the shots around it. But the ASL seems more an artifact of the numerous variables an editor juggled in a given scene, frame by frame, rather than a tool used in editing.

    I tend to find generating a shot length chart for a given scene (much like the Cinemetrics charts, but I do mine via Premiere and Excel) and then keeping it visible in my visual field as a kind of map while I roll back and forth through the scene (again in Premiere). Then again, the goals of my analyses may be different than yours.

    All that said, this is a really interesting experiment. I could see it being useful to quickly highlight shifts in rhythm on a clip during a lecture or a talk. Kudos.

  2. 2 Kaitlin Fyfe

    Now that I think about, you kind of are using ASL competitively here, by juxtaposing actual shot length with average shot length and visualizing that with a speed change. Again, an interesting technique.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: