directed by: Hsiao-hsien Hou


IMDB link: http://www.imdb.com/title/tt0826711

Submitted by Ian Jones on 2009-05-10

Ian Jones's comment:
The following is a classification of the shots of Hou Hsiao-Hsien's Le Voyage du ballon rouge (Flight of the Red Balloon) in terms of camera dynamics. The key to the eight categories is as follows:T+P - Prominent tilt(s) combined with prominent pan(s), from fixed camera positionTilt - Prominent tilt(s), with either no pans or only minor reframing pans, from fixed camera positionPan - Prominent pan(s), with either no tilts or only minor reframing tilts, from fixed camera positionS or MR - Static camera, or fixed position with only minor reframingST+P - Sideways track (left to right and/or right to left; the former is more common), accentuated by pans (with possible minor reframing tilts)F/BT + TP - Forward and/or backwards track, combined with prominent tilts and prominent pansST+TP - Sideways track (left to right and/or right to left; the former is more common), combined with prominent tilts and prominent pansIrreg - Irregular (a catch-all for the small number of shots that do not conform to these categories; see shot comments for further specifics on each shot)A note on the "Static or Minor Reframing" category: Normally, when dealing with a Hou film, it would not make sense to compound these two, as they would each make up a significant chunk of the overall makeup of the film. In this film, however, due to the paucity of shots that are actually shot with a static camera (complicated by screens-within-screens on two occasions), I opted to combine the two categories to make room for types of shots more frequently used in the film. However, I have tried to use the shot comments to provide an opportunity to be more specific about the shots in this category, just as I have done with the "Irregular" category.Additional note: The opening and closing credits for this film are plain while Helvetica titles over black. There is no imagery. I have opted to omit them.

S or MR
Number of shots:
MSL 56.1 22.4 101.4 18.7 34.4 100.1 117.5 60.3
StDev 101.7 10.1 61.1 43.9 22.1 86.9 104.3 68.9
Min 11.7 21.7 22.2 5.4 19.9 39.9 9.4 9.6
Max 482.5 43.4 207 204.2 77.7 294.3 295.3 225.2
CV 1.12 0.35 0.66 1.21 0.53 0.68 0.75 0.83

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

Users' comments:

Author: Yuri Tsivian Date: 2009-05-24

A technical advice to users: change Height to 1000 and Vertical resolution to 5 to 1 in order to see the graph of this extremely slow-cut film in full.

Author: Ian Jones Date: 2009-09-13

Note to readers:  These are some selections from a larger paper written on Flight of the Red Balloon.  I fear that it contains several instances of what I now consider to be methodological problems, including, for instance, an over-reliance on anthropomorphizing narration without exhibiting enough caution towards the potentially problematic hazards involved.  However, the section still works as an extended group of notes that adds extra background information on point of view in Hou.


These notes are best understood if the reader refers to both this graph and my subsequent graph of Flight of the Red Balloon in terms of location (2009-06-05).



Instances of Character Point of View in Flight of the Red Balloon


... James Udden, in his article entitled “This Time He Moves!,” charts the ways in which the changes in Hou’s style since Good Men, Good Women have both his disrupted previous predictable trajectory (up through The Puppet Master he seemed to be heading fairly consistently in the direction of longer average shot lengths and less camera movement) and detached him from his central place in the style that can be loosely termed “Asian minimalism” that he helped pioneer.  Udden warns that, “by this measure at least,” (referring to the stylistic features of long takes and little camera movement), “Hou has actually become ‘less’ distinctive, not more so” [1].  The sudden introduction of character POV shots in Flight of the Red Balloon, a bit of classical continuity that one would not find in a Hou film from the era of 1987 to 2003, may, in fact, be argued to be one more way in which Hou is becoming “less distinctive.”  However, before such sweeping judgments are passed upon the filmmaker, there is important work to be done analyzing concrete, textual mechanisms, as Branigan would say [2].




As was previously noted, the inclusion of a camera when introducing character POV is only a dominant device within the first six instances of character POV shots in the film.  After these six (marking roughly the first hour of the film), the tendency of the “A” component of the point of view setups to feature a camera drops off completely.  Shots 47, 48, and 49 of the film constitute the last use of this device, and make an interesting case study not only for their important placement in the trajectory of point of view in the film, but also because they contain interesting idiosyncrasies that work to provoke and probe some of Branigan’s assumptions.


Shot 47 begins with Simon holding Song’s video camera and pointing it to the right, towards the (off-screen) dining table in the apartment.  As he continues a conversation with Song, the camera pans over right to show what he is filming: she is at the table, making pancakes.   (Since both Simon and Song are framed in profile in this shot, and there is no attempt to emulate Simon’s perspective, this counts as a “reveal” rather than an “A”-“B” POV transition.)  The camera lingers on Song, then pans back to Simon, still pointing the camera.  With the cut to shot 48, the viewer is shown the red-curtained window directly behind Song: the film’s titular red balloon bobs outside, apparently “peering in.”  The cut to shot 49 then brings a return to the setup of 47:  Simon, still holding the camera, glances up, seemingly past Song and to the balloon outside the window (Fig 2.3).  Here, the narration suggests the viewing hypothesis that shot 48 is a “B”-category “point/object” shot, and the return to Simon in shot 49 provides a delayed “A” in which he notices the balloon.  (The viewer does not have to take shot 48 as Simon’s literal perspective; it more fittingly appears to be the “ideal position” type “B” that Branigan names in connection to what he calls the “cheated” POV schema) [3].  This hypothesis gains support as Simon shifts the camera, seeming to now film the balloon.  As Simon breaks his gaze and lowers the video camera, the camera pans to the right again, this time past Song, to the window, through which the viewer sees the balloon drift away.


And yet this hypothesis as to point of view is almost immediately called into question:  When Suzanne’s lawyer Lorenzo enters a moment later and asks Simon what he’s been filming, Simon answers only “I’m filming what she’s doing, how she makes the pancakes,” with no mention of the balloon—not a mention of having filmed it, or even having noticed it.  Shot 48 is now detached and unclaimed.  All evidence for the hypothesis that it was a character point of view shot has seemingly evaporated, and its status is now highly ambiguous.  According to Branigan, it is not an “error” to label the shot a character POV: it was merely a hypothesis constructed according to available evidence.  But this does not answer the question of what, now that the evidence for the hypothesis has been called into question, one can actually say that the shot is.  The narration has taken on the characteristic of being either somewhat of a baiting trickster (manufacturing arbitrary evidence to cue the viewer to hypothesize character POV when none was in fact in operation), or, alternately (an perhaps more interestingly), one who points: the visual narration suggests a narrator that does not remain “neutral,” but that can attempt to prod a character to notice something (such as a balloon outside a window), but cannot ultimately intervene upon these characters and their world.


Following this moment, the hegemony of the camera is quite suddenly toppled: no further character POV shots in the film make use of the device.  It is as if the narration is now more “comfortable”: limitations have been accepted, and the need for mediation has been cast aside.  From this point on, the film explores subtler systems of visual sympathy and alignment between the point of view of characters and the point of view of the film’s underlying narration.  


Towards a General Theory of Point of View in Flight of the Red Balloon


Shot 71 of the film provides a good example of how Hou is able to suggest what might be called non-literal point of view, or a general mode of “sympathy” between the film’s visual narration and character: or, as it were, object, for in this case the “character” is the titular balloon.  


Throughout the film, the camera has remained on ground (or at least floor) level: despite occasional tilts towards the sky to catch the balloon in flight, there have been no shots looking down on the city that could conceivably pass as the balloon’s point of view.  Shot 70 finds the balloon drifting away from Suzanne’s apartment, where Simon has just fallen asleep, over the roofs of Paris, accompanied by some non-diegetic piano music.  The balloon dips down towards the roof of a building, and there is a cut to a new shot, shot 71:  Here, the viewer is confronted with the first genuine high-angle shot of the film, looking down through a high window at Simon and a group of students touring the Musée d’Orsay.  Two things are worth pointing out here.  First, since Simon was last seen in his bed, some time has elapsed: two distinct moments have been bridged by the shot of the balloon’s drift through the sky, and the transition has been smoothed over by the continuation of the non-diegetic piano theme.  Secondly, the building the balloon floats down towards at the conclusion of shot 70 is most definitely not the Musée d’Orsay.  And yet, still, there exists here an undeniable connection between the camera angle and the balloon.  It is not correct to say, as Branigan might, that the viewer hypothesizes a character POV shot where none in fact exists.  Rather, it would be more fitting to say that here the viewer recognizes that this is a moment in which the narration has lost its detachment and become observationally aligned with the balloon.  It is not that the visual narrator “is” the balloon, but the viewer understands that it is somehow sympathetic with the balloon.  Also, although it might be best to generally exercise caution concerning assertions following Wilson's model, here seems accurate to say that, at this moment, the "implied (version of the) filmmaker" becomes sympathetic with the balloon: in a few moments, Simon’s teacher points out to her group of students, who are studying Félix Vallotton’s painting Le ballon ou Coin de parc avec enfant jouant au ballon, that the painter has portrayed his subject from “kind of a high-angle shot”—a sudden point of intersection between two artists [4].


This is something of a climax of the film.  Overall, if one were to characterize the underlying narrational point of view of Flight of the Red Balloon,  it seems most fitting, at least throughout much of the film, to term it a “furtive observer”—perhaps a polite guest, restlessly curious, but exercising caution so as to not be too intrusive.  This is especially true of the scenes in Suzanne’s apartment.  Examined separately, the scenes in the apartment have the lengthiest ASL of any location in the film, well above the ASL of the entire film (164 seconds, compared with 76.5 seconds for the entire film; in a distant second of ASL by location is the puppet theater, at a mere 79.5 seconds).  Long takes might connote a general sense of stability or comfort if they were more static, but the shots in the apartment are the most visually restless of any in the film:  Fully 100% contain major camera motion (panning and/or tilting), compared to 71% of shots in the film as a whole.  Additionally, 47% of shots in the apartment contain tracking motion (forwards, backwards, or sideways), as opposed to only 15% of shots in the film as a whole (23% of the shots already containing major camera movement).  Practical answers could be offered for this stylistic choice: for instance, the fact that the apartment was probably the most easily closed and controllable location used in the film, making the laying of dolly track substantially easier.  Here, the pro-filmic realities are, however, not as important to discuss as the effects the resultant style has upon the viewer.  Long takes that constantly drift back and forth, panning and tilting to and fro, create the overall impression of the point of view of a visual narrator unable to settle down, unable to engage intimately and inter-subjectively with the characters it presents to the viewer: the point of view of a restless visitor.


The impression created in these apartment shots is extended and magnified by the film’s frequent recourse to images shot through windows, often with a myriad of layers of reflections blocking any clear view.  Such shots constitute 24% of all the shots in the film, translating to 14% of the film’s total running time.  Shots such as these serve the function of presenting a visual counterpoint of staging-in-depth to the vertical staging present throughout much of the film (the balloon hovers over the city; Suzanne’s apartment is multi-tiered in its construction, and the major narrative concerns that take place there have to do with upstairs-downstairs conflicts between Suzanne and her tenets).  But they also emphasize a muddied dialectic of visual penetrability and impenetrability: the sometimes successful, sometimes frustrated desire on the part of the film’s narrator to peer into something it is not quite privy to. 


This style of detachment fully segues into an aesthetic of “sympathy,” “alignment,” and genuine intimacy only in the final ten minutes of the film:  It is as if the visual narrator has finally been “let in” to these characters and their lives.  Following shot 63, the piano-tuning scene, by far the longest shot in the film at an astounding 482.5 seconds (the runner-up, shot 33, encompassing the scene in which Suzanne realizes she’s misplaced her copy of a tenet agreement, is a mere 299.8 seconds), there occurs a precipitous drop in shot lengths for the remainder of the film (see Appendix B Fig. 3).  This speed-up in cutting rate is seemingly helped along by the entrance of music around shot 66, as well as the fourth and fifth appearances of the balloon.  Faster cutting alone cannot communicate a more intimate point of view:  This uptick in cutting rates is joined by a complimentary tendency towards closer shot scales, a greater concentration of shots with a stable camera, and an explosive increase in the concentration of character point of view shots.  These trends climax slightly after the high-angle shot 71, with a series of medium shots of Simon sitting in the Musée d’Orsay trading glances with the red balloon, hovering outside the museum’s skylight, framed not with a restless pan but in a shot pointing straight up, accented with a tiny but vertiginous bit of drift.


By the film’s conclusion, it is apparent that the processes of visual narration have become increasingly integrated with, and welcomed into, the world the film presents.  This more than simply a matter of identification with, say, the point of view of the character of Simon.  To put it in such terms is to accept Branigan’s dichotomy of “subjective and character-based” narration versus “neutral, objective” narration.  It has been the position of this study that these two poles can be productively negotiated between, allowing for a narrational presence that is not “subjective,” in the sense of being attached to a specific agent (an onscreen character or explicit off-screen narrator), and yet certainly not simply a neutral backdrop or colorless filter, but rather an isolatable, if somewhat elusive, point of view: following Wilson, the distinct character of the viewer’s visual access.


Ultimately, how essential is the task of connecting the character of visual access, the “implied (version of the) filmmaker," to the actual filmmaker?  In the case of Flight of the Red Balloon, such connections can flow quite freely after the initial groundwork has been accomplished:  The implied filmmaker seems to observe as a “visitor” to the scenes; Hou is a visitor in France.  The implied filmmaker at first seems most comfortable allowing access to character point of view in the presence of a mediating camera; Hou, of course, has a certain comfort with cameras.  There is no denying that a certain amount of insight into the film can be gained from such connections—and indeed, both of the above connections have been made to a certain degree in this very paper.  However—and here one can keep in mind Branigan’s hesitancy towards the concept of the “author”—the true value of such connections does not lie in a simplistic decoding of the film in terms of expression of biography, but rather in using information about the film’s creation to illuminate the ways in which the situation of a film’s creation resonates with (or fights against, if the case may be) the formal systems at work within the film itself.


After bemoaning the possibility that Hou might becoming a less distinctive director, James Udden acknowledges that “change has in fact been the norm for Hou Hsiao-Hsien from the beginning,” citing his transition from popular musical comedies to art films, and subsequent shifts in subject matter throughout the 1980s and 1990s [5].  This is one observation about Hou that can be said to ring true.  Despite Udden’s best attempts to isolate an overall trajectory of Hou’s style in an isolated phase of his career, Hou’s recent work has proved to be irrepressibly diverse.  Hou shows a general trend towards longer ASLs, and yet there are major outliers here: Three Times (2005) contains his quickest ASL (35.6 sec, disregarding intertitles) since Daughter of the Nile (1987) [6].  Tracking shots seem to be an increasing component of Hou’s vocabulary, reaching their current peak in Flight of the Red Balloon, and yet the recent Café Lumière features no tracking shots at all.  If anything can be pinned down about Hou, it is that he is difficult to pin down.  Techniques that were developed as a specific response to a specific problem—Hou’s long-take aesthetic developed as a side effect of his attempt to squeeze more naturalistic performances from his actors in the 1980s—have lingered on long past their expiration date; others, such as Hou’s static camera—a technique Hou made recourse to because he believed that a moving camera couldn’t adequately capture the feeling of the past—have been gradually phased out in importance [7].  Is there anything that can be made of such developments?


One answer to the riddle is to recognize that Udden’s search for an overall logic to Hou’s development as an auteur has the potential to be every bit as limiting and reductive as the attempts by other authors to pinpoint the essential “Chineseness” of Hou’s style.  A study of stylistic developments is important not because it delineates and comprehensively defines the boundaries of a director’s entire means of expression, but because it allows one to recognize the ways in which each film stands alone, as a text formed in response to a specific set of opportunities (for instance, in Flight of the Red Balloon, an invitation by Musée d’Orsay) and a specific set of limitations.  In this way, one is able to study concrete, textual mechanisms, à la Branigan, while simultaneously refusing to ignore the figure of the author.  As it has been shown, point of view in Flight of the Red Balloon has been one area where it has been profitable to undertake such an analysis.



[1] Udden, James. “This Time He Moves! The Deeper Significance of Hou Hsiao-Hsien’s Radical Break in Good Men, Good Women.”  Cinema Taiwan: Politics, Popularity and the State of the Arts.  Ed. Darrell William David and Ru-Shou Robert Chen.  New York: Routledge, 2007.  pg 92.


[2] References to Branigan throughout are references to Branigan, Edward.  Point of View in the Cinema: A Theory of Narration and Subjectivity in Classical Film.  New York: Mouton, 1984.


[3] Ibid, pg 116.


[4] Wilson, George M.  Narration in Light: Studies in Cinematic Point of View.  Baltimore: Johns Hopkins University Press, 1986. pg 135.


[5] Udden, pg 198.


[6] Ibid, pp 200-201.


[7] Udden, James. “Hou Hsiao-Hsien and the Aesthetics of Historical Experience.”  Ph.D. diss., University of Wisconsin-Madison, 2003. pg 111.  Udden, "This Time He Moves!," pg 190.

Add your own comment:
Your name:

Enter code from above picture: