The following is a piece I wrote in my senior year in college as the final project for a directed study in film semiotics. You can read the entire article -- which includes screenshots from the film -- at the Second Reel Annex.
Maintaining that illusion requires that scenes (this will refer to Metz's definition throughout) be constructed of many shots — and many edits — in order to constantly reassure the spectator that s/he has complete visual control of the diegetic space at all times. Raymond Bellour has thus observed resultant symmetrical patterns within the scene, complex patterns which structure the telling of the narrative and which belie the scene's apparent simplicity.2
It is true, however, that not every shot of every scene is a static presentation of a character's point-of-view. At these times, the lack of a diegetic agency onto which camera control can be displaced is recuperated by the camera's ability, as Stephen Heath has theorized, to "narrativize" itself; that is, the camera, when not allied with character point-of-view, simply hides its presence behind the narrative action that it always keeps in front of it.3 Even when the camera moves, the threat of that movement (and thus of the camera) becoming noticeable as a unique entity is subdued by the presence of narrative action before it.
When, however, camera movement becomes very intricate, the threat of the camera proclaiming its autonomy becomes greater. This situation arises in those long takes which present an entire narrative moment, an entire "spatio-temporal integrality experienced as being without 'flaws,'" in one shot, without editing. The autonomy of the camera is proclaimed in these shots when the moving camera leaves characters behind, abandoning them outside the frame, or when characters are allowed to leave the frame on their own. The former instance demonstrates the autonomy of the camera by showing that the camera controls the process of narrativization by selecting its own objects. The latter calls attention to an offscreen space that the spectator can not see.
Long takes of this type, then, do not adhere to the same suture techniques of the scenes allied with character point-of-view, or of those shots that are highly narrativized. This does not mean, however, that these long takes are necessarily devoid of the complex structure that Bellour observed in the Metzian scene. On the contrary, two shots from the Orson Welles-directed Touch of Evil (1958) demonstrate that the highly-mobile long take can also be structured upon those very threats to suture and narrativization, and that the resultant structures are very complex indeed.
Continue to Example 1: The Opening Shot
All footnotes can be found at the end of this essay.