Friday, September 16, 2011

Analyzing Action

A film critic named Jim Emerson recently deconstructed an action sequence from Dark Knight and claimed that it violated filmmaking rules and spacial relationships. He thought the sequence was indecipherable and that Christopher Nolan, when it comes to action, is essentially incompetent. 

As gentlemanly as that sounds, the methodology he uses is inaccurate and misleading. I’m not out to destroy this guy, but his observations are so flawed it just can’t be left in the universe like that. With the internet being what it is, I'd rather not have this misguided laugh track become the Loose Change of film geeks. Plus, Emerson sounds like a prick.

I don’t have a problem of criticism. In fact, I love film criticism and think it is vital to the historical and cultural dialogue of cinema. Great criticism helps define our interpretation of art and is a canon of thought that shapes both future artists and audiences. A good critic is just as valuable to the cinema as the movies themselves, and I have learned as much reading as I have watching or making (note: criticism as in critical analysis is not always the same as "reviews").

However, there is a point where criticism is so flawed and the author is pontificating on an area that is so out of his area of expertise, he is basically talking out of his ass. That is Jim Emerson’s knowledge of film space and the “rules” he has regurgitated out of a basic textbook.

I am going to go over his major points and explain the flaws. I’m going to skip over as many “opinions” as I can as he has every right to hold them, however smug they may be. This would be things like his personal preferences of how guns should fire, marksmanship, or how characters should act. Instead, I simply want to address points he explains as “rules” – which are not. Basically, anything he states as a “violation” is what I’m after here.

We’ll start with an accusation from him, and then my response. I’ve occasionally embedded snippets of his video.

INTRO: “A filmmaker has two tools to convey information visually – composition and cutting.”

This is old film school thought. It’s not even oversimplification, it’s wrong. It stems from the technological origin of silent filmmaking. Film began as “film,” literal strips of pictures moving through some sort of isolated gate that unified an image in sequential progression creating an illusion of movement. 

The addition of actually physically editing different strips created the idea of a “cut.” The delineation of composition and edits exists as a technical solution, but there is no conceptual reason why these two must live apart. The concept of a “shot” is a technical term that makes it easy to produce, but modern techniques like morphing and inverted blends can create cinematic ideas that are neither edits nor separated.

The idea that cinema exists “in the cut” implies that all filmmaking is based on some subconscious, comparative analysis of images. While the mind does naturally have a pattern recognition mode, it’s not the edit that triggers this. We are always doing this, in ALL ART, at ALL TIMES. In fact, as you get out of bed your mind is triggering a historical pattern response to orient you and make sure you don’t knock into your bedstand. It is not the placement of two images that creates the unique experience that is cinema. A typical motion graphics flash page on a website can fuck this concept in the ass.

Many of our old critical thoughts on the aesthetics of film are defined by the limitations of old technology and workflow. This includes sound, which was originally added as an afterthought and curiosity, but is as fundamental to cinematic technique as the concept of a “picture.” We again only separate because we must shoot sound separately from the picture, but this has nothing to do with the actual unit of a cinematic idea. David Lynch for one composes images with sound, and his images are inseparable from the audio.

As modern technology makes it easier and cheaper to access picture and sound and manipulate it, digital filmmaking will release our critical thinking from being so closely tied to shorthand production terminology, and reveal a purer cinematic ideal that is not afraid to embrace how we really experience movies.

Any analysis that views film from only from the prism of composition and editing, and excludes sound, has made a completely arbitrary line in the sand that does not reflect that actual totality of what you actually saw.

Now, Dark Knight.

CLAIM: Nolan violates the 180 degree rule, and this is confusing.

Emerson never explains why the 180 degree “rule” supposedly works. I have yet to read any precise technical papers on this, but it’s pretty obvious that as we tend to view images scanning on a horizontal plane during the course of a normal day - cars, lions and OJ Simpson will attack you from ground. 

As a survival mechanism, evolution has trained the human race to stare at the white’s of each other’s eyes to decipher danger and direction. Thus when someone on a screen stares left, we will instinctually look left to follow their eyes. A person then on the left looking right will reciprocate an interaction as our eye shifts to their direction.

However, Emerson has discombobulated this with action sequences, demanding that they strictly adhere to eyelines and a constant screen direction, as if once set they can never change (why not?). Even on an evolutionary level, the white-eye left/right instinct can be overridden. Stand on the ledge of a tall building and all you’ll look is down. Enter a cave of falling stalactites. Architecture, danger, and kinetics can change and reset where you look, and how you look. Action scenes can be solidly designed with this in mind, and it happens all the time.

Furthermore, Emerson confuses screen direction with meaning. An example is how he professes confusion over where Harvey Dent is sitting in the back of the truck. The sequence begins with Harvey walking to the back of the truck, and clearly shows him sitting on the passenger side. 

Emerson then claims he is completely lost as to where Dent is sitting for most of the action sequence because he is often “facing” the wrong direction in the chase. But Dent is an windowless contained space. 1) Part of his character’s perspective is to be isolated and confused to the outside world. It is ridiculous to flop around angles to match outside action as it defeats the subjectivity of his reactions  2) Yet the audience always knows he’s on the passenger side because 3) Nolan has chosen to constantly keep him framed on the identical right side to reinforce his positioning.

This is pattern recognition, and it overrides the 180 degree line. The audience cannot be convinced otherwise because Nolan locks it into their brains: Dent’s on the passenger side. And a character’s subjective perspective, how he feels about what he’s in, is something Emerson is tone deaf to in pursuing his eyelines.

Now I’m going to quickly analyze some of his criticisms:

CLAIM: “Again we don’t know but we assume since this is the only truck we’ve seen, that it’s the one Harvey’s in.”


Yes of course we know that’s Harvey’s truck because we’ve just cut to the “only truck we’ve seen.” Also, it has a fucking spotlight on it. He brings up a false question then answers it.

CLAIM: “This feels like a reverse angle which means this guy is now sitting where Harvey should be.”

Weird claim because by his own AXIS OF ACTION rules the cop is looking correctly from left to right, as the car is driving from left to right. 

And Harvey’s reverse pivots correctly.

CLAIM: “This little bit doesn’t accomplish anything. Why would you cut away from a convoy just when it’s getting going?”

This question of why we cut away isn't really a question but an opinion - an opinion of how he would have personally directed the scene. Very cute. But masquerading as technical analysis, the opinion shows shocking disconnect from the plot mechanics of the story. This obviously establishes the Joker is setting up his plan. Hitchcock himself stipulated tension is built by showing a ticking bomb to an audience at the beginning of a scene - instead of just randomly blowing it up. This scene sets the danger of the Joker as the timer and now the audience gets to anticipate his arrival instead of passively watching.


CLAIM: “The Joker shows up so brief it’s not really effective.”

Not effective? A guy in spooky white face and black eyes just popped out and kills a cop with a shotgun. What the fuck is he talking about?

CLAIM: “It would have been more ominous to just go straight into this next shot.”

Aside from ignoring Hitchcock for his directorial preference, this is where ignoring sound design would make you miss a wonderful detail: the shotgun blast echoes and raises the guitar hum in the next shot like an ear ringing. You still feel the Joker in the next sequence because you still hear him. This is sophisticated sound design working in tandem with the edit. All tools of cinema are at play here.

CLAIM: “A minor quibble, the previous shot sets this up to be a POV shot, and it isn’t a true POV shot. If it were, it would have to be three vehicles ahead of here...the imprecision of Nolan’s camera placement creates much more serious logistical problems later.”

Emerson just completely fucks this one up. There is simply no rule in cinema that states cutting to a person's face and eyeline automatically turns the next shot into a POV. In fact, it takes a lot of effort to convince the audience of any POV shot. It generally requires a hand held or steadicam feel amongst other signifiers. Establishing a POV from a car has a library of signifiers that clue the audience - namely is that there is a piece of the windshield to lock the viewer's position.

Furthermore, this is clearly a reverse master shot of the entire scene as the camera is placed higher than eye level and tracks in smooth, omniscient manner. One can only interpret camera meaning this poorly if your sense of visual aesthetics is completely broken.

On another note, the timing of the flame glow on the cop faces connects with the wreck as they pass. Emerson may not understand this, but that is a directorial choice that lead to a complicated rig of a moving car and special effect lighting. It is not a decision made lightly. Contrary to his assertion, Nolan’s camera and timing are methodical.

CLAIM: “Did I say 2, no in the next shot there are 3.”

He points out that in one shot there is an extra cop car in the back. This is obviously a continuity mistake in one shot. Shit happens. But you can clearly see they tried to edit around so that the third car is obscured quickly. In normal speed you really don’t see it. It is only seen for a couple frames before obscured by the lead car, and that microedit is the ONLY time it happens. This is precisely the type of editing fix that a filmmaker agonizes over to see what they can get away with but Emerson now uses a freeze frame as if the entire sequence is compromised.

CLAIM: “Where is this expendable guy?”

Emerson does have a point here (throw enough darts, I guess). This is the one beat where the action can legitimately be misconstrued. The guy is in the car in the front looking in his rearview mirror. Then the truck rams second car beside it, possibly insinuating he was in it.

However the solution is not to cut out the close-up as he implies (if he directed a chase sequence it would apparently never have a cutaway to any drivers). It needed a shot of the second car catching up to the truck to clarify, or the hit could have happened in his rearview mirror connected to his look. Or maybe not, it certainly isn't a deal breaker. But then Emerson fucks his point up with-

CLAIM: “We’re introduced to people in close-up or medium shot with no context, just a second or two before they’re dispatched.”

Context is highly relative by his terms. But adding reactions of drivers in multi-car chases is very common, humanizing, and not a sign of incompetence.

CLAIM: “You would think he would be looking in the direction he got hit, but wherever he is Harvey knows which way the van is traveling, why don’t we?”

Emerson makes a point that the truck got hit and Harvey faces the wrong direction for his reaction, but our critic has pulled a fast one. Using the same angle we use to establish Harvey’s seat in the truck, he actually turns first to his right toward the back during the hit, and then as he looks anxiously toward the drivers Emerson FREEZE FRAMES on his left turn, misleading the audience. Very unethical.

He then claims confusion as the editing violates the 180 degree line. But here is another example of how that “rule” is successfully broken. The action cut of Harvey being jolted is split with his head turn and the after shake of the SWAT drivers. The ACTION MATCH CUT clearly seams it together as the only thing we really react to is the impact. 

The simple question that must be asked is what happened in this scene? Truck hits SWAT. Harvey jolts. SWAT drivers react and keep driving, and we know they are driving the truck. This is very clear without Emerson’s obfuscation of the 180 degree line.

CLAIM: “If we think of this less as a three dimensional space than a two dimensional graphic space, like three flat comic frames, then the shots do make a kind of sense.”

He uses this completely random thought game to justify his argument that the sequence doesn’t make sense. But not only is film not two dimensional, it’s not even three. It’s got the fourth dimension of time, and like any mathematical equation the simpler the dimension the less you can solve. That’s why it doesn’t work in his freeze framed two dimensions, but works in our active moving four and allows to factor in kinetics like shock cuts.

CLAIM: When the semi hits the SWAT van, it flies into the river in the wrong direction.

This claim is inexplicable.

The sequencing starts with the semi hitting the van and pushing it to the right.

The driver is thrown to the right off the impact as the van turns right.

The semi twists into the next lane, acting as the force of the impact, to the right.

The van flies off into the river to the right.

It’s all correct. Emerson is just confused.

CLAIM: “Didn’t the SWAT truck and the other two trucks used to be where the semi is now? With the river on the right?”

Water is on both sides, as bridges over rivers are apt to do. This is just trolling.

CLAIM: The Joker shooting gun/bazooka sequence violates the Axis of Action and is confusing.

This adherence to this generic axis of action just doesn’t correspond to how we grasp images, especially in high velocity car chases. The reality is that each action sequence can have it’s own internal architecture in which a number of factors can influence how we perceive it. Just as much as standing on the ledge of a building would shift your awareness downward from left to right (making the axis of action up and down), factors like the actual shape of the environment can rearrange how we orient our screen direction.

In this case, this is very precise geometric construction where two cars are  driving down two narrow lanes, side by side, separated by a divider with columns that whiz by between them.

This helps orient us like standing on the the ledge of a tall building in that we are constantly calculating where our boundries are. The vice-like pathway also is abundant in movement reminders with passing background and passing columns in foreground. We clearly see how reversing shots between the Joker and SWAT drivers relate by the flipping movement of the backgrounds. If you only analyzed this sequence from a two dimensional axis of action, you would miss those four dimensional movements that frame each shot.

Then we factor that the semi truck looks very distinct from the SWAT truck, which reinforces symbolic, object, or pattern recognition (whatever you want to call it). 

The problem with a clinical 180 degree line technique of filmmaking is that ignores human memory, as if all we did were look at screen direction. If this were so, POV shots would never work. Instead, POV’s work off of association where the audience naturally interprets the shot as from within someone’s head. The human mind is powerful: it can decipher information from many different ways. Symbolic association is one that Nolan is very good at, and the sequence is confidently made.

In fact, there is an actual line cross that Emerson praises. When the Joker fires into the truck, he exclaims that the cop inside is finally linked to what side he is sitting on. The cut from outside (right to left) to inside the truck (left to right) is a line cross that would “confuse” which directions he is going in, but works because it is a MATCH CUT on the bullets. But Emerson thinks this finally clues him into where he is oriented – using the association the rest of us already use during the whole sequence.

Here, axis of action gives an overall sense of direction, but only as an overall pointer.

CLAIM: Emerson has recut a sequence in where Batman crashes into the garbage truck. He claims he has fixed Nolan’s edit and that it is better.

All because he misunderstands a shot where the Batmobile crosses under the moving camera to build acceleration. Objects get faster as they come closer to camera. This is just a variation of a car flying over camera mounted to a road.

Also, that editorial “fix” is ridiculous.

CLAIM: The Batmobile jumping over the car with the explosion has a continuity problem with a disappearing SWAT truck.

Nobody’s perfect.

CLAIM: He claims Salt is better directed than this.

Nobody's perfect.*

*Yeah yeah, I know. I should talk.