A funny thing happened when I watched this clip. The sound on the clip was muted; basically I could hear sirens and the radio voice-over, but everything else was too faint to be understood. I tried several times, hoping it were just a bad server connection, or, I don't know--solar flares? The result is that I watched it several times, basically in silence. The funny thing is, I got a lot out of it. So I can say that the first person POV was exceptionally helpful. And amazing camera work.
We know there's been a prison break the second fingers appear on the rim of the barrel labeled San Quentin . Daves immediately ramps up the tension with the prolonged barrel rocking (fall, already!), instills fear when the barrel starts rolling down a steep hill, and disorients us with the POV shot rolling down said steep hill. Our disorientation is intensified by the use of an iris (the center of a wheel spins faster than its periphery).
The steady iris shot that follows has the opposite effect. By limiting the field to our escapee, who thankfully survived the tumble, we get to reorient as we watch him doing it himself.
Watching him carefully hide his prison shirt (instead of just tossing it) and then look back at San Quentin, we see him realize that, yes, he's out. Free, though? We'll see. He gets a reminder he's being chased; sirens, a quick pan, and motocycle cops coming around the bend tell the story. He has to get out of here quick, and now he knows where the road is. And which way San Quentin lies.
Low camera angles and tracking shots show that he is creeping along the fence by the road, waiting for a car going in the right direction, and keeping an eye out for cops. When he sees the jalopy, he knows his chariot has arrived! But he's not out yet. He gets in the car, and the camera pans down to his muddy shoes. Slightly suspicious to have wet shoes walking along a dry road, isn't it? Just have to play it off, so pan back up to the driver's face, where the camera stays, as if we were glaring at him, looking for any sign he might be catching on. He may be a chump, but he might not be stupid on top of it. And those signs come through the facial expressions of the driver. Where, of course, the camera stays trained until the driver asks a question the escapee doesn't want to hear. The camera pans to the front; we break eye contact. Oh, but wait. Isn't that slightly suspicious as well? He pulls it together, and the camera pans back to staring at the driver, as do we. Hoping.
But this guy looks like he's getting more and more suspicious. If he's going to need to do something drastic, he won't want any witnesses, so the camera pans to the back, to check on traffic. Seeing none, the camera pans right back to the driver's face. Clearly he suspects something is wrong here, but he doesn't know exactly what. And we are in the same boat; we know what he doesn't know, but we also know something is wrong, not what it is, exactly.
So when the radio announcement comes (the only voice-over that was audible in the whole clip) and we see the road signs, we see in his face that he's figured it out. So he has to go and we have to steal his car. Next we see our fists pummeling the face we've spent the last few minutes glaring at.
So why was so much time spent on the driver's face? It's justified as a plot device, because an escaped convict would want to keep a close eye on someone who could potentially turn him in. But I think it serves a few other purposes as well. The most interesting character in a car sequence, by far, is the driver. He is interacting with two worlds simultaneously; the road and the passengers. In this clip, the driver was even more interesting for the danger he (unknowingly) represents to this passenger. Focusing on the driver's face allowed the background to float along peacefully, in counterpoint to the tension that every other element so far has built up. It also contrasts with the road shot facing forward, with the background rushing past, heightening confusion and tension, and the shot to the back, with the world quickly receding. And spending so much time on his face made the beating that much more jarring.
This opening incorporated noir elements we've seen in the previous three. In M, we first see the evil protagonist as a shadow, just as here. La Bette Humaine spent a lot of time showing the world from the POV of a locamotive. And The Letter made ample use of gently flowing backgrouds to counterpoint dastrardly deeds and scheming. They each ramped up the tension from 0 to 100 in the space of 4 minutes. They each leave us with more questions than they answer, an irresistable need to answer those question, and certainty that those answers will not be pretty. Still, we're going to watch.