Proximity

This is a celebration of camera-to-subject distance. Powerful and misunderstood in equal measure, it’s an incredibly important property of any shot. We’ll go through some basic theory, I’ll describe the psychological implications I believe it to have, and we’ll look at how understanding these can affect the way you design shot structures for a scene.

In normal life, things that are closer to us appear bigger and things that are further away appear smaller.

Let’s call depth the dimension passing through the lens and receding from the camera in the direction it is pointing. The apparent size at camera of an object of a given size depends on its distance to the camera. The depth difference is the distance between two objects along the depth. A depth difference amongst two objects of equal physical size will give rise to a difference in apparent size at camera.

I want to call depth rendition what some people call depth compression/expansion and yet others refer to as perspective, perspective distortion or one of any number of terms. I use it to describe the difference in apparent size caused by a given depth difference between objects of equal physical size. If we create a depth difference between two identical objects, a frame in which these two objects appear very different in size exhibits expanded depth rendition compared to a frame in which they appear similar in size. The latter is said to be compressed in rendition.

Take a situation where objects A and B are the same physical size, but there is a depth difference (A-B) between them giving rise to a difference in apparent size at the camera. The ratio of A’s apparent size to B’s apparent size is equal to the ratio between their distances to the camera. It can thus be said that the difference in apparent size depends on the ratio between the depth difference and the distance between the camera and the first object. It’s not the absolute distance between the objects that counts, but rather how significant this distance is in comparison to the camera-to-subject distance. If the camera is a hundred feet away, a one foot separation between two objects is not going to make them look very different in size in the frame. But that same one foot distance is going to produce a massive size difference if the camera is two inches away from the closer object – why? Because that same one foot is large in comparison to two inches, but tiny in comparison to a hundred feet. Since depth rendition is nothing but the apparent size difference created by a given depth difference, we can say that depth rendition is a function of camera to subject distance.

It’s the same mathematical relationship according to which focus has to be pulled faster as the subject gets closer to the lens, or printing time differences in the darkroom have to be measured relative to the total exposure. Someone who prints in the darkroom will understand all this discussion in a pinch, because if I ask them “How many stops extra is 2 seconds printing time?” they’ll reply “It depends what your basic exposure was. It was 2 seconds, another 2 seconds is a stop. If it was 160 seconds, you won’t see that extra 2”.

(The mathematically inclined can enjoy a more rigorous proof here.)

Assuming you understand the above, now consider what most photography students hear in a classroom, and therefore carry through to their careers: “Depth compression/expansion is the effect whereby the lens used alters the relationship between depth difference and perceived size difference from what is seen to be ‘normal’. Wide lenses exhibit depth expansion, whereby small differences in depth give rise to large differences in perceived size, so that an object can be made massive in relation to another just by moving it closer to the lens, and size differences are exaggerated. Long lenses exhibit depth compression, whereby differences in depth do not give rise to large perceived size variations, so we can move something close to the lens or put it way in the background and its perceived size does not change much.”

In much the same way that non-experts in aviation have a mistaken view of how wings work, non-experts in photography and cinematography have a mistaken view of how focal lengths affect depth rendition. Depth rendition is not affected by focal length. It depends only upon camera-to-subject distance. If we maintain our camera-to-subject distance and change focal length, depth rendition does not change. The only way focal length affects depth rendition is if we decide to keep subject size in the frame the same, because then we have to move the camera further from or closer to the subject and this changes the depth rendition. So a wide lens seems to expand perspective because it forces us to move closer. A long lens seems to compress perspective because it forces us to move further back. In fact, there is no depth rendition difference between using a 100m and cropping into a 32mm shot to obtain the same field of view.

Think about what would happen if this wasn’t the case: if each focal length had a unique depth characteristic, and if using different focal lengths changed depth rendition even if we did not move. Different sensor sizes force us to use different focal lengths to achieve the same field of view, so every different sized sensor would have different depth rendition characteristics. Very small sensors would massively expand perspective whereas Imax cameras would compress them like crazy. And we know that’s obviously not the case – if I stand a 2/3” chip camera next to a 35mm camera, at the same distance from the subject, and use whatever focal length is required to achieve the same field of view, the depth rendition will be the same on both cameras even though I’m using a much longer focal length on the larger chip camera.

Let’s be clear – we are not wrong when we swing a wider lens because we want more depth expansion. But unless we realise that it’s not the lens itself, but the fact this lens allows us to move closer than a longer lens to attain the same subject size in the frame, then we have an inaccurate picture of what’s really going on, and the reason I make such a big deal out of a seemingly small incompleteness in understanding is that without understanding this properly, it is impossible to make use of the phenomenon I’m about to describe. Read on!

We’ve established that depth rendition depends only on viewer to subject distance. This is true for the human eye, and it’s true for the camera. It must follow that the visual system is used to using the depth cues in a scene to help determine distances. No matter what lens you use, the distance between the camera and the subject is ineluctable and is forever imprinted in the image in the form of depth cues which the viewer’s visual system is subconsciously reading all the time and using to draw conclusions about distance. This is the central contention of this article. It is the reason street photographers are told to ‘get in there’ rather than use a long lens. Once a director told me she wanted to get ‘close’ to the characters by using a long lens. She may have a unique way of seeing, but it’s more likely she misunderstood lenses; the way to get close to the character is to move the camera closer!

On a short I shot earlier this year, our protagonist spots two characters interacting across the road, and stops to look at them. We started the scene tracking with him on a 35mm, and stopped with him. The director then wanted a shot of the two characters across the road. I just walked across with the 35mm on and lined up a two shot, and the director commented that it just didn’t feel right. After thinking about it for a second, I realized that it was a question of narrative point of view. We are telling this scene from this character’s narrative point of view, so even if the shot of the characters across the road is not a literal POV, it will only feel like the scene is consistently in our lead’s narrative point of view if we keep the camera close to him and switch to a longer lens to shoot the characters across the road. We shot it on an 85mm. The camera to subject distance is many times greater than the distance between any two elements in the frame, creating our so-called depth compression. Our brain is used to reading these depth cues continuously, so it effortlessly and subconsciously realizes that it is seeing these characters from some distance away, and concludes that this must be something like what our character across the road is seeing, even though the size of the two characters in the frame is identical to how I had them in the 35mm version. The only thing that has changed is the depth cues produced by moving the camera back across the road.

Instead of using the term camera-to-subject distance, I’d like to call it proximity. I owe this terminology to director Oscar Nobi, who came up with it whilst discussing depth rendition in his film "Sorinne". To the writer Gul Davis, on the other hand, I owe the term narrative point of view, which I used in the example above. This doesn’t refer to what we filmmakers understand as ‘POV’, where the camera is the character’s eyes, but to whether we are telling the story from the perspective of a particular character at any given point in the film. If there is considerable physical distance between two characters and we choose to keep the camera nearer one of them, we are telling the story from his narrative point of view, no matter the focal lengths we use or what we point the camera at, because the eye reads the depth cues.

Consider instead a scene with only one character. What happens when we vary the proximity? There’s something I’ll call emotional distance, which is what it sounds like – how close we feel emotionally to a certain character. In real life, if we are physically close to someone, chances are we have an emotional connection with them, and conversely, we don’t usually share emotional moments across canyons so long distances tend to evoke emotional detachment. This is not a straight relationship - it depends on the context – but more often than not, emotional distance is linked in some way to proximity.

Sometimes you wish to maintain a consistent proximity between two shots for various reasons: narrative point of view, emotional distance, whatever it is. In the example from my short, it was a question of coherent narrative point of view throughout the scene. In this case it would make sense to use a similar proximity but change the lens. Sometimes you wish to modulate proximity throughout the scene, for example to get increasingly emotionally close to a character as the scene goes on. In this case, physically moving the camera between shots will do the trick. What about zoom moves versus dolly moves? Pushing in to a character using a dolly alters the proximity, whereas zooming in does not. Zoom moves are largely out of fashion nowadays, and that’s a shame, because there is a unique feeling to a zoom move that can be very appropriate in certain situations – we’re making things bigger or smaller in the frame whilst keeping the proximity constant.

I want to emphasize here that once we have established what proximity we want and why, and put our camera down in the desired place, the feeling created by that proximity is not affected by our choice of focal length. How I like to think of it, at least for the scenes in which we are clearly in a character’s narrative point of view, is that given the correct proximity for that point of view, different focal lengths create different impressions of focus of visual attention. Here’s a very good example: in Casino Royale, there’s a scene at the airport in which Bond is among the airplanes looking for the villain and spots a dead body next to a fuel truck. We have a shot of Bond, and cut to something like his point of view, and then have a series of shots cut together quickly, where the camera stays where it is and a longer and longer lens is used for each shot (or a zoom is punched in between each shot). This sequence is incredibly effective in giving us the feeling that bond is ‘homing in’, suddenly registering the importance of what he’s just seen, and focusing his attention on it. It’s the feeling of a zoom move but done in a series of cuts rather than a zooming in; it has a starkness and urgency that fits the scene.

Formalizing the above, it could be said that when we are roughly seeking to mimic the human eye with the camera, proximity controls the depth rendition; the feeling of closeness, and focal length controls the impression of focus of visual attention. This is why I think the term ‘normal lens’ is bogus. Back in my photography days one of the big discussion points was whether the 35 or the 50 was the ‘normal lens’. Attempting to find a focal length that matches the human eye is missing the point. We can’t be trying to find a focal length that matches the depth rendition of the human visual system, because we’ve established that depth rendition depends on where we stand and is independent of focal length. So we must be trying to find a focal length with a  field of view that matches the human visual system. But the human visual system is a dynamic entity that adapts to the situation at hand and cannot be thought of simply as a camera with a prime lens on it, fixed at a certain field of view. If we strain to be aware of our entire field of view, we can see at least 180º horizontally and vertically, way more than an 8mm! But if we, like James Bond, suddenly spot a dead body in the distance, our field of view, perceptually speaking, decreases drastically as our attention becomes focused on that area directly in front of us. When simulating the eye, instead of trying to find a ‘normal lens’ we should think about what the perceptual focus of attention would be for the situation at hand. As I said, changing focal lengths is an excellent way of simulating various degrees of perceptual attention on what is in front of us. Bond sees a dead body indicating the presence of the villain: BANG, BANG, BANG, we cut to progressively longer focal lengths, the perfect analogy for the focusing of attention on a small area of our field of view.

Now that I’ve explained the importance of proximity, let’s pitch it against focal length in terms of how working cameramen line up shots.

Assuming we’re on the same camera throughout the show, we have three factors: subject size in frame, proximity and focal length. We will typically choose two of these to our taste, which dictates what the final one should be.  I realize that real world shooting situations are more complicated, that we may be blocking things to the camera, etc, but I believe that most of the time we are, either consciously or subconsciously, tending towards one of two mindsets:

1 )   Knowing the frame size we want, choosing a focal length, and moving the camera into a position that gives us the desired frame size with the chosen lens on.

2 )   Knowing the frame size we want, knowing the proximity we want, and selecting the focal length that gives us the desired frame size from the chosen camera position.

Although both of these mindsets can be executed easily with either a zoom or primes, it is easy to imagine how someone who wasn’t too conscious of how they were making these decisions would be prone to lapsing into mindset 1) if using primes and mindset 2) if using a zoom.

Common wisdom has it that zoom lenses are bad training for a DP,

because you'll have a tendency to keep the camera where it is and just 'punch in and out' on the zoom to reframe the shot. But just like using a zoom lens may reinforce the habit of staying where one is and punching in, destroying any field of view consistency and proximity variation, using primes may reinforce the habit of staying on one field of view and moving around, destroying any proximity consistency and field of view variation.

I’ll give an example. A character is in a room, another knocks and leans against the doorjamb on the other side of the room. We want close-ups of both. Your standard ‘prime trained’ shooter may elect to shoot both on the 50mm, matching camera-to-subject distance between setups. But what if we want to tell this scene from the narrative point of view of the character in the room? We could shoot his close-up on a 35, and just turn the camera around and pop an 85 without moving for the other character’s close-up. The audience will subconsciously read the depth cues, even if the head sizes in frame are exactly the same, and as we cut between the two angles they will subliminally perceive that the scene is being told from the narrative standpoint of the character in the room.

Another James Bond example for the above: In the Macau casino segment in Skyfall, bond spots Severine from the other side of the room. The first time we see her, it’s in a wide from Bond’s position. We then have a close up of Bond, shot on something like a 27 or 32. For all we know they kept the camera where it was, same lens, and spun it around. Then we cut to a long lens two-shot of Severine and her bodyguard from the same camera position as the wide. She turns around and sees him. Cut back to Bond’s close-up. The fact that the camera is kept where it is throughout this small sequence means that we are firmly in Bond’s narrative point of view. Up until now in the film, the camera has always been closer to Bond than it has to Severine, so that when she finally approaches him it is the first time we have a close proximity to her.

Isn’t that a higher level of storytelling with the camera compared to popping a 50 and matching distances? Some will say: “But you’ve introduced focal length inconsistency and camera-to-subject distance (depth rendition) inconsistency!” But what we have maintained is consistent narrative point of view: Bond’s. And that can be a much more powerful storytelling tool than focal length or depth rendition consistency. Of course, you may choose to treat both characters with equal emotional distance, in which case the same lens and same camera-to-subject distance may be entirely the right thing. But you’ve arrived at the solution through a much more considered process than “Here are the close-ups, get the 50”.

Obviously beginning shooters don’t really use a conscious thought process, they just go by eye and try to do the best they can. It is my contention that once you are past being a total beginner at lining up shots, there is a tendency to really focus on the focal length and get into that same focal-length centric mindset that then leads you to say things like ‘The zoom lens forms bad habits’. Zooms can form bad habits, but so can primes. The real perversity of the whole anti-zoom school is that they will usually say that using primes really teaches you about lenses and their unique depth characteristics whereas the truth is that unique depth characteristics belong not to focal lengths but to camera-to-subject distances, proximities, awareness of which might well be brought about better by shooting with a zoom that with shooting with primes! Despite this, I do ultimately agree that a zoom is a bad tool to learn on, because a beginner should always be aware of what lens they are on. Although sometimes they should stay where they are and swing the lens rather than move closer, awareness of exactly what focal length they have on, rather than just tweaking the barrel, solidifies your knowledge of focal lengths. But that knowledge consists of awareness of the field of view and depth of field characteristics of different focal lengths and not their ‘depth characteristics’, which as we now know, don’t belong to focal lengths at all but to camera-subject distances, to proximities.

Just before I go, I need to bring you back to reality. In this article, I have made distinctions stark and situations exaggerated, because things are so muddled up that if you don’t point the conceptual lines out clearly then they’re impossible to see. The reality is that you’re going to be framing for a shot size most of the time, and if you’re experienced enough you’ll know the proximity a certain focal length will force for a certain frame size. So a competent DP asking for a 32 could just as well be thinking “That’s the lens dictated by my choice of frame size and proximity” as “I know this is a lens that’ll give me roughly the correct proximity for my frame size”. The important thing is to remember the difference.

And finally, a disclaimer. By heavily intellectualizing all this, I don’t mean to suggest that the only valid approach on set is an intellectual one. You could have two DP/director teams; one working on a very intellectual level, another on a very instinctive level; both could be capable of producing excellent work. Whatever works for you. I find that the more I think through these things logically off set, the more spontaneous and instinctive I can be on set while still producing good results.

L.

London, April 2013

Look under the rule

In this short piece, I will propose that ‘rules’ in cinematography can be thought of as a vertical ‘stacking’, in which every rule represents an easy way to follow the rule under it, and in which we move from the theoretical to the practical as we move up, but also forget about possible alternative ways of fulfilling the rules below. If this is a valid way of thinking about the topic, then it is important that we keep this ‘stack’ in mind instead of just following whatever rule tops it off. In a phrase, always look behind the rule – or rather, look UNDER the rule!

To demonstrate this idea, I’ll take one of the most common ‘rules’ you are taught when you start out as a DP: Avoid full frontal lighting – that is, don’t put the light right behind the camera.

Why do we normally avoid full frontal lighting? It gives no shape to the face. We want some contrast in there, some variation in brightness. But who’s to say the variation in brightness has to be across the face and not within the wider frame? Given that the audience sees the whole frame, what else can we do to give it some variation in brightness? We can introduce contrast - a brightness difference - between the face and the background as opposed to between the two sides of a face. How do we do that? By bringing the source close to the subject so it falls off rapidly. If we’re on a longer lens, by bringing the light down at a slight angle so it doesn’t hit the portion of background in shot. That can look beautiful. And what use is your ‘no frontal lighting’ rule now?

The ‘Avoid full frontal lighting’ rule arose because it introduces variation in brightness across the face, which is one of the easiest ways to introduce contrast into the frame. So we’ve taken a rule – ‘There must be some contrast in the frame’ – and created a rule above it that gives us an easy way to fulfil it. But in the following the rule above rather than the one below, we forget about other ways of fulfilling the rule below – for example, using full frontal lighting with fall-off to introduce an illuminance variation between the background and the subject. After all, in moving ‘up’ the stack, we’re going from the theoretical to the practical. It is easy to see why any practical solution is only one of a number of possible approaches to a theoretical problem.

Let’s dig deeper. If the rule behind ‘no frontal lighting’ is ‘There must be some contrast in the frame’, then what’s the rule behind that? Why must we have some contrast in the frame? At the basic level, no contrast difference, that is, the same foot-lambert level coming from every bit of the frame, would yield an image with no information. A real subject under completely even illumination will introduce foot-lambert level variation through the different reflectances provided by its texture, so although the same level of incident light strikes every part of the subject, every bit of the subject will reflect a slightly different amount towards the camera. So we have to have (and will inevitably have) some contrast variation. Why do we want a substantial amount of it? Because it guides the eye. It separates elements within the frame. 

So we’ve looked a level deeper, and we have: ‘We must separate elements within the frame’. But again, in going from that to ‘There must be some contrast in the frame’, we’ve forgotten about alternative ways to separate elements within the frame: depth, focus, colour contrast. You can compose wonderful frames lit flat but with colour contrast provided by the elements within the frame, or light them flat but compose them in depth, using geometry and/or focus, or have a colour contrast provided by the light rather than by the elements within the frame.

I’m going to stop here, but I invite you to look deeper. What’s the rule behind ‘We must separate elements within the frame’? Do we always want to separate elements, or is that rule just the easiest way to fulfil a deeper rule? Maybe we want the subject not to stand out. Maybe we don’t want to guide the eye. Maybe we should all aim to get to the point where we only use the full contrast range of the medium when it serves our purposes to do so (see Darius Khondji’s incredible night work in Amour).

I’m going to wager that most cinematographers are going to find this at the bottom of every rule stack: ‘Tell the story’. In a way, the ideal cinematographer holds the ‘rule stack’ in his semi-conscious mind at all times, so he is aware of the rules at the top but is always capable of starting from the bottom in order to produce imagery that serves the story rather than the rulebook.

L.

London, 1 April 2013