LIGHT METERS: What are Incident Meters Good For, Anyway?
Incident meters don't tell you how bright things in your shot are, but it's surprising how handy they can be if you know a few simple facts about light and reflectivity.
By Art Adams | November 13, 2012In film school we're taught that incident meters are best at averaging light. The method most often taught is to measure key + fill by aiming the ball at the camera. This assumes that key and fill are both fairly frontal, which is not always the case.
In this case, with a key of 100fc and fill at 25fc, and a camera or film stock rated at ISO 400, we'll end up with a stop of T5.6 +1/4. (The next stop up from 5.6 is 8, which requires twice the light at 200fc, so one quarter of the way from 100fc to 200fc--25fc--is 1/4 stop.) If the light meter is held in front of a face then the part of the face that is brightest, in that it is being hit by both key and fill, will be rendered fairly accurately at T5.6 +1/4.
It's generally best to denote stops in positive terms, such as T4 +2/3 or T8 +1/2. If we start throwing in negative values, like T8 -1/3 or T4 -1/2, we take the risk of our assistant misunderstanding us and setting the stop incorrectly. If we ALWAYS speak in terms of a stop PLUS a fraction then that removes a significant source of error. Otherwise we risk our assistant mishearing "T5.6 -1/4" as "T5.6 +1/4", resulting in a half stop underexposure.
One of my mentors told me years ago that the human eye can detect a brightness change of 1/3 stop but it has a hard time seeing smaller changes. I've met DPs who will dial in their exposures to within 1/10th stop, but I think that's overkill. I stick to full stops, 1/2 stops, 1/3 stops and occasionally 1/4 stops. Often 1/3 stop is close enough. It's easy to get caught up in the details, so making your T-stop decisions coarser will make your life a little bit easier.
It's important to know that T-stops are different from f/stops. F/stops measure the actual diameter of the aperture, or iris, in a lens whereas T-stops ("transmission stops") compensate for light loss as light passes through a lens. F/4 on one lens might really be T2.8 +2/3, because the T-stop accounts for the fact that about 1/3 stop worth of light is lost between the front of the lens and the film plane or sensor.
T-stops are measured and set at the factory and are always more accurate for exposure than f/stops, although f/stops are always more accurate for determining depth of field. Nearly all modern lenses are calibrated in T-stops.
While we're talking about techie things, it's always best to set the lens aperture by opening it up all the way and then closing it down to the intended T-stop. If the iris leaves have any slop in them you'll be erring on the slightly open side by doing that, and a little more exposure is generally better than a little less.
The incident meter merely averages the light hitting its dome, so it can be deceptive if we always read it with the dome facing the camera, as taught in film school. Here's an example you can try with your meter:
Imagine a situation where you're in a dark room lit by one very distance light source that's at 90 degrees to the subject. Aiming the dome of the incident meter at the camera means that one side is lit by 100fc of light, and the other side is completely dark. The meter will average both sides of the dome and tell you that the proper exposure at ISO 400 and 24fps is T4, because half way between 100fc and 0fc is 50fc.
If you then use that exposure then half the subject's face will be one stop brighter than normal and the other side will be black. The meter won't tell you what the "proper" exposure is because that's a creative choice. It'll give you whatever information it can, but you have to determine how to use that information creatively.
For example, if someone is sitting by a window it may be appropriate to overexpose their skin by a stop, especially if they are sidelit, because we'll often accept overexposure as long as only part of the subject is overexposed. Our eyes do this naturally by averaging the amount of highlight and shadow within our vision and showing us a decent compromise between the two, usually by overexposing the bright side a little bit while allowing the dark side to remain a little dark.
In that case it's useful to know that flesh tone usually falls somewhere between middle gray and two stops brighter than middle gray. If the person near the window has very light skin and you're shooting with a camera that doesn't have a lot of overexposure latitude you may not want to overexpose them at all, as you can guess that their pale skin is already at +2 stops reflected and another stop of exposure may cause their skin to clip. If you're using a camera with a lot of overexposure latitude, say five stops above middle gray, overexposing the person by two stops won't clip their skin tones, although they may become flat and compressed by the top of the gamma curve.
You still need to know, basically, how your camera responds to light with whatever settings you have dialed in, but you can rough in lighting very quickly without running around and spot metering everything in the frame. Too much data can get in the way of the creative process, so it's often better to light by eye first and then use a meter to see if your eye and the camera will see the same thing. If not, then you can adjust. Over time you'll get a sense for how to light for that particular camera and your setups will come together much more quickly.
If a scene is dark and moody then it may be appropriate to create this feeling by reducing the level of fill light dramatically or by underexposing the shot overall, although overall exposure can look "wrong" unless it is justified by the appearance of something bright in the scene. A bright highlight tells our brains that the dimness isn't a mistake, as if our brain wants the darkness to be balanced by a highlight so it can "average" the scene the way a meter does.
The important things to remember are:
Caucasian flesh tone is normally one to two stops brighter than 18% gray. Dark skin is normally 18% gray or, occasionally, one stop darker--although that may be a bit too dark for some cameras and may require a little lighting boost.
The brightest matte surface in the shot that is being lit by the same amount of light as your light meter can only be 2.4 stops brighter than 18% gray. (This does not apply to shiny or reflective surfaces, which can be much brighter.) That brightness level appears as white, but probably not a really bright white depending on the camera. A camera with crappy dynamic range will probably put this value close to 100% on a waveform monitor, while a better camera will probably place it at 80% using a WYSIWYG gamma curve or even 60-70% on a log curve. That leaves lots of room for bright highlights. (Sony's S-Log log curve places an object with 100% reflectance at around 62% on a waveform monitor.)
The darkest matte surface in the shot that is being lit by the same amount of light as your incident meter is generally 2-3 stops darker than 18% gray, unless it is made of a material that is known to absorb light such as black velvet. (This does not apply to shadows, which can be much darker.)
Incident meters can allow a more intuitive way of working than spot meters because too much analysis can often result in confusion and distract from the point of lighting a set, which is to use technical tools toward a creative end. For example, with a bit of experience it becomes fairly easy to light a set by eye so that it looks good and then hold an incident meter in the area where you want tonal values to look "normal." As people move around a set they naturally become brighter and darker, and you can take control over that by using an incident meter to determine where those "normal" spots are and make sure the "non-normal" highlights and shadows fall within the camera's dynamic range.
For example, a person with average caucasian skin will normally appear one stop brighter than 18% gray. If they wander over to a window and catch 1.5 stops more light then we know they'll be at 100% reflectance, which is effectively white but still with texture. Modern cameras can hold at least that much overexposure; most can hold more and some can hold a LOT more.
The trick with incident meters is that you have to know the brightness of the thing you're measuring. For example, if one person's skin tone is one stop brighter than middle gray and another's is one stop darker than middle gray, an incident meter will tell you how to expose them so that they look roughly the way they do in real life: one has much brighter skin than the other. It doesn't actually tell you how bright their respective skin tones are, it just gives you an exposure that will make them look "correct" within the 5-6 stop range within which most matte surfaces will reflect light.
That's why it's handy to know that the dynamic range of the average Chroma Du Monde or MacBeth chart is easily within range of the average camera, and that exposing that chart with an incident meter will result in an image that looks "correct." Not artistically "correct" but technically "correct." You'll see all the patches, and none will look terribly over- or under-exposed.
The artistic part is knowing how far to shift things away from the meter's exposure in order to capture different looks. That takes a bit of practice.
For example, filling at two stops below the T-stop set on your lens means that average caucasian skin will drop from 36% reflectance to 9% reflectance, or one stop darker than 18% gray. That's still perfectly visible, and in fact it's probably too bright for most dramatic purposes. Fill levels of 2.5 to 3 stops under the lens T-stop often look very nice indeed. (Our brains like contrast as we're programmed to appreciate changes in our visual field over sameness. Flat white walls are boring, but walls textured with shadow are quite compelling.)
Filling a person with dark skin at the same light level may not yield desirable results.
Note that I say "two stops under the lens T-stop" instead of "two stops under key" as many people phrase it. The key is variable. Sometimes you'll expose it a bit hotter than normal if someone is standing very close to a light source, like a window. Sometimes you'll expose it darker, for a moodier feel. There is no rule as to how the key light--whatever it may be, or however it falls across the set--should be exposed.
I also never talk in terms of key-to-fill ratios, because those tell us how bright the fill light is in relation to the key but they say nothing about how I've exposed the scene. If I expose the key at T4 and fill at T1.4, that's technically a key-to-fill ratio of 8:1, but it doesn't tell you whether I've set the T-stop on the lens at T4 or T2.8 or T2 +1/2, all of which will yield different looks.
The best part about using an incident meter is that you can determine the exposure for a position in a set without having a person standing there. I don't know about you but I rarely get to work with stand-ins anymore. An incident meter helps me dial in the exposure for the parts of the set where I know an actor will stop and deliver a line--and I can do this long before the actor gets there.
If actors have dramatically different skin tones then I'll put a piece of camera tape on the back of my meter and make notes as to how bright or dark to light them, based on how the first few shots of the day look on the monitor, and I'll refer to that over the rest of the shoot to rough in lighting before the actors step in.
Turn the page for a final trick or two and a closing thought...
Get articles like this in your inbox: Sign Up
artadams - Wed, May 22 2013 - 8:34 pm
New article: "House of Cards" ironically restores my faith in humanity, at least when it comes to composition. http://t.co/jr0NH0yLSd
artadams - Wed, May 22 2013 - 8:19 pm
I don't normally post commercial links (unless I shot them) but I can't resist. Three words: "Bomb-defusing sloth." http://t.co/doK8T9q2yL