Ever since my days working in film I’ve tried to find some way to equate my sense of brightness to real world exposure. I learned to do this well with film, but HD has been a mystery… until now.
On a recent Alexa shoot I gave up on the idea of running to the monitor and scrutinizing the waveform for every shot. I had to work too fast and the lighting had to be perfect. Instead I relied on my incident meter. This is unusual for me because I’ve never trusted incident meters.
Reflected meters are more my thing because they appeal to my anal retentive need for total accuracy. As a camera assistant I became obsessed by the Ansel Adams Zone System and as I started shooting film I learned that I could expose images very, very reliably. My dailies generally turned out exactly the way I wanted them, although I spent more time on set than I should have peering through my spot meter trying to analyze everything I saw.
A wise old DP imparted some advice to me: “Don’t read everything. Just read the brightest highlight, darkest shadow and middle gray, and if they all fit within film’s exposure limits then put the meter away and shoot.” What he meant was to spend less time collating data in my head and instead rely on my eyes and creative intuition when lighting and exposing. In short: light so the set looks pretty, check to make sure the exposure will reproduce on film the way I want it to, use the meter only to determine a shooting stop, and then shoot.
One DP I knew simplified things considerably by simply exposing everything important in the scene between -3 and +3 reflected stops from middle gray. The unimportant highlights and shadows he set by eye and didn’t worry so much about. This system also worked well for me. At the time I didn’t understand why, but I do now. More on that below.
Over time I learned to light by eye and then find something in the set that that looked close to middle gray and meter that to determine a shooting stop. Then I checked the brightest highlight and darkest shadow to see if they would hold. This method worked very well, although it was hard to pre-light a set to a specific stop this way as I tended to light first and determine an exposure later.
I had a hard time equating different shades of gray to exposure values. The Zone System contains 11 steps, from 0 to 10, and each step is one stop different from the next. The idea is that if you know, by eye and memory, roughly what each step looks like you can use a light meter to expose a real world object at that value. For example, if I wanted a wall in the background to go dark I would take a reading and then underexpose it by a couple of stops. A white wall that reads T2.8 on a spot meter will look gray or dark gray when shot at T5.6 or T8.
Using words helped me determine what those values were, as it was hard for me to look at an object and say for certain that it was Zone 3 or Zone 4 (both of which are dark tones) or Zone 7 or 8 (light tones). I used the following terminology to translate tones to values in my head:
Zone 10: Overexposed beyond visibility.
Zone 9: White with texture. (I’ll still see some detail if there’s detail to be seen.)
Zone 8: Dark white. Still bright, not yet gray.
Zone 7: Light light gray.
Zone 6: Light gray.
Zone 5: Middle gray. Just gray, not light gray or dark gray.
Zone 4: Dark gray.
Zone 3: Dark DARK gray.
Zone 2: Light black.
Zone 1: Black, possibly with texture and detail.
Zone 0: Underexposed beyond visibility.
I can also translate this in reflected light values:
98%: Barium sulfate, the brightess diffuse (non-specular) white possible. Zones 7-8, about 2.5 stops brighter than middle gray.
72%: Bright skin tone. Zone 7, two stops brighter than middle gray.
36%: “Normal” caucasian flesh tone. Zone 6, one stop brighter than middle gray.
18%: “Middle” gray, the average of black and white. Zone 5.
4.5%: About the darkest gray an inkjet printer can print on a matte surface.
2%: The darkest black you’ll typically see in a diffuse surface that isn’t black velvet or a deep shadow. Zone 2.
By the way, this is how Sony tells us the maximum brightness a gamma curve can handle. A “480%” gamma curve, for example, can render detail in objects stops brighter than middle gray if we do a little math and know that doubling a brightness value is the equivalent to an increase in exposure value by one stop: 18% -> 36% -> 72% -> 144% -> 288% -> 576%, so a 480% curve will clip between four and five stops above middle gray.
Any object that reflects more than 100% is a highlight in a shiny surface, so Sony is also telling you that a 480% curve gives you about two stops of latitude to hold detail in those highlights. (Sony’s current log curve in the F55 is rated at 1300%, or slightly more than six stops above middle gray.)
When I tried to use my reflected light meter system in HD I found myself consistently confused. My word games no longer worked. Early HD cameras clipped at 2-3 stops brighter than middle gray. The camera’s ISO frequently didn’t match my meter’s ISO. Eventually I came to rely on the waveform monitor as I could use it to dial in very precise exposures… but that required seeing the subject in the shot the entire time in order to see how bright objects and people actually were. This isn’t a very efficient way to work as often we don’t get to see how actors look until they step in for an actual take. (Features and TV shows often hire stand-ins to help with lighting, but most of my work is in commercials and we never get stand ins.)
As HD cameras evolved I periodically tested my reflected meter against newer cameras. Some worked marvelously, some didn’t. I learned I could expose RED ONEs perfectly using only a reflected meter, while Arri Alexas in Rec 709 mode disagreed with nearly all of my reflected exposure calculations. Sony cameras changed their responses depending on the gamma curve. It was relatively easy to find the under- and overexposure limits of the camera but much harder to determine how the shape of the gamma curve affected highlight and shadow compression. It’s helpful to know that a specific camera slips at, say, six stops brighter than 18% gray, but that tells me nothing about whether this particular curve boosts shadows or compresses highlights.
Film stocks had similar characteristics but I really only had to know a couple of them really well. Now every camera is a hundred different film stocks with a built-in lab whose sole employee is me. (This is not the case with log curves, which tend to be much more consistent between cameras.)
I became frustrated. I didn’t want to rely on the waveform monitor anymore, especially since the advent of raw and log cameras meant producers didn’t want to rent them anymore even on shoots where we weren’t shooting raw or log. The scopes built in to most on-set monitors are okay but not great, and they still require me to actually see actors in the frame to set an exposure. In a fit of desperation I decided to dig out my old incident meters and give them a try. This didn’t, initially, appeal to me.
I’ve never trusted incident meters. They told me how much light was falling on a scene but nothing about what was actually happening in the scene, so I found them a bit useless. One DP taught me that they were great for flesh tones as an incident meter is not going to compensate for differences in tonality, which makes sure that the brightness values of different kinds of skin actually come through on film accurately. (The HD equivalent of not doing this is setting everyone’s flesh tones to 60% on a waveform monitor, for example.) They’re also great for roughing in lighting before actors arrive on set, but an incident meter won’t necessarily tell me how bright the back wall of a set is.
Then I got into the test chart consulting business and learned some interesting things about printing technology.
The brightest diffuse (non-shiny) white material in existence is considered to be barium sulfate, which reflects around 98% of the light that hits it. If we know that middle gray reflects 18% of the light hitting it we can quickly suss out that 100% reflectance is about 2.4 stops brighter than 18% gray. This means that the brightest diffuse (non-shiny) object I’d ever run across in nature is 2.4 stops brighter than 18% gray. I’d never known that before. In the grand scheme of things, diffuse white just doesn’t seem like it’s terribly bright at all. 2.4 stops? That’s it? Huh.
What’s equally surprising is to know that the darkest matte black a printer can print is about 2-2.5 stops darker than middle gray. A glossy black might pull that value down to 3 stops.
When that one DP told me he basically exposes to keep all the important information within a +3/-3 stop range from middle gray I had no idea that’s where most values in nature, short of specular highlights and really dark shadows, actually fall.
Why is this important? Well, remember when I told you that I didn’t trust incident meters because they didn’t say anything about how bright objects were? Well… if I take an incident lighting reading and I know that the normal range of brightest diffuse white to average dyed or printed black is about +2.4/-3 stops from middle gray, I now have a good idea of where values in the set fall just by looking at them. Black dress? Probably 3 stops below middle gray. White shirt? No more than 2.4 stops above middle gray. Knowing then that my camera can handle, say, +6/-8 stops above and below middle gray I can use an incident meter and know, fairly accurately, where values will fall without actually seeing anything through the camera.
I can also shift them around and know how far I can push them. If my camera can see detail in something six stops brighter than middle gray then I know that I can overexpose a white dress by three stops, according to my incident meter, and still see some texture. This is handy in case the actress walks past a practical light source or a sunny window.
Another good example is flesh tone. If I know that flesh tone is generally between a half stop and two stops brighter than middle gray, and I know that something that’s black in the real work probably isn’t darker than three stops below middle gray, I can quickly figure out how far down I can push someone’s face in order to make it barely visible in the background of a dark shot in a horror movie. If the person has really light skin I can assume it’s probably around two stops brighter than middle gray, which means I can underexpose it four stops (-2 from middle gray) and still see something. If I want to blast an actor with light and make their skin glow then I can quickly determine that I need to add three stops of light to a light-skinned person versus five stops to a dark-skinned person.
I can do this with a spot meter as well, but the spot meter requires an actor to stand in whereas an incident meter doesn’t. If I have an idea of how bright the actor will be based on my knowledge of how reflective things are in the real world, I can dial my lighting in really quickly without needing a stand in.
At the beginning of the article I spoke about using my incident meter on a recent Alexa shoot. Due to a quick turnaround we shot in Rec 709 WYSIWYG mode and I’ve had a hell of a time getting my reflected light meter to work reliably with Alexa in that mode as its Rec 709 curve is really contrasty compared to what I was used to in film. Items that looked to be +2 stops from middle gray looked way too bright when viewed on a monitor. The highlights still held beautifully but the mid-tones felt way too crunchy.
When I started using my incident meter I discovered this problem largely went away. The incident meter doesn’t care about how bright objects are, only how much light falls on them, so rather than try to place values in the scene on specific zones I lit by eye and read the most important part of the shot with my incident meter. I got very good results. It still freaks me out a little as I really want to predict the tonality of everything in the shot before looking at a monitor, and my trusty reflected meter doesn’t really do that well with HD cameras due to the variations in gamma curves, but trusting one or two readings with my incident meter and simply judging everything else by eye is working very well for me now.
It’s tough because I have to turn off part of my right brain and turn on part of my left, but part of working with a light meter means learning to simply trust it. Once, when doing a job that required rappelling down a 200′ drop for a shot, our mountaineering supervisor told me, “Some people learn to trust the rope. Others don’t.” For me, learning to trust an incident light meter is like learning to trust the rope. It’s not giving me all the detail that I’ve come to expect from a spot meter, but… somehow it still works. And knowing the real world brightness values of a few key things I now have the confidence I need to estimate what those values are without actually reading them.
There’s a certain kind of freedom that comes from lighting a set more-or-less by eye, holding a meter in the light source where objects should look normally exposed, putting that stop on the lens and seeing a beautiful image on the monitor… and knowing that everything I shoot within that set will be consistently exposed.
People still look at me a little funny when I pull out a light meter on video sets, but my dailies are solid and I don’t have to look at a waveform monitor much at all. That makes me both fast and good, and those two qualities keep me happily busy.
About the Author
Director of photography Art Adams knew he wanted to look through cameras for a living at the age of 12. After spending his teenage years shooting short films on 8mm film he ventured to Los Angeles where he earned a degree in film production and then worked on feature films, TV series, commercials and music videos as a camera assistant, operator, and DP.
Art now lives in his native San Francisco Bay Area where he shoots commercials, visual effects, virals, web banners, mobile, interactive and special venue projects. He is a regular consultant to, and trainer for, DSC Labs, and has periodically consulted for Sony, Arri, Element Labs, PRG, Aastro and Cineo Lighting. His writing has appeared in HD Video Pro, American Cinematographer, Australian Cinematographer, Camera Operator Magazine and ProVideo Coalition. He is a current member of SMPTE and the International Cinematographers Guild, and a past active member of the SOC.