How light meters work

Camera light meters try to expose to make all pictures average out to be a middle tone.

A Gray card, a Black card, and a White card
photographed with automatic TTL flash

Gray card with TTL flash

Approximately correct. The small paper was in fact pink, and the couch looks normal. Middle gray is made to be middle gray (all things are made to be middle gray average).

Black card (a fun foamy) with TTL flash.

The metering used much more flash power (exposure) to make the black card be middle gray. It makes couch and pink paper be white.
We must manually apply about 1.5 stops underexposure (-EV compenstion) to make it appear Black.

White card (a fun foamy) with TTL flash

The metering used much less flash power (exposure) to make the white card be middle gray. Makes couch black and pink paper dark.
We must manually apply about 1.5 stops overexposure (+EV compenstion) to make it appear White.

These three cards that were used really were black and white and gray. All propped up on a couch background. THIS scene more nearly averages out to actually be middle gray, more what light meters expect to see.

These are flash pictures metered by the camera's TTL automation. We would get similar results outdoors in sunlight, or in any other light.
This is simply what reflected lightmeters do.

Get some black and white paper (from craft store, or any dark and light objects), and try this yourself, to believe it, and then to expect it.
This is simply what reflected lightmeters do.

It is very good to know this, and to plan for it. More samples and explanation below, like What Light Meters Do near page bottom.

The camera meter is a "reflected" light meter. It points to the subject, and reads the light reflected from the subject. But handheld "incident" meters also exist, which stand at the subject and aim back at the camera, and directly read the incident light falling on the subject, totally independent of the color of the subject. Next section immediately below is about an incident meter.

Beginners assume their light meters ought to always naturally give the correct exposure, but alas, life is not that sweet. Metering is an art.

There are two types of light meters, reflected meters (at left) aimed at the subject from the camera or from the camera direction, which reads light reflected from the subject, and incident meters (at right) aimed at the camera from the subject position, which reads the direct light incident upon the subject (independent of the subject). Cameras are not at the subject, so they can only use reflected meters.

Reflected meters read the light reflected from the subject. A black dress will read much lower than a white dress (so, we must realize what we are metering). Incident meters read the actual light source, and are usually close to correct, because incident measures the actual light, instead of the variable reflection from the subject. Meters like Sekonic L-358 or L-308S offer both metering methods, for continuous light or for flash.

Incident meters tell you how bright the light is. Reflected meters tell you how much of it the source is reflecting to the camera. The article below compares a reflected meter in a camera, with a Sekonic L-308S incident meter.

A Histogram is not a light meter

Four sheets of craft paper (red, white, black, and gray), roughly 20x25 inches, is normal stuff from the craft store. One smaller 8x10 inch photo gray card is added. Paper is standing up against a very black velvet background.

Using a Nikon D300 and one SB-800 flash in an umbrella, to show the difference between incident manual hand metering, and reflective TTL flash metering (i.e., how light meters work).

First picture used Manual flash power which was metered with a Sekonic L-308S incident hand flash meter. 1/4 flash power level reads f/7, ISO 200, in this umbrella at this distance.

First, this top section shows what incident metering can do. A hand held flash meter requires true Manual flash mode (and is incompatible with the Nikon CLS iTTL or AWL commander system). The same manually metered f/7 exposure was used for all pictures in this first section, with the same 1/4 power manual flash power. It provides the correct exposure, no matter what the subject is, no matter how the subject is changed. Then the point (in the lower section) will be to demonstrate that the reflected light meter (TTL metering) in the camera always strives to give a middle gray result. Middle gray is the only one result the reflective metering system is capable of doing. The overall scene is averaged to come out middle gray, which is in fact usually about correct for average scenes (but not every time). This middle gray result is the reflected light meter's only goal. It benefits the photographer greatly to realize this, how light meters work. We watch that result, and always stand ready to provide a little compensation, as necessary.

An incident meter (the flash meter used here) simply meters the light source itself. It does not point at or see the subject, so it does not matter if the subject is white or black, the light meters the same. Black things will come out black and white things will come out white, as we would expect. The subject is not a factor in the metering, except that incident meters must be used at the subjects location (they meter the light on the subject), so they cannot be built into cameras.

But reflected meters are NOT that way. A reflected light meter meters the light directly reflected from the subject (seen from the camera position), and it becomes very important if the subject is black or white (or is very reflective or not so much, etc). Bright things are bright only because they reflect a lot of light. Dark things are dark because they don't reflect much light. The reflected light meter only sees that degree of reflected light, and it does not have a brain to know how bright it ought to be. Its only goal in life is that all scenes are to come out as a middle gray average.

Useful detail to know: Note that in the Photoshop or Elements Levels tool (CTRL L), you can HOLD down the Windows ALT key while dragging the Levels White Point or Black Point slider, which will show you what pixels in the photo that you clipping (In ACR too, Exposure is White Point, Blacks is Black Point). This is how you identify pixels in the histogram. Note that relative height of the histogram data represents area in the picture, the scaled number of pixels with this tonal value.

Continuing use of incident metering, with different subjects

Gray paper, incident meter, with same Manual flash power and f/7, by removing one paper sheet from top at a time. There was only the one original incident meter reading for everything, since the lighting and distance and situation did not change. The subject is not part of the incident meter reading, only the subject's location and lighting is a factor.

Note that it would be very incorrect to "expose to the right" here, on this subject. "Expose to right" means to increase exposure so that the histogram data is shifted to almost touch the right edge. That is indeed often approximately correct for "average" scenes which actually contain something white, but not on special cases like this. The histogram is NOT a light meter. The histogram simply shows where your tones have mapped, but it does not know where those tones should map.

Black paper and many black fabrics do not come out very black if there is light on them. Black velvet is one major exception which is very black (example in the top background of first picture) - the paper is standing against velvet without any separation.

FWIW, to make black look jet black, we must keep the light off of it. Not done here, but to make a black background look black, the usual way is to place it way back (like ten feet behind subject), so the lights are weak on it back there. Or we can move our subject very close to the lights, within a foot or two, which requires we turn the lights down, which makes the distant background darker. Or both of course.

Do not misunderstand me, I am all for "exposing to the right" for most normal average scenes where it might work. But also in that is the need to see and understand what the abnormal subject is, to know when it might work.

White paper, incident meter, with same Manual flash power and f/7. Needs a bit more exposure to be bright white, but the white approaches 248 here, and if trying to show detail in a bride's white dress, be careful, this is about enough.

Exposing to the right would work here, because this subject contains some white which should actually be at the right.

The larger point is that the histogram is not a light meter. The histogram merely shows you the result, where things came out. It does not know where things ought to come out.

All RGB colors have three color components. This red paper has a strong bright red component, but with weaker but distinct green and blue components. For example, this box is a "pure" red, RGB=(255,0,0).

The Nikon and Adobe histograms show the three RGB channels superimposed (on top of each other), but often you can select only one channel to view, to see what each is. Multiple histogram values (like this) for the individual RGB components are common in items of one color, for example like say a manila mailing envelope color (which is three colors too).

White and Black and Gray (the pure tones of them) are unique because their RGB components are equal (no color cast).

Shifting to TTL flash metering by the camera meter

Light meters 101

Next, the flash is simply switched from Manual flash mode to TTL flash mode, and the same pictures were repeated. They are metered from the cameras reflective light meter now, instead of by the handheld incident lightmeter above. REPEAT: The difference now is Reflected light meter instead of Incident light meter. The idea is to show how the cameras reflective light meter responds to these same scenes. The first section above showed how it should look, and this section shows how it infact will come out (because these are NOT the typical scenes expected by the lightmeter). Camera was manual mode M and the flash was TTL flash mode. These are simple normal TTL flash pictures of the paper. Center metering mode was used.

I used the D300 camera's automatic TTL flash metering, working back the other way, replacing papers one sheet at a time. Of course, this subject is about the metering of that paper. Still f/7 and ISO 200, but now the TTL metering is controlling the flash power automatically, so it is no longer necessarily 1/4 flash power level.

Red paper, remetered by TTL flash automation. This is about 2/3 stop underexposed by TTL. We would normally adjust the flash power (flash compensation) to "expose to the right" on the histogram, but I did not do that here on any frame (because these are NOT "average" subjects).

It is a bit underexposed, but from the lightmeter's viewpoint, it is simply more centered, striving for a middle gray look.

Yes, no mistake, this is the white paper, and this is simply how reflective light meters meter white objects. The light meters goal is that everything comes out averaging near middle gray, which is normally correct for mixed average scenes. But this is NOT an average scene. We need two or three stops of +EV flash compensation (overexposure, relative to the metered exposure) to bring this paper up to white. And if we are shooting some scene like this, that is our job to do it. We know what the scene looks like, and when this will be necessary.

The only one thing unique here is that this subject is 100% white (full frame is 100% white). It is not a mixed scene containing many tonal values in it which would average out to a darker value than pure white. But if your scene is "largely" white (like a white background wall), or of course a bright sky background, or simply any highly reflective scene, you will get the same underexposed result.

The light meter does NOT know this subject is white paper. It has no clue that it should come out white. The meter is just some dumb silicon chips, it has no human brain. It only sees some reflected light, and its only goal it so make all scenes come out middle gray. The metering told the TTL flash to use less power to keep the white down at middle gray. This blank white paper is of course nowhere near an average or typical scene. It may be an extreme case, but it is an attempted photograph.

Preaching, but the photographer's brain knows what the scene is and looks like, and he learns to expect this natural result, and he must plan on compensating it as necessary. Beginners unfortunately expect their light meter to always give the correct exposure automatically. But the real world does not work that way, and they are going to be very disappointed if they don't learn this fact. Middle gray is simply what reflective light meters do, in all cases. The middle gray term here just means middle intensity, it might be red or green or blue. Most typical or average scenes do come out about right (the numerical average value of all their pixels is in fact near middle gray, meaning, middle intensity) - see large yellow box details below. But our human brain always has to pay attention, and help it now and then, because the light meter has no brain, and no clue what the subject is, or how it ought to look. Its rule is that the average ought to look middle gray.

Black paper, remetered by TTL flash automation. Yes, this is the black paper, and this is simply how reflective light meters meter black objects. The light meters goal is that everything should come out near middle gray, which is normally correct for mixed average scenes. But this is NOT an average scene, and the metering told the TTL flash to use more power to move the black up to middle gray.

"Exposing to the right" would definitely be the wrong thing to do for this black scene. We need a significant -EV flash compensation (underexposure) to move this paper down to black. It is only our human brain that can recognize that. Our human brain also understands that the camera light meter will make it be middle gray.

Film was much more difficult, since we had to handle this properly before the picture was taken. Digital is much easier (instantaneous feedback), we simply look at our result and compensate as necessary to redo it.

Gray paper, remetered by TTL flash automation. Meter still tries to make all things come out middle grey. The subject need not be gray for that to be true. This is a lighter gray paper, but it comes out middle gray here.

In fact, the point here is, it does not matter what color it actually is, the reflected light meter will make the overall scene average out to middle gray. Or a middle shade of red or a middle shade of green, whatever. And this is normally very near correct for an average scene containing many different elements, but is not so great for special scenes, like the proverbial black cat in a coal mine, or the white polar bear on the snow. You need to be aware of this when spot metering too (the spot metered area will come out middle gray too, which if you are unaware of this, may or may not be your intention).

Exposing to the right would be incorrect here too. This one partially more to the right maybe, but fully to the right is only true of scenes which contain white or bright, those which ought to be fully to the right. But exposing to the right is often suitable for average mixed scenes.

Added gray card. This does not add much except to show the gray paper result is in fact near middle gray.

All of these automatic TTL exposures come out a little below histogram middle point, which the camera considers to be the correct place for gamma encoded middle gray. The histogram midpoint is not middle gray (see Techie Part, next).

The meter is working right. If we had a wide scene containing a wide range of tones, like sky and mountains, or even a human portrait, this overall average middle gray result would likely be near correct. This paper is a special limited test, and is NOT a typical scene, and we cannot expect a typical result. So we photographers must pay attention to what we are photographing. Knowing this fact (that light meters aim for a middle gray result) will help you to know what to expect BEFORE you shoot.

Make no mistake, this usual middle gray average is the only hope we have, this is simply how things work. It requires a human brain to comprehend "this is white and it should come out white". Or "this is black and it should come out black". This brain power does not reside in the light meter.

Histograms:

Histogram bar charts are shown 256 pixels wide on video screens, and thus always represent 8 bit color (our monitor simply cannot display a 12 bit histogram 4096 bits wide).

Histograms can show only the Red or only the Green or only the Blue channel, but they typically show a three channel combined "gray" value (another reason I refer to "middle gray" when it is red.)

There are two types of these combined histograms:

Threre is a "RGB" type (Adobe and Nikon are examples, shown above) where the three RGB channels are simply shown superimposed on the gray histogram, so that any red pixels are still shown at the red pixel values. They may not be distingishable there as such, but they do not shift position. This is the common case.

And there is the "Luminosity" type, where the RGB channels are first converted to gray scale data using the NTSC televison standard:

Luminosity = 0.3 R + 0.59 G + 0.11 B

Example, a pixel of RGB value (200, 100, 50) would map at 200x0.3 + 100x0.59 + 50x0.11 = 124 gray.

This single resulting value is where this RGB color would appear in a grayscale conversion image, or on B&W film. It is intentionally based on the human eye being more sensitive to green and less sensitive to blue. Photoshop also has this option to show, but it is not its default. It is the default in a few programs (typically older scanners).

Techie Part

OK, hang on, here we go...

The term middle gray is an older term coming from grayscale in B&W photos. Today, that term usually simply means the middle intensity value in RGB color images, whatever that representative color is. All images, when all the pixels are numerically averaged together (blue sky, snow covered mountain, green grass, dark shadows, shiny faces, etc), have one single representative color value result (Photoshop menu Filter - Blur - Average). This is frequently infact a middle tone, which is "middle gray" if converted to a B&W image. Light meters make that assumption, expecting that typical result, which is often near correct for typical average scenes. The light meter samples some part of that total area, a tiny area for a Spot meter, a much larger central area (and a bit of all of it) for a Center meter, and an algorithm unknown to us for a Matrix meter. But some area is considered important and assumed to average "middle gray". The goal of the reflected light meter is to try to position that value at the middle point, called middle gray here.

Middle gray should be somewhat lower than the histogram midpoint. A gray card is said to reflect 18% of the light that hits it. So this result should be at 18% of the histogram too (certainly less than 50% - it is 18% after all). And it should of course be at 18% in the linear RAW camera sensor data. That is the meaning of 18%. This is not blasphemy, 18% means 18%. Not 50%.

The light meters instead work to about 12%. Whatever scene they see, they assume its average value ought to be a middle gray tone, which they try to set it to be 12%. For reasons explained next below, that is why the middle of the above histograms is a bit left of center (the above are all single tone special cases). Kodak says if you use their 18% card for exposure metering, to then open up aperture 1/2 stop, which then converts as if it were a 12% card (See more about 12%, and about lightmeter calibration constants).

However (a biggie), by the time we can ever see it, the image data has been gamma encoded, and the histogram data shows gamma encoding too, and so we see 18% middle gray at 255x(0.18 ^ 1/2.2) = 118 on the histogram, which is just a numerical coincidence that it is a bit left of the center 128 value.

There is NO RELATIONSHIP between 18% gray cards and histogram middle point. 18% is 18%, it is NOT 50%. The 18% gray card may appear middle gray to our logarithmic human eyes, but it is 18% to the camera. Midpoint may be what the eye perceives, but it is NOT what the camera sensor or film sees. This 18% value has no special meaning to the linear camera sensor, for two reasons, 1) the camera does not have a human eye response (our human eye will see it later), and 2) because the later gamma encoding confuses everything we see.

The term linear here is in the math sense (graphs as a straight line), but in video usage, linear also mainly implies "not yet gamma encoded", after which the data is exponential, no longer linear (and which represents 100% of the RGB image data that humans will ever see, by convention).

Said another way, white fully exposed to be at the 255 right end of histogram, but then we change things to expose it one stop less (one stop underexposure now), that 255 data point will move down to the 128 middle point (linear). In the linear camera sensor, 50% is the meaning of one stop. However, by the time we can see this data, it has been gamma encoded, and we see that 50% at 255x(0.5 ^ 1/2.2) = 187 on the histogram. We see one stop underexposure to be at 187 (about 73% of 255), and not at 128 (50%).

The 18% gray card should come out at 18%, or at 117, which is about 46% of the 255 range after gamma encoding, which is near enough the midpoint that it confuses us, but the coincidental reasons are quite different. The midpoint 128 has been moved up to 187 at 73% now anyway. The ANSI standard for light meters is 12% reflectance, not 18%. Kodak says metering from their 18% gray card should be compensated with 1/2 stop more exposure (which converts it to 12%). All RGB images are gamma encoded, we never see anything else. See Adobe PDF Link. Their chart shows gamma stops as five equal steps, which is only a crude approximation. Perhaps that bypasses their need to explain that the actual stops (of underexposure, down from 255) computed for gamma 2.2 are at:

0 stop = (1.0 ^ 1/2.2) = 1, x255 = 255    100%
-1 stop = (0.5 ^ 1/2.2) = 0.73, x255 = 187    73% of 255 full scale
-2 stops = (0.25 ^ 1/2.2) = 0.54, x255 = = 137    54%
18% Gray card = (0.18 ^ 1/2.2) = 0.46, x255 = 117    46%
12% Gray card = (0.12 ^ 1/2.2) = 0.38, x255 = 97    38%
-3 stops = (0.125 ^ 1/2.2) = 0.39, x255 = 100    39%
-4 stops = (0.0625 ^ 1/2.2) = 0.29, x255 73    29%
-5 stops = (0.0312 ^ 1/2.2) = 0.21, x255 54    21%
-6 stops = (0.0156 ^ 1/2.2) = 0.15, x255 39    15%
-7 stops = (0.00781 ^ 1/2.2) = 0.11, x255 29    11%
-8 stops = (0.00390 ^ 1/2.2) = 0.08, x255 21    8%

What Light Meters Do

NOTE: My Middle Gray term here does not necessarily mean gray. When discussing exposure, it could be middle red or blue or green (just a mid-tone tonal concept, between darkest and brightest), and it refers to the equivalent tone value present in the (gray) histogram. Photography and its terms used to be black and white (grayscale). Most histograms still show the equivalent one grayscale value (instead of three RGB values). The term "middle gray" usually simply means the grayscale equivalent... even if it might be red or green, etc.

Camera TTL light meters do not have human brains to know what we are pointing them at. The meter cannot recognize a snow covered mountain from a black cat. The meter just sees some pixel values with no clue what it is or how it ought to look. Human brains know the expected result instantly, but the meter cannot recognize the subject, and does not know if the subject should be black or white or pink, or if it should be light or dark. They only see some degree of reflected light intensity coming back through the lens. But it turns out that most average scenes contain a range of various degrees of brightness distributed over the full width of the histogram (a shiny face and dark hair for example, or sky and mountain for example), and specifically, all the tones in most scenes typically do average out to be near a middle gray overall.

Reflective light meters simply always assume every scene averages out overall to middle gray, and the light meter's goal in life is to make the picture exposure average out to be middle gray. Normally this is correct for average scenes, which typcally do in fact average out to be about middle gray, but there can still be exceptional non-typical scenes when this fails somewhat. Humans can recognize those, light meters cannot.

The design of any reflective light meter is that when metering something all black, or all white - or anything, whatever it might be - all scenes are supposed to come out all middle gray. This is how simply reflective light meters work. It may not be the way humans think however. :) Humans want to imagine some "correct exposure", but light meters only know middle gray for all scenes. Reflective meters are designed to measure an overall composite scene which OUGHT to average out to middle gray, which is often true of many things, like landscapes or portraits, but which may not always be the case. If you meter bright things, they will meter bright and exposure will be reduced to be middle gray. If you meter dark things, they will meter dark and exposure will be increased to be middle gray. It is pretty good to know this. This is shown above, and again below.

Spot meters are reflected meters too. So when we take a spot reading of the subjects face, it does NOT mean we get a correctly exposed face. We get a middle gray face, every time. Not aways what you want. It can however work if you meter something that ought to be middle gray. Spot meters are not designed to give a good exposure of a scene. Spot meters are designed to let the photographer prowl around in the scene, measuring bright things and dark things, to determine the range of the scene. Center and Matrix metering have much to be said for them, averaging the overall scene to middle gray.

Incident light meters (flash meters are one example) are a big deal in the studio, and the concept is more about "correct exposure". They are instead are pointed the other way, at the camera instead of the subject. They instead directly measure the brightness of the light incident on the subject (at the subjects location), and give a corresponding exposure so that middle gray things should come out middle gray, and white things come out as white, and black things come out as black. This is a much more consistent reading but cannot be implemented on the camera. Incident meters must meter from the subject.

Matrix metering

Here is a third sample using Matrix metering this time (months later, always the same results, this is simply "how it works"). This meters these white and black cards again (at right). D300 with hot shoe SB-800 bounced on ten foot ceiling. TTL flash mode of course. 100% frame, no cropping. Same result, this is how reflective light meters work. Reflected meters only have one way they can work - with zero knowledge what the subject is, or how it ought to look, they can only always strive for a middle gray average.

We certainly need to know this, so this is one of those things which you should try at home. If you are metering a rather light scene (meaning, one which reflects much if its light effectively), it is going to come out underexposed. If you are metering a rather dark scene (meaning, one which does not reflect so much), it is going to come out overexposed. Both will come out averaging middle gray. Our human brain can recognize this and plan ahead. The least we can do is to look at the result and compensate. What we ought not do is to act surprised. :)

Black card, Center metering	Black card, Matrix metering.

White card, Center metering.	White card, Matrix metering.

The point is, this is simply what reflected light meters do... tries to average every scene to middle gray (meaning middle brightness). This exposure will usually come out near correct for average scenes. But it can guarantee underexposure for bright scenes, and overexposure for dark scenes (speaking of the "color" and "reflectivity" of the scene, not the light that is on it).

This example uses camera TTL flash metering, but you would get similar results with tungsten or sunlight... any reflective meter strives to give middle gray result for any scene it sees. Most typical scenes do in fact average out about at middle gray, so it normally works out fine.

We need to understand this, regarding the amount we compensate the flash. The light meter has no idea what the scene is, or how it should look, it only sees the overall average, which it assumes should be middle gray. But we humans know what the scene is, so it becomes our job to correct. When metering white things (say a wedding dress), you probably need +1EV compensation to keep it white. If metering black things, then probably -1EV compensation to keep it black. This depends, so check your results.

Seeing this here is one thing, but to really grasp it and make it yours, you should of course repeat this simple experiment yourself... to actually "see" it happen.

How Light Meters Work

A Gray card, a Black card, and a White cardphotographed with automatic TTL flash