Ant Smith


Exposure Operations

By Christ but photography's easier for you kids today than it ever was for me!

In this article you will read about

  • The importance of good exposure

  • The nature of tones and the subjective act or recording them

  • What a Histogram is and how to use it

  • What under and over exposure looks like

  • How to use exposure compensation.

Probably the single most significant aspect of the act of photography is the exposure itself. This is why we take photographs in the beat between breaths. The moment of it certainly, but oft times more subtely and with greater endurance, the tonality the exposure brings.

The moment and the tonality.

Black and white darkroom print of the sultans elephant

Sure, geometry and composition matter too (and we'll get to those), but tonality is where the image can become sumptious. So that we want to lay in it for some time.

Or not, depending on what you're trying to do with the image. But let's pretend for awhile that what we want, is to make something that is sumptious.

You know, it is in exposure where the photographer dances. Creating a print from a negative in the darkroom requires an exposure to be made onto the photographic paper. Ansel Adams would execute his darkroom performance in as many as 20 steps per print, in each his hands dancing to sculpt the cone of light that his enlarger shone. You can feel it yourself when your hips wiggle as you twist about trying to get the reflections in the shot under control.

So lets dance.

I'm going to mostly talk in terms of Black & White. That's how I learnt and it never did me any harm. To be honest I'm at heart a B&W photographer. Even though the greatest majority of my work is in colour, it always feels just a little like a sop to the clamouring masses not to print it in B&W. Not too much though. Just the odd 'Hmmm' to myself. You should probably take caution, there may be a bias in my approach here that may not suit you; or worse, could corrupt you.

The exposure we choose creates the tones that end up in the image. These tones have some resemblance to those of the real world scene we shot, but are by no means the same.

There is no truth in photography. The only truth an image may or may not have is that which the artist and the viewer care to share between themselves. Images are fragile. You can put all the truth you want into an image and all it takes to ruin it is one ill-fated context. Pictures and statistics, condemned to speak our lies for us. All you can do is work with truthful intent and hope for the best.

Which begins with owning the truth that the camera's view of the world is not our view. It won't tell the truth for you. It can't.

With all that said, let's look at how we convince the camera to approximate the world in a manner to our preference. Which of course means first understanding the world we wish to approximate.

You might think it's pretty dark when you close your eyes, and pretty bright when you look directly at the Sun (which really, you're best not to). But that just demonstrates the limits of our own vision. Things get much darker than any darkness you can perceive. The real world, beyond the scope of your perception, gets darker than the darkest of your dreams. My point is, there are degrees of black.

In fact, there are three levels of what we might think of as black:

  • Pretty black, but we can just to say make out a few details.

  • Quite black, we can't really make out any details but we do get a sense of texture.

  • Really black, too dark to even make out any texture, let alone details.

It seems appropriate to let The Black Sparrows demonstrate this:

demonstrating shades of black

What About The Exposure Triangle?

The exposure triangle is a great tool for avoiding the need to properly teach the basics of exposure...

Because there are 3 related things, they gotta make sense if you stick 'em together in a triangle, right?

Well, actually, no.

First off there are 4 things, not 3, that determine the resulting lightness (or darkness) of an image:

  • How sensitive the camera is to the light the sensor receives.

  • How much of the available light is allowed to pass through the aperture to reach the sensor

  • How long the available light is allowed to fall on the sensor, and

  • How much light there is!

So this 'triangle' concept ignores the most important aspect of exposure, the amount of light we are dealing with.

But, even if we accept this limitation and decide the triangle is useful 'for a given light level', we find other problems with it.

It puts ISO, aperture and shutter speed on an equal footing, as though they are completely and freely interchangeable. But these 3 things are not on an equal footing. Aperture and shutter speed both effect the aesthetic quality of the image, ISO does not.

The triangle makes it clear that there is 'some relationship' between 3 (of the 4) things that affect the image lightness, but doesn't really demonstrate what that relationship is. Eg. halving the area of the aperture (increasing the aperture by 1 stop) is equivalent to halving the shutter speed (in terms of exposure).

Finally, it's called the exposure triangle but it isn't even about just exposure. Exposure is about adjusting the aperture and shutter speed in response to the amount of available light (which the triangle doesn't even mention). The triangle does mention ISO, which certainly effects the lightness of the final image, but isn't actually anything to do with exposure!

The triangle looks simple at first glance, but because of all of its shortfalls things can get confusing when you try to apply it in practical terms.

Worst of all, the triangle is really only presenting one very simple idea, which really isn't worth all of the shortfalls and confusions it creates.

Forget the triangle. Just remember:

Halving the area of the aperture (eg. stepping down from f/4 to f/5.6) requires doubling the shutter speed (eg. switching from 1/125th to 1/60th of a second) to maintain the exposure; always keeping the ISO as low as you can.

Equally there are 3 levels of white: solid white; white with a hint of texture; white with just a few details showing through.

Everything else is grey.

The dark greys of deep foliage or wet granite. The mid grey of a northern sky on a winter's late afternoon. The light grey of shadows falling on a sunlit snowscape.

In digital photography we find the camera is better equiped to capture shadows than highlights. It can reliably scoop up the few photons that might be bouncing around in the deepest of shadows, details we can then bring out when we 'print' the image. Highlights are different though. Once the sensor has been fully saturated we have no hope of recovering the detail that was present in the scene.

Film, unlike digital, creates a negative image so we find the opposite is true. With digital we need to be certain we have preserved the highlight details because we'll never get them back if we don't; we can get the shadow detail back later. With film we need to be certain we have recorded the shadow detail; with the highlights being sorted out by the development we use.

Because the highlight details are so much more critical than the shadow details (in digital photography), I keep my camera exposure compensation set to around -⅔ of an Ev (-0.7). This tells the camera to knock a bit off whatever it thinks the right exposure is and pretty reliably guarantees that I haven't 'blown out' the unrecoverable highlights.

Of course, that's just my starting point. After every shot under new conditions I check the histogram. Note, I said 'under new conditions'. I do not check the histogram, or the camera's rear-display at all, after every shot. I think its called 'chimping', because of the body language it exhibits, that checking of the LCD screen after every single shot. As an action it takes you further from your connection with the environment you are photographing. Photography has an implicit level of seperation from the 'object' of the shot, simply because you are looking at it through glass. You could do with not exacerbating that disconnection by then immesing yourself in the camera's display. You need to fight to maintain a presence in the thing you are attempting to approximate.

And anyway, when I were a lad we all shot on film and had to wait weeks before we could see anything at all!

Don't be a 'chimp' chump, champ!

The Secret Lore of Photography, #69

Do check your histogram though.

Histograms are simple and make a great deal of sense. As you know the image is made up from millions of pixels, each one presenting a given tone (in fact, a given colour). Each pixel has a red, a green and a blue 'channel' (RGB values). Channel might seem a funny term but if you think that Radio 'channels' are defined by their frequency in the electromagnetic spectrum (eg. Radio 4: 92-95FM) it makes more sense. 'Red', 'blue' and 'green are of course just other frequencies within the same electromagnetic spectrum, so we might as well call these channels too. It's a term that crops up in post-processing software so we might as well get used to it.

A black pixel will have a value of zero in all three channels, whilst a white pixel will have maximum red, maximum green and maximum blue.

In a finished image (for examle a jpg file) there is one byte (8 bits) available to store the value of each channel; so it takes 3 bytes to store each pixel. Because each RGB channel is stored as a byte, they can have a value of between 0 and 255 (that's how bytes work); ie. they can each represent 256 distinct shades. Mixing the three channels together means we can represent over 16 million different colours. Seems a lot.

Until you realise the camera uses 12 or more bits for each RGB channel of each pixel. My D750 can use 14-bits for each, which allows over 4 trillion different shades to be represented; way more than the human eye could ever differentiate.

But no matter the bit-depth (8, 12 or 14...) there are a finite number of different possible shades.

To draw a histogram all we have to do is count how many pixels have each of the possible, limited number, of shades and then plot each shade's pixel count as a bar-graph. Here's the histogram for the earlier photo of The Black Sparrows:

Showing how image is represented as a histogram

The histogram shows the number of darker pixels towards the left, and the number of lighter pixels towards the right.

You can see a little hillock towards the right-hand side of the histogram that I've labelled 'T'. These are high tones of base player Rob's T-shirt; probably with some of Dan, the singer's, shirt.

As you might expect from such a dark image overall, there's a large peak on the left-hand side, in the shadow area of the histogram. Just as this is tailing-off we find the dark tones of the band's jackets (labelled 'J').

So that's how histograms work... this one though is for a finished image, where I've deliberately driven the contrast up and pushed a lot of the image down into the shadows. It's not at all the kind of histogram I want to see in the camera.

I make the fewest possible compromises in-camera. My aim is to ensure I have adequately recorded the full tonal range, with nothing blowing-out in the highlights or utterly blocking up in the shadows. I don't care what the image looks like in camera, I just make sure no highlight or shadow details are lost. Which means I aim to have a little bit of space at the very edges of the histogram.

Demonstrating over and under exposure

After shooting this glass, I took photos of my camera's top-plate information display at each of the settings that were used. In these you can see the small +/-0 exposure indicator; just above where the ISO is displayed, and beneath the shutter speed. This is how my camera offers its guidance on the exposure setting selected, based on its own light-reading. There are 'major ticks' at -3, -2, -1, 1, 2 and 3 EV; with 'minor ticks' at ⅓ Ev steps. You can see the camera has advised that the first image is under-exposed by 2 stops. The middle image was given 2 stops additional exposure by extending the shutter speed from 1/30th, through 1/15th to ⅛ of a second. At this point the camera thinks the settings are fine (and I agree!). The last image had a further 2 stops of additional exposure (from ⅛ through ¼ to ½), and you can see what the camera thought of that!

Other cameras will display this stuff differently, but at least you know what you're looking for!

When looking at the camera's advice regarding your f-stop and shutter speed settings, you might have been told to 'keep it at zero' - and you can see from the above experiments this is sound advice. But the histogram should be the final arbiter.

There's an important difference though, between what the camera advises based on its light-reading and what the histogram says about the photo taken. The light reading is looking at the actual scene, and is available before you take the photo. The histogram is looking at the actual data captured and can't be known until after the photo is taken.

Clearly, we're most interested in knowing that the digital image is correctly exposed. The light-reading at the time helps us to ensure that; but the histogram tells us (after the fact) if we did indeed achieve that.

So why use the light-reading at all? Why not just take the shot, look at the histogram then adjust and re-shoot?

Well, you can do exactly that, especially when shooting inanimate or dead things that don't move. But mostly in photography we're capturing life itself, and life has a habit of moving, constantly changing. So if you don't want to miss the shot, pay attention to the light-reading. But don't take it as gospel, your histogram is your gospel.

To be honest, when something catches my eye, I'll often fire a quick shot off immediately then take the time to consider my settings and re-shoot if I think I need to. Most of the time I find it is that very first shot that I end up using.

So when juggling your aperture and shutter speed, keep your exposure indicator round about the zero mark; unless the histograms you're getting are looking bad (bunching up on the left or right, 'falling off' either end). A histogram bunching up on the left (in to the shadow area) indicates underexposure; choose a smaller f-stop number (for a wider aperture) or a longer shutter speed. Conversely a histogram bunching up on the right-hand side (into the highlights area) inicates the image is overexposed; select a larger f-stop number (for a smaller aperture) or decrease the shutter speed.

Of course if you're shooting in semi-automatic aperture priority mode you can't effect the exposure by changing the aperture because the camea will compensate by using a different shutter speed in response. Similarly for shutter speed priority mode. In these cases either switch to manual mode or inform the camera of its error by adjusting the exposure compensation. If the histogram is telling you that the camera is underexposing your images add some compensation (eg. +1Ev). If the camera is overexposing, subtract some compensation (eg -1Ev). +ve Exposure Compensation will shift your histogram to the right (towards the highlights). -ve Exposure Compensation will shift your histogram to the left (towards the shadows).

You might think its easier just to shoot in manual, and in some respects that's true because you are in full control and the light-reading the camera takes is purely advisory. There's no need to faff on with exposure compensation.

But working in full-manual is slow, and the ambient lighting conditions can constantly fluctuate (especially when shooting outdoors). There are times when it is appropriate to work slowly and methodically. Certainly in a studio set-up, where things don't randomly change; and oft times when employing a tripod, which usually implies a rather painstaking set-up. But when shooting on-the-hoof, especially in street-photography, the semi-automatic modes are tremendously useful.

When shooting in manual mode:

  • Take a test shot with aperture and shutter speed set so that the exposure indicator is at the zero mark.

  • Inspect the histogram and adjust the exposure as needed, even though the camera thinks you are under or over exposing.

When shooting in aperture or shutter speed priority mode:

  • Just shoot.

  • Check your histogram from time to time (but don't chimp)

  • Adjust the exposure compensation if the camera is tending to over or under expose.

Looking back now at the 'reasonably exposed' photograph of the red glass (above) you can see that there is a lot more space to the right of the histogram (in the highlight area) than to the left. The image is perhaps just a little underexposed, even though the exposure indicator from the camera's light reading is at the zero mark. Well, that's because I always expose with the camera's exposure compensation set at -0.7Ev. Well, not always, but usually and by default.

This is because mostly I shoot out of doors. Odd patches of sky and clouds in the shot are typically much brighter than the scene around me. Consider the image below, where I have super-imposed the histogram. You can see quite a long tail of tones reaching all the way into the highlight area on the right. If I hadn't deliberately underexposed by ⅔Ev this area would have reached even further into the highlights and those white clouds would have completely 'blown out'. Skies are tricksy, and a patch of clouds blown-out to pure white with no texture or detail is often the reason a shot is completely ruined.

Image and histogram showing effect of sky highlights

So for my photography routinely underexposing by -0.7Ev serves me well. But I have a lot of experience with my own photographic style and my own equipment. Things will be different for you.

You might come across advice that you should always Expose To The Right: ETTR.

This means selecting an exposure that pushes the histogram as far to the right (into the highlight area) as you can, without causing it to bunch-up and lose highlight detail. Ie. that you should pretty much always overexpose.

This is based on the idea that low signal levels (that we get when recording shadows, because there isn't a lot of light in the shadows) are more adversely affected by the noise of the circuitry than higher signal levels. By overexposing we boost the signal levels and so improve the signal-to-noise ratio.

So should we overexpose to reduce shadow noise, or underexpose to ensure we don't burn-out the highlights? Of course, 'it all depends'.

All such advice, even my own (keep the exposure compensation set at -0.7Ev), is utterly ludicrous. Because such advice does not take into account your approach to photography, your style & intent, your equipment. The only advice worth listening to is you camera's light-reading coupled with your own understanding of the resulting histogram.

Rules of thumb are dangerous if you don't understand what lies behind them. But you have read this far so I will wrap-up by saying my rule of thumb is: keep the ISO low-ish (200, 400) and underexpose a bit (-0.7Ev). There, I've said it. Let the pedants pounce, I'm ready to dance!