Under what situations would you purposely underexpose a photo?

[Mike Buckley]

Dave and Remco,

You provided excellent responses to my comment that, for me, the epitome of a low-key or high-key image is one that has little contrast. The operative word, especially for the purpose of providing an example to Christina, is "epitome." I'm not suggesting that all low-key or high-key images have to have little contrast.

The amount of contrast might depend upon the luminosity of the subject as much as the treatment being given by the photographer. As an example, a high-key black-and-white photo of a person with light hair might use high contrast to emphasize the eyes. The photo of the rose that I provided would be an example of a high-key photo with little contrast because the subject simply has so few tones that are dark.

One of the confusing aspects about this is that the Internet is full of highly contradictory information about the definition of low-key and high-key photography. In my search for good examples for Christina, I saw one article that even contradicted itself.

[Mike Buckley]

It just now occurred to me that the issue of contrast is very important when answering Christina's question. As an example, if you want a low-key image with high contrast, I can't think of a reason to underexpose the image; if you did that, you would lose the contrast provided by the few very bright tones in the image. Yet if you do want to lessen the contrast, you could underexpose to the point that the bright tones are rendered darker. Naturally, you would take care not to underexpose so much that the detail is lost in the darkest tones.

[Shadowman]

Christina,

Here is a short article on the subject and they included another reason, to shoot silouhettes.

http://betterphotography.in/2012/03/20/breaking-norm/

[Brownbear]

Thank you everyone. Lots of great advice, information and resources

[Brev00]

I would say I have two approaches that I use that might answer your questions. First, I don't often use aperture priority. I usually shoot in manual. While the exif of my images will show matrix metering, I use my fn button on my Nikon D90 to spot meter most scenes and come up with the exposure I want from there. I developed this approach partially because when I used aperture priority, I often had to apply exposure comp of some kind to avoid the highlight blinkies. That led me to spot meter the highlights which led me to the manual mode. Why bother with the trouble of reacting to the camera's guess when I could proactively come up with my own? Now, I still use aperture priority but mainly when the scene has few tonal extremes. Just the other day, I shot a scene in aperture mode that at first needed no comp. The scene involved late light over water. The contrast between the water's surface and the sky varied depending on the position of the sun, how much sky was in the shot, etc. When I shot with the sun behind me, I need no comp. But, as I shot with sidelighting, I needed about -1 ev comp. If I wanted to shoot with the sun closer to or in the frame, I needed more. If I shot mostly water, I needed less. Most often, I will just shoot in manual and adjust the shutter speed. That seems so much simpler to me. The shutter is controlled by one movement of a nice big dial while changing the ev is more complicated.

I will give you another example where spot metering in manual mode works well for me: shooting a cityscape. Shooting in aperture priority, the scene will be well exposed except for the windows and lightly painted building surfaces. They will always blow out unless I apply comp or switch to manual, spot meter the highlights, and expose from there. I prefer the latter.

[Manfred M]

Thinking about things, I do remember deliberately underexposing landscape shots when shooting slide film to get deeper, more saturated colours. I shot a lot of Kodachrome 64 (which had a ISO 64 speed) and I would occasionally shoot it at ISO 80, i.e. 1/3 stop underexposed to get better saturation and more detail in the sky.

I haven't tried it in digital, but might do so now that I think about it.

[Dave Humphries]

I have never felt the need to understand "the zone system" for my own photography, but this thread caused me to go read the luminous article Bill linked to in post 11 above, to see what it was all about.

Well, now I know what it is

It seems to me that if digital photographers just read the recommendations there for slides and ignore all the bits about negative/print (that just 'fills your head' with stuff you don't need to know), it suits quite well for digital shooting.

Perhaps someone that understands it (more fully than I do) could write up a simplified 'digital version' for future reference?

I used to shoot slides for the last few years of my film use before I took the 'photographic sabbatical' that career and family life forced on many of us. Perhaps that explains why I didn't have too many problems adapting to digital shooting (exposure wise), I was already well used to the need to 'protect the highlights' when reading a meter.

Cheers,

[Inkanyezi]

I think we are in a way leaving the original question here, or perhaps not. Of course you need not read about and understand the zone system in order to expose correctly with a digital camera. The digital camera has many aids for exposure, and the final image, which is always the goal, will often show up on your screen seconds after pressing the shutter.

The core of the zone system has little to do with the zones per se, but visualisation of the final image, the one that you will see printed or displayed. The zones are a kind of crutches for our imagination, to make it easier to evaluate what is possible to do with the scene we have before our eyes. Learning the system, as I did, as an introduction to photography, is a mental tool to understand how the photographic medium responds. What more my teacher worked on, from start, was to understand how to build an image with contrast, how to compose it with the aid of lighting and objects, with their shadows and highlights.

We worked with black and white film in the early sixties, developing our film and printing on paper. We even used sheet film, in order to take just one exposure, develop it and see the outcome. We also took just a few frames with the miniature camera, cut off the film in the darkroom and developed the exposed few frames, much in the same way, to evaluate what we had done without waiting for 36 exposures to be shot. We did it to learn how our various lighting setups would work.

I think it is good to have started that way, and that understanding the limits of the medium is valuable knowledge. The difference between film and digital is not too large, essentially I work in the same way now as fifty years ago, exposing my digital chip just as I would expose a film back then. With the zone system, I learned to work within the limitations of the medium, and we also learned just how much those limitations could be tweaked by different exposure and development. Much the same thing now, although the very limits are somewhat different, the most prominent difference, that there is a hard limit in the bright end.

But the zone system applied to digital will tell you another story about the ISO setting than the most pronounced one, that sensitivity would be altered by setting ISO. In fact, the ISO setting does nothing to the sensitivity of your chip, it only alters amplification, which raises contrast and pushes all values a bit to the brighter side, at the same time pushing the highlights over the edge to be lost forever. It is much akin to the "pressing" we did with our Tri-x in the sixties. "Pressing" by developing more, higher temperature, more concentrated developer and longer time, after using a higher exposure index, raised contrast and blocked the highlights, decreasing dynamic range, at the same time as grain increased, just as inreasing ISO does in a digital camera, if you swap grain for noise and accept that the raised contrast results when you set the black point in post processing.

So maybe there could be a reason to make some simple to understand article about the zone system for digital shooting, and just as the original zone system, visualisation of the final image would be the core point, while we would never have to think about any negative, but would found it entirely on what's on the chip and what we can find out from the histogram. But as photography is a visual art, this technical knowledge may be sidestepped and its core values understood without referring to the zone system. By the end of the day it would probably boil down to exposing to the right at all occasions where you have to utilise the full dynamic range that is available; keeping in mind that dynamic range decreases with increased ISO setting, i.e. exposure index.

[Brev00]

One more thing: I will purposely underexpose a lowlight action shot to get the movements crisp when slowing down the shutter will result in a mess of blur. I can raise the exposure and shadows later along with reducing the noise. This is not a favorite way to work but, given the limitations of my gear and the light, it can make for a decent shot.

[Inkanyezi]

/.../ I can raise the exposure /.../ later /.../

So many times have I wondered whether this is a helpful way of explaining a technical quirk that has nothing whatsoever to do with exposure. We cannot alter exposure the slightest bit after releasing the shutter. Exposure is what takes place during the short moment the shutter is open. No software can alter exposure in post production, nothing less than magic could, but we are not magicians. I think it was a mistake to stamp that name to a software control that does something entirely different.

We cannot raise the exposure later. Exposure takes place when we release the shutter. Afterwards, we may alter tonality, we may alter the tone curves, set "blackpoint" and "whitepoint" to desired values, and we can manipulate the curve in a way that resembles changing ISO. In fact, we can set a higher ISO in post production, but we cannot change exposure. The drawback when increasing ISO in post production, instead of in the camera, is that it invites posterisation, as there will be fewer digital levels to play with.

So software controls can be used to cope with some of the issues regarding exposure. Even if exposure will not be changed, we can make image tones brighter if they came out dark. We can bend the tone curve in a way that imitates increasing ISO, although when ISO is increased in the camera, there are many more digital levels in the dark areas.

[Mike Buckley]

In fact, we can set a higher ISO in post production

I'm not an expert when it comes to understanding this stuff, but I don't think that's true. Just as we can use the software only to emulate many of the effects of having used a different exposure at the moment of capture, we can only emulate the effects of having used a different ISO. The primary software that I use, which is made by the camera manufacturer, has no adjustment capabilities that even refer to the ISO value.

[Manfred M]

So many times have I wondered whether this is a helpful way of explaining a technical quirk that has nothing whatsoever to do with exposure. We cannot alter exposure the slightest bit after releasing the shutter. Exposure is what takes place during the short moment the shutter is open. No software can alter exposure in post production, nothing less than magic could, but we are not magicians. I think it was a mistake to stamp that name to a software control that does something entirely different.

We cannot raise the exposure later. Exposure takes place when we release the shutter. Afterwards, we may alter tonality, we may alter the tone curves, set "blackpoint" and "whitepoint" to desired values, and we can manipulate the curve in a way that resembles changing ISO. In fact, we can set a higher ISO in post production, but we cannot change exposure. The drawback when increasing ISO in post production, instead of in the camera, is that it invites posterisation, as there will be fewer digital levels to play with.

So software controls can be used to cope with some of the issues regarding exposure. Even if exposure will not be changed, we can make image tones brighter if they came out dark. We can bend the tone curve in a way that imitates increasing ISO, although when ISO is increased in the camera, there are many more digital levels in the dark areas.

Urban, while I agree with you philosophically, from a technical standpoint this is really the digital equivilent of "pushing" film in wet processing. Leaving the film in the "soup" for an extended period meant that we were able to convert additional silver salts into metallic silver, regardless of the manufacturer specified film speed, albeit at a cost of graininess and resolution.

This is even more so the case when we look at RAW images; those that can exceed the dynamic range of our output display by perhaps four or five stops. Shifting the raw data around in post actually gives me a cleaner image than I ever got out of pushing film.

[Inkanyezi]

Urban, while I agree with you philosophically, from a technical standpoint this is really the digital equivilent of "pushing" film in wet processing. Leaving the film in the "soup" for an extended period meant that we were able to convert additional silver salts into metallic silver, regardless of the manufacturer specified film speed, albeit at a cost of graininess and resolution.

The standpoint is almost purely technical, and there is actually not much philosophy involved. I regard it as a deeper understanding of the properties of the photographic media, whether silver salts or semiconductors.

Using different exposure indexes with film or electronic sensors is indeed exactly equivalent, whether we do it with the camera control that amplifies the signal or amplifying it later with software. Analogue amplifying in the camera renders a cleaner output than digital amplifying. And it should be stressed, that the "additional silver" that was converted corresponds exactly to what happens with the digital image, where more contrast is fetched from the darkest areas. And, importantly, this does not change the "film speed" or the sensitivity of the image chip. It only changes the tonality of the image. Film speed remains the same, chip sensitivity remains the same, although we set a higher exposure index. What happens when "pressing" or raising ISO, is that dynamic range shrinks when we apply either of those processes, and it is the highlights that are lost in both cases. When we output an image with full tonality, contrast is higher than when normally exposed and processed.

So we fool ourselves by thinking that our 400 ASA film became 1600 ASA by processing hotter and for longer time. Its sensitivity remained the same, although it no longer corresponded to the ASA standard, as it could not capture a scene with normal contrast. Highlights were blown, and shadows became pitch black or grainy with no structure resembling the scene structure, but all structure that came out in the black was from the film itself. The same exposure index could have been achieved by exposing identically with normal development, printing on a higher grade paper, although it would then not be quite as grainy, and it might have been impossible to find such a high grade paper.

When you understand the zone system, you will fathom this and know what actually happens when processing your film harder. We did the same when using the zone system, but with a different purpose, to raise contrast for a low contrast scene.

This is even more so the case when we look at RAW images; those that can exceed the dynamic range of our output display by perhaps four or five stops. Shifting the raw data around in post actually gives me a cleaner image than I ever got out of pushing film.

I think it would be worthwhile also to use more distinct terminology regarding raw files and derived images. There is no RAW image, there are only RAW files. The image is calculated by software in a process where several things are done to the file. When the RAW file is downloaded from a Bayer chip, it must be demosaiced in order to allot correct colour to each pixel, as they would otherwise be pixels of the three basic colours, green, red and blue. A RAW file may have captured a larger dynamic range than can be displayed or printed, and there are software solutions to cope with that in various ways, mostly by altering tone curves or by tone mapping. When we shift the data around, we do nothing to the sensitivity of the chip, but we can alter the amplification, in a similar way as the analogue ISO control in the camera, with the difference that in the camera it is done before A/D conversion, but in software, we do it to the digital data, that is more akin to steps of a ladder while the analogue signal has a gentle slope.

There is actually not much philosophy about this, but they are technical properties. It helps to understand them, because then you know what you can do and why. I do apply the zone system to digital photography, in a way that understanding how the medium can capture a certain dynamic range, I know how it can be rendered in the final image. That is achieved in the mental process of visualising the final result. I never worry about noise, I don't bother at all about it. The noise is there from start, it is always there, it is only one property of the electronic image making that we have to deal with. I do however care about dynamic range. I won't raise ISO to very high values unless my scene has low contrast and I know that it will be improved if I raise it. When I need a high dynamic range, I set a low ISO, much in the same way as when pull-processing film for higher dynamic range when using the zone system that Ansel Adams outlined.

From this knowledge of what tonality I can actually get out of a certain dynamic range, I have a better understanding of the dynamic range curves presented for example by DxO. Understanding these inherent properties of either film or image chips helps me to understand what I will sacrifice if I override any of the limitations set by the media. When I let a highlight blow out, I know that it is lost forever, and I do it on purpose, because it was not all that important to the image, but perhaps I needed a shorter exposure time to stop motion. A blown highlight or a dark shadow then is a trade-off, done deliberately.

And "philosophically" there is no difference compared to film. The main difference is that each frame in digital photography can be individually processed, just as when we used plates or sheet film. Or even better, because we can process it in as many different ways as we please for years on end as long as we keep our RAW file.

[Letrow]

Lots of answers already Christina, so hopefully that helps you. For my bit, I would trust the camera in most cases. I was a bit surprised to see that you generally upped the exposure in most cases, so I hope I understood that correctly. As Mike pointed out, it is easy to correct (modest) underexposure in PP, but overexposure can be more difficult.

Anyway, I underexpose deliberately when my camera makes a dark scene too bright and I want to capture that dark atmosphere. The camera is right of course in doing what it does, but it might not please me. In case of snow, or a lot of white I would do the opposite and overexpose if needed to capture the white instead of getting gray.

It is important to look at what you are measuring exactly. If you do a sunrise or sunset you should measure the sky next to the lightsource, not the lightsource itself. You can get great results with that, especially if you use AE-L (if your camera has that button). The camera measures for the sky (not the lightsource), you lock exposure, recompose to get the lightsource in your photo and shoot.
Hope that works for you.

[Inkanyezi]

How does matrix, center and spot metering come into play in this situation?

I return to this question, which I feel is not fully answered, if it ever can be.

Spot metering is fairly straight forward, because with the spot meter, you will know what you measure, at least pretty close. The companion of the spot meter is understanding the foundation of the zone system, the zones in the final picture, printed or shown on screen, and how they relate to the measured value.

Centre weighed metering is a fuzzier concept, and I regard it as a "general" measurement. I cannot trust my understanding of what it actually is measuring and just how the different areas are weighed against each other. Would very bright highlights in the peripheral area influence the reading, and how much? There are more questions than answers, so in essence, for me, it is some kind of average metering with fuzzy borders.

Matrix metering is a quite different animal. Just as with the centre weighed metering, we cannot know what it does and why. Exactly what areas are measured, and how do they weigh in on the verdict? A nagging thought also is that maybe it is just a marketroid lie, an unimplemented feature that looks very nice to have. Does your camera actually have hardware that can distinguish various fields within the image? Quite clearly, Nikon did implement such a feature, by providing a minuscule camera into the position where the light meter cells were usually placed in many older SLR TTL cameras, close to the ocular. But to my knowledge, no such hardware is implemented in any Canon camera; their only system camera that has the physical ability to distinguish fields in the image area being the EOS M. Of course the possibility exists also in other models when Live View is used, but still we would not know exactly how the readings are averaged. To me, the situation is very akin to the centre weighed scenario; I cannot know what it does, so I regard it as a slightly unreliable concept. However it gives the promise that it can avoid burnt out highlights. Nikon has such a feature, Active D-Lighting, Canon purported one, Highlight Tone Priority, which at least in the EOS 40D is a blatant marketing lie. No such feature is implemented in that camera, and no Canon DSLR camera has the necessary hardware to implement such a feature. It may be done as others have, by simply exposing a bit less than an average reading, and I think that they might have done so with their "HTP" feature in later models.

A true matrix metering is possible in all cameras that take their reading directly from the image chip, as compact cameras or system cameras without an optical viewfinder. Whether it is done, and how it is implemented is obscure. No camera manufacturer has produced whether evidence or functional description of how matrix metering should work, and what calculations it does. Only Nikon has presented the concept regarding their little camera in the viewfinder, and claim that the image is evaluated against a board of thousands of examples. When the camera is set to Active D-Lighting, it would be fair to think that it is programmed to expose to the right, ETTR.

So, if you have a Nikon DSLR, their Active D-Lighting should be expected to take care of compensation for you, making sure that you don't burn the highlights. Whether you can use the jpeg or not depends on your D-Lighting settings, but your RAW file should be exposed to the right. What you could expect from other makers is veiled in even much more fog, and personally, I think of the different metering modes as gimmicks with no real substance. When taking pictures, the photographer has to learn how the tools function, and we have to bridge all shortcomings in one way or another. I have my own idea of how to do it, and I don't trust any DSLR system to set exposure or sharpness correctly, and that is one of the major reasons why I use an EVIL camera. I have real time blinkies and histogram, and I know what result I can obtain from a spot reading of a highlight or any other shade that I may place in a particular zone of my zone system.

So, I don't trust metering systems that I don't really know what they do, and I don't trust marketroids.

[Brownbear]

Thank you Peter.. Yes, all the replies are enormously helpful. I've learned a lot. I should not have said that I overexpose regularly, I meant to say that I often in increase the exposure compensation a bit (usually for bird shots) to obtain the correct exposure. I will try it out on my next sunrise/sunset shot. thank you

Thank you, Urban. Very educational and extremely helpful. I do have a Nikon DSLR and today I will be looking up Active D Lighting.