Exposure is not how dark or how light an image looks!

Exposure is not how dark or how light an image looks!

When it comes to online discussions, it's helpful if everyone has the same meaning for a word. In the world of photography, there has been a lot of confusion in just what is meant be "exposure".

Traditionally, "exposure" refers to the amount of light reaching the film, measured in light per unit area. If we think of light being comprised of photons, then we can think of "exposure" as how many photons per unit area reached the film, while the shutter was open.

Note that this is not the same as how "bright" the light was on the film. You can halve the brightness, and leave the shutter open for twice as long, and you get the same exposure.

The three primary factors in exposure are:
- Subject Lighting
- Shutter Speed
- Aperture

With film the relationship between the exposure and the film "speed" determined how dense a negative you got. If the exposure was too high (too many photons), then you got a dense negative. Too low an exposure, and the negative was too thin.

Most negative films have an "S" shaped response curve. You got the highest quality negative by getting the mid tones of your image into the middle of the response curve. If the negative was too dense or too thin, you would lose contrast and/or shadow/highlight detail.

None of this controlled how dark or light your final print looked. That was determined at print time. You could easily make a dark or light print from a thin, normal or dense negative. How dark or light the print looked was determined not by the exposure, but by the person (or automated machine) making the print.

When it comes to digital, exposure refers to the amount of light reaching the sensor instead of the film. The vast majority of digital cameras capture raw data from the sensor. Very roughly speaking, the raw data contains a count of how many photons were seen by each pixel while the shutter was open.

How dark or light the image looks is determined when that raw data is converted into a traditional image (usually a JPEG). That conversion is done in the camera to build camera-produced JPEGs. You can also have the camera store the raw data in a "raw" file, and do the conversion later on with a computer.

The ISO speed affects the "context" for interpreting the raw data. At a low ISO setting, you need a high photon count in order to produce a light area in the JPEG. At a high ISO setting, a much lower photon count produces a light area.

For instance, on a particular camera, a photon count of 10,000 might produce a dark pixel at ISO 100, and a light pixel at ISO 10,000. In both cases the exposure is the same (same number of photons seen), but the different ISO settings yield different interpretations of those photon counts.

In summary, "exposure" refers to the light reaching the sensor. the ISO setting drives how we interpret the recorded data.

Keep in mind that the above is a very rough approximation. In the real world, raw data is a little more complicated than simple photon counts. Some cameras will scale the readings based on the ISO settings. Furthermore, ISO is not the only factor in how the data is interpreted, other settings (White balance, Highlight Tone Priority, Auto Lighting Optimizer, etc.) also play a roll.

Just expanding on Michael's post with a practical example.

For a given scene and lighting:

1. f/8, 1/200s, ISO 100
   and
2. f/8, 1/400s, ISO 200

will both output the same image lightness but shot 2 will have had only half the exposure of shot 1 because less light from the scene reached the sensor.

Consequently, shot 2 will have more visible noise than shot 1 because of its lower exposure but that is a topic for another thread.

Image lightness and exposure, although related, are two different things and the two terms should not be used interchangeably as if they had the same meaning.

Not necessarily. If both shots are still dark and you push the development to the same brightness level then with many cameras the ISO 200 shot will have marginally less noise than the ISO 100 shot.

The question is visible noise. With current cameras 100 to 200 ISO may not produce extra visible noise.
Pushing the development is another discussion altogether.
Properly exposed pictures are in eyes of the photograph.
Not an easy subject to discuss.
And sometimes the viewer may feel this image is too... whatever.
For beginners though, this is critical learning issue. For older people, I suspect too many rely on the 18% grey AE. Sad, as we miss huge creative exposure opportunities.

DannoB stated that ISO 200 produces more visible noise. Whether that noise is visible or not is besides the point I was making, my point is that beginners (and others) should understand that always opting for the lowest ISO does not always result in the best image quality.

I'm not sure what you mean here :-(
Are you referring to the actual amount of noise or how much of the actual noise is visible.
If you mean actual amount of noise then yes, the actual amount of noise in the ISO 200 shot will be less in many cameras because as shown in the charts at photonstophotos, raising ISO reduces the read noise in the raw data.

But more of that reduced amount of noise in the ISO 200 shot will be visible because of its lower exposure*

* exposure - amount of light striking the sensor per unit area while the shutter is open

Yeah, I had a brain-fart.

Yes, I have those too occasionally but I prefer to refer to them as my senior moments :-)

That's a great idea :-)

Sort of.

When you "push" the development, you change the ISO.

Back when I was shooting film, I often used a film that was rated at 400 when developed normally, but was rated at 800 when push-processed in a particular way.

But most people reading this are likely shooting digital.

As it turns out, the ISO speed is not defined for raw files. Remember, the primary purpose of the ISO speed is to give a context for processing the raw data to produced a JPEG image. If you change the raw processing parameters to brighten (or darken) the image, then you are literally changing the ISO Speed.

In the real world, it's more complicated. For many cameras the ISO setting actually performs three functions:
1. As discussed, it affects the context for interpreting the raw data.
2. It sets a target exposure for the camera's metering system.
3. It lets the camera optimize internal processing to the target exposure.

We've already discussed setting the context for interpreting the raw data. See the first post in this thread.

When you set a fixed ISO, that sets the target exposure for the camera's metering system. For instance, if you set the ISO to 400, the metering system targets an exposure that matches that setting. The higher the ISO setting, the lower the target exposure. If you are relying on the camera's meter, then selecting a higher ISO results in your getting a lower exposure. It's possible to ignore the camera's meter and get exposures that are not a traditional match for the ISO, but that's not typical.

It turns out that some cameras can optimize their internal processing for various situations. For instance, the camera might have a mode where it adds less noise to the image, at the expense of a lower limit on the maximum exposure. Often, this mode will be entered when you set a high ISO. The camera assumes that there will be a corresponding low exposure, and therefore it's OK to have a low maximum tolerable exposure. At low ISO settings, the camera configures itself to tolerate the high exposure, even though it means the camera will add a little more noise to the image. That isn't an issue as it isn't the absolute lever of noise that's important, but the ratio of the signal to the noise. If the signal (i.e. the exposure) is high, then a little bit of noise isn't an issue, as the signal is much higher. If the exposure is low, then a slight reduction in the noise can make a noticeable difference.

Consider two images of the same subject, both taken at f/4, and 1/500. The camera is set for RAW+JPEG.

For the first image, the camera was set to ISO 3200 and produced a good looking camera-produced JPEG.
For the second image, the camera was set to ISO 100, and the camera-produced JPEG was far to dark. However, you took the ISO 100 raw file, and adjusted the processing parameters to match the image lightness of the ISO 3200 capture.

Both captures had the same exposure. Surprisingly, with many cameras, image taken with the camera set to ISO will be a little noisier!

As it turns out, with low light photography, the biggest factor in image noise is usually the "shot noise". This is the noise inherent in the quantum nature of light. Shot noise becomes less of an issue as exposure increases. As both captures have the same exposure, they will have the same shot noise.

However, many cameras add less noise to the image when they are at higher ISO settings. Thus the image taken at ISO 3200 may have less noise than the image taken at ISO 100.

Note that this is true because we have specifically chosen as example where the exposures are the same. In the typical case, the ISO setting is correlated to the exposure. As you set higher ISO values, you normally also select lower exposures. These lower exposures increase shot noise, and hence we normally see noisier images when we choose higher ISO values.

Thinking that higher ISO values are the cause of the noise is like thinking that people walking around with umbrellas cause it to rain. Yes, if you see a lot of people walking with umbrellas, it will likely rain, but if you somehow convinced everyone to carry an umbrella, it wouldn't make it rain.

Yes, we normally see noisier images with higher ISO settings. But selecting a lower ISO, without an increase in exposure, won't reduce noise.

I am with you 100%, Michael !!

I am one of the "how much light whacked the sensor" crowd.

On the other hand, the terms over- and under-exposure are about how dark or how light an image looks.

I totally disagree because in the scenario where I set the optimal exposure**, as defined below, the image could still be too light or too dark if I or the camera set an inappropriate ISO for my desired image lightness.

For a given scene lighting and exposure* I can vary the image lightness by varying ISO without altering the exposure* at all.

For example, after setting the optimal exposure** it is impossible to add more exposure* because then either my DOF or motion blur requirements would be messed up and not met. In this case if the image is still too dark it is not because the exposure* was too low (underexposed) but because the ISO was too low.

Exposure* and image lightness are two different things, not two different terms for the same thing.

When someone says an image is under or over exposed what they really mean to say is that the image is too dark or too light, so why not just use the correct terms in the first place?

For me, an image is under exposed if more exposure* could have been added with the DOF and blur constraints still being met.

For example, for a given scene:

Say someone used,

A. f/8, 1/200s, ISO 400

to get the image lightness they wanted when they could have used

B. f/8, 1/100, ISO 200

and still met their DOF and motion blur constraints. Both settings will output the same image lightness but shot A is underexposed because less light than was possible was allowed to strike the sensor to meet the dof, blur and image lightness requirements.

* exposure - amount of light striking the sensor per unit area while the shutter is open
** optimal exposure - the maximum exposure* within dof and motion blur requirements without clipping important highlights.

They certainly were not about that in the days of shooting traditional film negatives.

In those days, over- and under-exposure were about whether the negative was too dense or too thin, not how dark or light the print looked.

How dark or light the print looked, depended on how you printed it. Drop your film off at a drugstore with an automated lab, and the prints from over- and under-exposed negatives will have the same lightness. The machine had a target lightness, and would try to hit it for every frame, independent of exposure.

Similarly, when I was printing in the darkroom, I could make a dark or light print from just about any negative, whether it was over-, under- or normally exposed.

When it comes to digital, using "under-exposure" for an image that is too dark can easily mislead a beginner. For such an image, the exposure may have been exactly appropriate for conditions, but the camera's ISO setting was too low.

Remember, with digital, how dark or light the JPEG looks is dependent on the relationship between exposure and ISO setting. If the image looks too light or too dark, the issue may be "ISO set too low" or "ISO set too high". It's not a good thing to suggest to beginners that the issue must have been improper exposure.

The terms exposure and optimal exposure are well-defined.

However, the terms overexposure and underexposure are ambiguous, IMO. I would like to see an attempt to define them in digital photography before the argumentative DPR members move to this forum :).

By "how dark or how light an image looks," I meant "how dark or how light the raw or JPEG file initially looks."

A beginner would probably work with SOOC results, but I agree that the terms over and underexposure are ambiguous. They can be clear only in a wider context of what is being discussed.

Does over/under exposure mean that the exposure is improper?

That would likely mean that 99% of our images are underexposed, as we do not have raw histograms, and almost nobody brackets 1/3 stops to get to the exposure just before clipping of relevant highlights.

The terms are regularly used in the context of over/underexposing by n stops, making the initial raw/JPEG darker or lighter than the metering suggests (neutral exposure).
We also use the terms to say that an image looks over/underexposed by judging its initial brightness. E.g., ETTR images often look overexposed.
But we also use the term overexposed when unwanted clipping occurs.

While the term "Exposure" is well defined, "Optimal Exposure" is a bit problematic. The exposure that's "optimal" for a given situation will vary by the parameters you use for measuring. An obvious example is that the "optimal exposure" for minimizing noise in the captured raw file, may not be the same as the "optimal exposure" for producing an artistic camera-produced JPEG at base ISO.

Some might define the "Optimal Exposure" for a situation as the highest exposure that does not blow out important highlights, yields sufficient depth of field, and does not result in unwanted motion blur.

Now, you might argue that there is only one true measure for judging "optimal" exposure, but other's may reasonably disagree with you.

If we are going to discuss how the industry uses various terms, it is helpful to discuss their history. There's an established history of over 50 years of "exposure" being used to mean the light reaching the film/sensor while the shutter is open. My point was that in the days of film, over/under exposure was not about how dark or light the resulting image looked. You believe we should discard that traditional meaning, and embrace a new meaning. Changing the meaning introduces new problems. By making "exposure" synonymous with "image lightness", we now need to create a new term to discuss the light falling on the film/sensor while the shutter is open. As you think that the word "exposure" no longer means that, can you suggest a new term to have the traditional meaning of "exposure"?

Yes, beginners frequently work with Straight Out Of the Camera (SOOC) JPEG images, If that JPEG is too dark, it may very well be that the had the best exposure for the situation, but their ISO was set too low.

I think the best way of describing a SOOC JPEG that's too dark is to tell the beginner that it was "too dark". Then explain this is due to the relationship between ISO and exposure. To make the image lighter, they should increase the exposure or increase the JPEG.

Here's a thought experiment.
If the JPEG is too dark, why not simply tell them that their shutter speed was too fast? After all, increasing the shutter speed will raise the exposure and increase image lightness. I suggest that we don't describe all dark SOOC JPEGs as have too fast of a shutter speed, as that's not the only possible cause. In fact, the cause may not be the exposure at all.

Similarly, we should not describe dark SOOC JPEGs as being "under exposed" as that's not the only possible cause of the issue.

Now we are getting into arguments of science vs. engineering. A scientist might point out that it was possible to increase exposure by 1/100 of a stop, without violating any constraints. The scientist may declare that the exposure was lower than what was desired, and therefore "under exposed". An engineer might point out that 1/100 of a stop wouldn't have made a visible difference, and therefore the exposure just fine.

If the exposure is reasonably close to the desired exposure, it is not crazy to say that the exposure is correct. Of course "reasonably close" can vary with the needs of the task at hand.

This is similar to the concept of Depth of Field. A scientist might point out that when the camera is focused at 10 feet, only objects exactly 10 feet from the sensor are in focus. A tenth of an inch further away, or a tenth closer, and the object is not in perfect focus. An engineer might point out that there is a range of distances where the focus is close enough, that under typical viewing circumstances, those objects will appear to the human eye to be in focus. That range is the depth of field. While objects in that range are not exactly in focus, the focus is good enough that they look in focus.

That may be how you use the terms. It is certainly not the traditional way the terms have been used, and it is certainly not a universal use. That's clear from the number of people who disagree with that usage.

If you really want to redefine "exposure", you may want to suggest a good reason why we need a new synonym for "image lightness" and you may wish to suggest a replacement term for what has been traditionally called "exposure".

For the sake of clarity and consistency what is the definition of the word exposure you use when you say an image is over or under exposed?

I explained earlier why I disagree with your opinion.