About the word "exposure"

I thought I would talk a little about the term "exposure" in the quantitative sense (that is, not in the "narrative" sense of "I took the exposure at just the right time)".

There are two different technical properties that both legitimately carry the name "exposure", and a third to which it is sometimes applied (although ill-advisedly). I'll discuss the first two.

1. Exposure - the joint effect of the exposure time (shutter speed) and the effective relative aperture (which we state as an effective f/number). It is this property that is reflected, in logarithmic form, by the APEX value Exposure Value (Ev). [Do not be misdirected by the common, but improper, use of "Ev" to mean "scene luminance" - another story for another time.] It applies to the entire "shot" process. I will call this factor camera exposure here to distinguish it from the exposure we will hear of next.

2. Exposure - the product of the exposure time and the illuminance on the imaging element (film or sensor). (More precisely, the time integral of the illuminance over the exposure duration.) This second kind of exposure is often called, in technical writing, photometric exposure. It applies at specific points in the image.

The third property often called "exposure" is what I call exposure result. It is what photometric exposure leaves behind for us. In the case of film, the physical property is density (at any point in the image). In the case of a digital camera, the property is the digital representation of the image (at any given pixel). Especially in color-filter array cameras, we need to distinguish between the raw data exposure result (which pertains to a sensel) and the developed image exposure result (which pertains to a pixel of the developed image).

So when we look at the histogram of a developed image, and comment on the "exposure", we are actually contemplating the exposure result, and either describing it colloquially as "exposure" or are actually referring to the photometric exposure that led to that exposure result, or the camera exposure that led to that.

When we develop a raw data set, the development software usually gives us the option to vary the transform between the raw exposure result and the exposure result that will be placed in the developed image. We often do this when it turns out that the camera exposure was not ideal (based on the distribution of exposure result we crave). It is, in a way, rather analogous to using non-standard development of film ("pushing" or "pulling").

When we do this, people often say, "I adjusted the exposure". But in fact, the exposure (in the narrative sense) has already happened, there was a certain value of camera exposure used, and a certain photometric exposure occurred at each sensel of our sensor, and was captured to the raw data. We cannot vary either of those at this juncture.

What we are really modifying is the set of exposure results.

In fact, because of the common expression, in some raw data development software, the "slider" used to vary the exposure result is labeled "exposure", an understandable (if slightly sad) decision by the interface designers. In others, it is labeled "brightness", and so forth.

Normally, in context, which of the three meanings (two legitimate, one not really) of "exposure" is meant is usually clear. But it is worthwhile to keep in mind that there are really three different properties that might be involved.

Next week: the Cherokee name for the red fox.

I have adopted the following terminology for use in my subsequent technical writings (with respect to digital photography):

• For "exposure" in the first sense (the joint effect of shutter speed and effective relative aperture): Camera exposure (CE). [Applies to the shot process.]

• For "exposure" in the second sense (the illuminance-time product on the sensor): Photometric exposure (PE). (This is the term ordinarily used in scientific writings.) [Applies to specific image "points".]

• For the raw data resulting from the photometric exposure: Primary exposure result (PER). [Applies to specific image points, in the sense of specific sensel outputs.]

Note: in the case of film, this would be the photochemical impact on the undeveloped film, the physical manifestation of the "latent image". We cannot observe this directly.

• For the developed image data resulting from that raw data: Secondary exposure result (SER). [Applies to specific image points, in the sense of specific pixels.]

Note: in the case of film, this would be the resulting density of the developed film.

I will of course be certain to define these terms before they are used in an actual article or note.

Best regards,

Doug

I read what you've wrote out, and understand you, however I think the technicalities of people labeling "exposure" seems perfectly valid to me.

Post-process controls like "brightness" or "exposure" often use different algorithms that affect the curve differently. I don' t know what the algorithms are in the various raw converters are, but in our compositing package at work there are alot of sliders that effectivly make the image brighter or darker in different ways, gain, gamma, white point, black point, lift, multiplier, offset, and a bunch of others in scale that has a variety of options by stop, light, density, cineon, panalog, linear, or whatever. So having the variety of options with explanations of what they do is fine because more technical users use this program, and may actually want/need to know what each of them do for a specific technical purpose.

However, for mass-used programs with casual users, I think "exposure" is a good enough term to be used. Because it simulates what the brightness level of the image would be if you had a brighter exposure, nothing else the user needs to know.

Hi, Daniel,

Daniel Buck said:

However, for mass-used programs with casual users, I think "exposure" is a good enough term to be used. Because it simulates what the brightness level of the image would be if you had a brighter exposure, nothing else the user needs to know.

Click to expand...

Of course. The part I've marked in red is the key.

My main concern isn't with the labeling of sliders, or even what people say when they are moving one, but rather to be able to be precise when it matters (in technical descriptions where the different factors are being spoken of and need to be unambiguously distinguished).

Thanks for your insight.

Best regards,

Doug

I think it's safe to assume with casual users that "exposure" just be assumed what would happen with different shutterspeed/iso/aperture settings as far as making the image brighter or darker.

If you want precise mathematical changes, then you probably need to use a program that allows for that. We use Nuke at work, you can do just about anything you want, full float, node-based, support for a huge amount of channels and data, changing of color spaces, and easy to script in Python or TCL for if you want to do your own mathematics above and beyond what is already generously given

Doug_Kerr said:

Normally, in context, which of the three meanings (two legitimate, one not really) of "exposure" is meant is usually clear. But it is worthwhile to keep in mind that there are really three different properties that might be involved.

Click to expand...

Hi Doug,

Thanks for taking the time to shed some clarity on this area of loose use of the term "exposure". I fully agree with the distinctions you made. Appreciated.

Next week: the Cherokee name for the red fox.

Click to expand...

Odd as it may seem, I would be iterested in that as well (actually I'm interested in a lot of things, so that might explain ;-) ). Our minds apparently work in similar ways, although you have Carla to help solve the pertinent question, and I don't.

Bart