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

As my grandfather taught us, for the chosen media and print size, overexposure is when important highlights are unresolved, ugly, or void; underexposure is when important shadows are unresolved, ugly, or blocked; and both over- and underexposure can (and often do) happen at the same time ;)

My definition of the term "exposure" in photography is "the amount of light reaching the film or the amount of light reaching the sensor." (partial quotes from OP)
However, over/under exposure talks about the amount of light hitting the sensor, with over/under meaning above and below a threshold. Overexposure sounds like an exposure that was larger than a neutral value or something that is excessive, maybe even harmful (overexposure to radiation, underexposure to sun).

I like those definitions. The terms define a harmful situation, not a suboptimal situation.

Looking at Google search results, many attach the terms over/under-exposure to the image's brightness, as I did in my post. I agree that is wrong.

It's a good definition IMHO, but I think an important thing to understand is that exposure itself is an artistic decision. For example, consider two photos of a dimly lit person against a bright background. One photographer may use a longer exposure to bring out the face, another may use a shorter exposure to create a silhouette effect. (I use the terms "longer" and "shorter" here for simplicity, of course you can control exposure with more than just shutter speed).

The choice of which shadows or highlights are "important" rests with the photographer, and just because the face is crushed into the blacks does not necessarily mean that the image is "underexposed" in terms of the purpose for which that exposure was intended.

If I understand your definition, a particular exposure might be an "under exposure" if the camera is set to ISO 100, and an "over exposure" if the camera is set to ISO 6400. Even though both exposures are exactly the same.

This suggests that your use of the terms "Over Exposure" and "Under Exposure" are not telling us anything about the exposure itself, but it telling us something about the relationship between the exposure and the ISO setting.

Using over- and under-exposure in this way, seems like it would be confusing to beginners. The terms themselves seem to be describing something about the exposure, when in reality it is describing a relationship between the exposure and something else.

Even worse, if the beginner is willing to adjust the resulting image, they can take a too light JPEG, darken it, and convert an "over-exposed" image to an "under-exposed" image. This introduces some logic problems because they are not changing the exposure itself.

For instance, if you have a light meter, and know the shutter speed and aperture, you can determine the exposure, but you don't have enough information to know whether that will result in an over-exposure or under-exposure.

Obviously, but the intended audience sometimes disagree with the artist ;)

I think the better "word" describing what my grandfather meant would be "under-resolved", which is also a subjective criteria. He would often say that meeting the customer / editor expectations is absolutely necessary.

You seem to be using exposure and image lightness interchangeably which is not technically correct to do.

For example if I set the optimal exposure** for my dof and blur requirements without clipping important highlights and the image is still too dark then imo it is nonsense to say the image is underexposed because no more light could have been allowed to hit the sensor while the shutter was open because the dof and/or the blur requirements would have been messed up.
The image was too dark because the iso was too low, not because the exposure* was too low since it had been maximised within my constraints.

On another note, even using the definition of the word exposure you posted then of two images with the same image lightness, one can be under exposed as shown in the example in my earlier post. This is inconsistent with your comments about over/under exposure which you say are about how light or dark an image looks.

* 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.

I differentiate between exposure and image lightness. I am not sure how you got the impression that I don't.
Yes, I initially claimed that over/underexposure is about brightness, as I saw those terms independently from photographic exposure. However, I renounced that opinion in one of the later posts (the one that you quoted a part of):

Yes, good point. The ISO 6400 shot could cause blown highlights with ISO 6400, qualifying the shot as overexposed.

I would say that the terms over/under-exposure tell us about the quality of the acquired data. So far, that seems the most consistent way to explain them. If you cannot recover highlights, you have overexposure.

I am interested in hearing a more sensible definition.

I disagree.
The most consistent way to describe images that are too light or too dark is to just simply say they are too light or too dark because even by using your posted definition of the word exposure, images can be optimally exposed** and still be too light or too dark as described earlier.

Maybe or maybe not.
You can set the largest or even a smaller exposure* for your dof and blur constraints but then clip highlights because either the camera or the photographer set iso too high.
In this scenario the exposure* did not cause the clipped highlights.

Using the term image lightness when describing an image as being too light or dark is consistent and applicable to all scenarios.

People using over/under exposure when referring to how light or dark an image is often get shown up for their inconsistencies as you are here for various scenarios.

I do not understand your reply. I was not talking about image brightness, and I do not think that over/under-exposure should be about image brightness/lightness.

While exposure is independent of ISO, the terms over/under-exposure seem dependent on ISO. Or is there a definition that would make those terms independent from ISO?
In that vein, the optimal exposure also depends on ISO. Depending on ISO, the optimal exposure is different.

I can change how they initially look by changing the Brightness setting in my Picture Controls or Image Settings. Overexposure of a JPEG is pretty straight forward, just look at it and the histogram and that will tell you all you need to know; OTOH, determining overexposure for a Raw file requires more than the eye-test or a histogram derived from a default setting JPEG.

With the help of other posters, I realized that I was wrong.

You posted your opinion on the most consistent way to describe over/under exposure and I posted why I disagreed with it.

You seem to be very confused and/or very inconsistent because first you say "While exposure is independent of ISO..." and then go on to say "the optimal exposure also depends on ISO" which is the total opposite. You can't have it both ways.

ISO is the very last setting I set after setting my exposure* with aperture and shutter speed. I describe how I set my optimal exposure** in my Why Are My Photos Noisy Thread.

You also say "...the terms over/under-exposure seem dependent on ISO..." which is not necessarily true.

In the A and S modes ISO sets the exposure* for the camera to target for that ISO setting to output an average 18% grey image lightness.
ISO is the relationship between exposure* and image lightness with the aim to output an average 18% grey image.

In manual mode ISO sets the image lightness to output an average 18% grey image lightness to suit the aperture and shutter speed settings (exposure*) set.
Michael explains the various roles of ISO very well in his op .

Agreed. As highlight clipping can be the result of sensor exposure alone, or in conjunction with excessive pre-ADC ISO brightening, there really needs to be a clear distinction made between the two (one is overexposure, and one is ISO clipping). Any discussions about ETTR, making good use of ISO invariance, and/or sorting out how Dynamic Range modes work etc. all rely on having a working understanding of how brightness brought about by sensor exposure is fundamentally different from any sort of brightening after the fact. Without this basic understanding and a clear definition of the terms being thrown about, it is nearly impossible to make any headway trying to explain any of this stuff.

I thought that we agreed that under/over-exposure should not be about image brightness.

Good point. Since the optimal/maximal exposure should be set before the ISO is set, we can view optimal exposure as independent of ISO.

Ditto. I maximize the exposure (aperture and shutter speed) and then set the appropriate ISO (or let the Auto ISO do its job).

I said that in the context of a fixed exposure. If overexposure is a situation where relevant highlights are clipped, it does vary with the selected ISO. The same exposure can cause overexposure or not depending on the ISO selected.

For a well written examination of exposure and its effects on final image appearance, I would recommend the technical paper written by Karl Lang for Adobe titled Rendering the Print: the Art of Photography kindly archived by Andrew Rodney. I assume it's clearly understood by all that increasing or decreasing exposure has a direct effect on latent image brightness encoding (digital) or density (film) and the resulting appearance of the final image.

The fact that final image appearance depends upon and can be altered thru image processing, does not negate the effects of exposure. There are limits to the range and quality of visible image detail, when rendered thru image processing with its various adjustments, which are determined by exposure.

We did but I posted why I disagreed with your earlier statement
"I would say that the terms over/under-exposure tell us about the quality of the acquired data. So far, that seems the most consistent way to explain them."

True for slide film such as Ektachrome. Not exactly true for negatives or for digital data.

Don