Exposure understanding and definition

[riveredger]

Useful definition of exposure is the one that is acceptably universal and helps achieving better quality.

 

[sybersitizen]

When talking about the absence of Exposure Compensation dial, the Zeiss manager explained: "If a photographer is manually controlling shutter speed and aperture, then the only way they have of further affecting exposure is ISO. So they can use the ISO dial as exposure compensation, effectively."

I don't care about the never-ending discussions concerning the precise meaning of exposure or why ISO does not affect exposure. That's a rabbit hole with no bottom simply due to confusion of terms. Unlikely to change anytime soon.

What bothers me there is the idea that an 'Exposure Compensation' control is deemed unnecessary in that camera. The purpose of EC is to override the 'normal' meter reading.

For example, what if you like to shoot at a specific aperture while leaving the shutter speed dial in 'A'? That's aperture priority mode, frequently used by many photographers. If you want to reduce the brightness of a scene from the 'normal' meter reading, how do you do that without an EC dial? Setting the ISO to something specific won't help because the camera will still just choose a shutter speed that leaves the overall 'exposure' (brightness) the same.

Apparently, whenever you want to override the 'normal' meter reading with that camera you will have to pick a specific aperture, a specific shutter speed, and a specific ISO.

I guess if you're lucky maybe you can instead spot meter on something that you're willing to designate as the 'normal' brightness value.

I don't see a PASM dial on this camera ...

There isn't one.

maybe you can select Aperture or Shutter Priority in the menus. Or this camera will work in Manual mode only then exposure compensation is not needed.

There are manual settings on each of the three dials and there are 'A' (Auto) settings on each of the three dials.

I have no objection to that arrangement, but it does not justify the lack of an EC control.

 

[Jerry R]

Exposure is the product of illumination, f stop & shutter speed. Changing "ISO" changes the amplification of the imaging sensor's signal.

 

[jonas ar]

When talking about the absence of Exposure Compensation dial, the Zeiss manager explained: "If a photographer is manually controlling shutter speed and aperture, then the only way they have of further affecting exposure is ISO. So they can use the ISO dial as exposure compensation, effectively."

If you think about it carefully, this statement is absolutely correct.

No, it's not.

If you set the shutter speed and the aperture, then changing the ISO will have exactly the same effect on the resulting image as if you change Exposure Compensation when in one of the auto-exposure modes.

No, that's false. Changing ISO doesn't change how much light is captured, while changing f-stop or shutter speed does. Since signal-to-noise ratio is controlled by how much light is captured, that's a very significant difference.

I am not sure this was the topic of the thread, but this part is correct, I agree with Tom.

Well, this isn't a debate, and the statement is wrong unless you want to change the definition of the word exposure.

The OP just wanted to mention the misuse of the term exposure.

Right...changing ISO does not, by definition, change exposure.

Indeed.

Btw: If I set my (Sony) camera to autoISO, turning the exposure compensation dial changes the ISO, and keeps the EV constant....

 

[hampshirejohn]

The problem as always is terminology. We describe an image that is too dark as "under-exposed" and one that is too light as "over-exposed". Some might argue, but those descriptions are correctly using the word exposure: the exposure that was used for the image was either too low (dark image) or too high (light image) for the sensitivity of the recording medium. Note that this does NOT say that sensitivty affects exposure.

So far as I am aware we do not have a generaly accepted single term that descibes the consequences of changing sensitivity at constant exposure. Words like "lightness" or "brightness" convey some of the idea, but I am not convinced either is entirely what is needed.

 

[Michael Fryd]

Exposure is the product of illumination, f stop & shutter speed. Changing "ISO" changes the amplification of the imaging sensor's signal.

It's not actually "amplification". The sensor essentially counts photons that have passed through a color filter. The JPEG file contains RGB lightness levels.

ISO is more like a grading curve for a high school math test. Whether 25 correct answer is an "A" or an "F" depends on the grading curve. When we change the grading curve we don't say we change the "amplification" because the inputs and outputs are different things (the input is count of correct answers and the output is a grade).

The ISO setting of the camera is like the grading curve. The input and output are different things (photon counts and image lightness). " Should a count of 50,000 photons be a dark area or a light area of the image?

Amplification" is not a good term to describe ISO.

====

The underlying question here is whether or not one should use mental models and/or terminology that are fundamentally wrong to provide a simplistic overview of a process. It's can be tempting to do so, because it can quickly give a beginner a handle on some general rules of thumb. The problem is that if the beginner tries to learn more about the topic, he has the challenge of unlearning the incorrect mental model.

In my opinion, if there is a choice, the better idea is to try to use a correct mental model rather than an incorrect one.

For instance, many beginners are taught that "high ISO causes noise." In my opinion, the better choice is to teach that "low exposures are inherently more noisy" (where "exposure" is the light reaching the sensor). Both explanations get the idea across. However the incorrect model suggests that one can reduce noise by using a lower ISO, and then brightening in post production. This is an incorrect conclusion due to an incorrect mental model.

When it comes to "ISO", my opinion is that the better mental model is that ISO is like a "grading curve". It's the mapping from photons seen to image lightness.

====

Getting back to the main topic, the Senior Product Manager of Zeiss might have been better off saying "...the only way they have of further affecting image lightness is ISO..."

The problem with that is that is raises the question as to why digital cameras use the term "exposure compensation" rather than "lightness compensation"

 

[ZodiacPhoto]

When talking about the absence of Exposure Compensation dial, the Zeiss manager explained: "If a photographer is manually controlling shutter speed and aperture, then the only way they have of further affecting exposure is ISO. So they can use the ISO dial as exposure compensation, effectively."

I don't care about the never-ending discussions concerning the precise meaning of exposure or why ISO does not affect exposure. That's a rabbit hole with no bottom simply due to confusion of terms. Unlikely to change anytime soon.

What bothers me there is the idea that an 'Exposure Compensation' control is deemed unnecessary in that camera. The purpose of EC is to override the 'normal' meter reading.

For example, what if you like to shoot at a specific aperture while leaving the shutter speed dial in 'A'? That's aperture priority mode, frequently used by many photographers. If you want to reduce the brightness of a scene from the 'normal' meter reading, how do you do that without an EC dial? Setting the ISO to something specific won't help because the camera will still just choose a shutter speed that leaves the overall 'exposure' (brightness) the same.

Apparently, whenever you want to override the 'normal' meter reading with that camera you will have to pick a specific aperture, a specific shutter speed, and a specific ISO.

I guess if you're lucky maybe you can instead spot meter on something that you're willing to designate as the 'normal' brightness value.

I don't see a PASM dial on this camera ...

There isn't one.

maybe you can select Aperture or Shutter Priority in the menus. Or this camera will work in Manual mode only then exposure compensation is not needed.

There are manual settings on each of the three dials and there are 'A' (Auto) settings on each of the three dials.

I have no objection to that arrangement, but it does not justify the lack of an EC control.

I see... I guess, this camera encourages you to switch to manual controls if you want to be creative.

In a regular camera, when you are in, say, aperture-priority mode, when you use exposure compensation, shutter speed changes.

With this camera, you will have to move shutter dial from Auto, and find the shutter speed that gives the desired exposure.

Also, if Zeiss could cooperate with Google, Samsung, or Huawei, and use their latest phone camera and AI software, while replacing the sensor and the lens with what is in this camera, the results may be interesting...

--
http://www.zodiacphoto.com

 

[ZodiacPhoto]

Getting back to the main topic, the Senior Product Manager of Zeiss might have been better off saying "...the only way they have of further affecting image lightness is ISO..."

Agree...

The problem with that is that is raises the question as to why digital cameras use the term "exposure compensation" rather than "lightness compensation"

It sis hard to change the terminology that photographers learned for generations, about the "exposure triangle", etc. Or we can say that the word "exposure" has more than one meaning (it certainly does outside of photography).

 

[Iliah Borg]

Useful definition of exposure is the one that is acceptably universal and helps achieving better quality.

What does this sentence actually mean? "...helps achieving better quality "?

It means that a useful definition of exposure helps to predict various technical quality aspects of the outcome of the exposure, instead of implying "high ISO noise" nonsense.

Glad you had no other questions.

 

[Iliah Borg]

It sis hard to change the terminology that photographers learned for generations, about the "exposure triangle",

There are no "generations" behind "exposure triangle", it is a recent thing (appeared circa 2003 and gained some popularity in 2007), and if anything, being recent, it indicates how easy it is to force adoption of new things, whether those things are right or wrong.

--
http://www.libraw.org/

 

[David5833]

There are no "generations" behind "exposure triangle", it is a recent thing (appeared circa 2003...

Surely you are referring to the term, not the concept.

--
https://www.flickr.com/photos/143821723@N06/with/46460832575/

 

[Ellis Vener]

There are no "generations" behind "exposure triangle", it is a recent thing (appeared circa 2003...

Surely you are referring to the term, not the concept.

 

[Iliah Borg]

There are no "generations" behind "exposure triangle", it is a recent thing (appeared circa 2003...

Surely you are referring to the term, not the concept.

The concept is demonstrably wrong (absurd would be a better word). Nothing like it was ever discussed among my colleagues.

The phrase was: for normal density, given film is rated xxx, you need yy exposure (a pair of shutter speed and f-stop, for example). Film speed was never understood as a part of exposure. The concept was density triangle, not exposure triangle; same like now it is default manufacturer rendering lightness triangle, not exposure triangle.

The analogy between high ISO grain in film and high ISO noise in digital is an easy one, but it is a lazy one and the wrong one.

 

[Michael Fryd]

....

The analogy between high ISO grain in film and high ISO noise in digital is an easy one, but it is a lazy one and the wrong one.

While wrong in specifics, there is a similar underlying concept: You need light to get a quality image. When you have low exposures (light reaching the sensor), the image quality suffers.

With film, you need to use higher speed film which typically has larger, more visible, grains in order to deal with low light levels. With digital, the issue is typically shot noise.

Thus "noisier digital images" and "grainier film images" have a conceptual similarity in that they are both the results of low exposures.

However, the underlying mechanisms are not at all the same.

 

[Jerry R]

The sensor only has one ISO (although some new Sony sensors have 2 ISOs). However, changing ISO on a digital sensor amplifies the sensor signal it does not change the ISO of the sensor. Similar to push processing film.

 

[Iliah Borg]

....

The analogy between high ISO grain in film and high ISO noise in digital is an easy one, but it is a lazy one and the wrong one.

While wrong in specifics, there is a similar underlying concept: You need light to get a quality image.

While the concept is correct and is the same for film and digital (not just similar, but exactly the same; to get a better recording for a given signal chain / processing and recording media, to start with a better signal is an obvious choice), the analogy is not about this concept.

What they are saying is, "just like with faster film speed, higher ISO in digital results in more noise" (and that's from numerous discussions I witnessed during 20+ years).

Thus the analogy is lazy and wrong, it implies that the root cause is ISO speed, substituting the reason (low exposure, simple enough reason) with the likely consequence (high ISO).

This consequence starts to live the life of its own. The reason and the consequence are now out of joint; the consequence is perceived as the reason in itself. As a result, I frequently, 2-3 times a week, receive emails mentioning in passing, as if it is some established knowledge: "I underexposed, keeping ISO low, to decrease noise".

That's terrible.

As you well know: in digital, higher ISO setting doesn't bring "more noise" on its own (raising ISO speed may even improve SNR, signal-to-noise ratio, resulting in what is often called "less noise"). Lowering ISO setting while keeping exposure the same of course doesn't help with noise, and may make the situation with noise visibly worse.

Few notes:

Grain strongly depends on emulsion. I can "overexpose" the film, pull it, and still have more grain than I would have if I started from a lower speed film and matched processing.

Effect is the same, but quantifying effects helps better understanding and putting things into perspective.

For a certain ISO 200 colour negative film, through 12 stops of exposure (0.6D to 2.6D on film), StdDev for D changes from 0.003 (mean 0.5D) to 0.009 (mean 0.9D); flat 0.0055 (mean 0.8D) for 6+ stops upper portion of the sensitometric curve. How strong does the input of photon shot noise look here?

With digital, the issue is typically shot noise.

Thanks for telling me, but Kodak explained and demonstrated it to us couple of decades ago. One Kodak guy even forced us to check it in a direct experiment

--
http://www.libraw.org/

 

[Tom Axford]

The sensor only has one ISO (although some new Sony sensors have 2 ISOs). However, changing ISO on a digital sensor amplifies the sensor signal it does not change the ISO of the sensor. Similar to push processing film.

Camera manufacturers normally give the ISO of the camera, not of the sensor.

"The standard specifies the measurement of light sensitivity of the entire digital camera system and not of individual components such as digital sensors, although Kodak has reported[65] using a variation to characterize the sensitivity of two of their sensors in 2001".

-- from the Wikipedia article on ISO.

 

[Ellis Vener]

The sensor only has one ISO (although some new Sony sensors have 2 ISOs). However, changing ISO on a digital sensor amplifies the sensor signal it does not change the ISO of the sensor. Similar to push processing film.

Camera manufacturers normally give the ISO of the camera, not of the sensor.

"The standard specifies the measurement of light sensitivity of the entire digital camera system and not of individual components such as digital sensors, although Kodak has reported[65] using a variation to characterize the sensitivity of two of their sensors in 2001".

Yes: sensor plus processor and programming.

 

[FingerPainter]

The problem as always is terminology. We describe an image that is too dark as "under-exposed" and one that is too light as "over-exposed". Some might argue, but those descriptions are correctly using the word exposure:

It is entirely possible to have an image that is too light but underexposed.

the exposure that was used for the image was either too low (dark image) or too high (light image) for the sensitivity of the recording medium.

Given that

• the ISO setting generally doesn't affect the sensitivity of the sensor,
• the user can set both the output sensitivity of the camera system and the exposure
• the lightness of the image depends on both those user-controllable factors

why must it necessarily be that an image is too bright because the the exposure was too high for the sensitivity of the camera, rather than the sensitivity of the camera system was set too high for the exposure?

When you take an image at 1/500 f/4 ISO 1600 that is two stops too light, the problem isn't that the exposure is too high. The ISO is too high. So that image is not over-exposed. It is too light, or over-brightened. If we could have taken the image at 1/125 without unwanted motion blur, the image is actually underexposed.

Note that this does NOT say that sensitivity affects exposure.

So far as I am aware we do not have a generaly accepted single term that descibes the consequences of changing sensitivity at constant exposure. Words like "lightness" or "brightness" convey some of the idea, but I am not convinced either is entirely what is needed.

What do you think is missing?

Image lightness is affected by two factors: exposure and ISO setting. I'd suggest that what is missing from "too light" in this case is the information regarding which of the two factors was needlessly high. However, "too light" is better then "overexposed" in this case, because it does not point the finger at the wrong factor.

 

[kiwi2]

The problem as always is terminology. We describe an image that is too dark as "under-exposed" and one that is too light as "over-exposed". Some might argue, but those descriptions are correctly using the word exposure:

It is entirely possible to have an image that is too light but underexposed.

the exposure that was used for the image was either too low (dark image) or too high (light image) for the sensitivity of the recording medium.

Given that

• the ISO setting generally doesn't affect the sensitivity of the sensor,
• the user can set both the output sensitivity of the camera system and the exposure
• the lightness of the image depends on both those user-controllable factors

why must it necessarily be that an image is too bright because the the exposure was too high for the sensitivity of the camera, rather than the sensitivity of the camera system was set too high for the exposure?

When you take an image at 1/500 f/4 ISO 1600 that is two stops too light, the problem isn't that the exposure is too high. The ISO is too high. So that image is not over-exposed. It is too light, or over-brightened. If we could have taken the image at 1/125 without unwanted motion blur, the image is actually underexposed.

Note that this does NOT say that sensitivity affects exposure.

So far as I am aware we do not have a generaly accepted single term that descibes the consequences of changing sensitivity at constant exposure. Words like "lightness" or "brightness" convey some of the idea, but I am not convinced either is entirely what is needed.

What do you think is missing?

Image lightness is affected by two factors: exposure and ISO setting. I'd suggest that what is missing from "too light" in this case is the information regarding which of the two factors was needlessly high. However, "too light" is better then "overexposed" in this case, because it does not point the finger at the wrong factor.

Wow. Amazing.