Is Changing ISO Just Adjusting for Incorrect Sensor Exposure?

[toktik]

After about 20 years of not engaging in much photography, I returned to it about a year ago via Fujifilm. While refreshing my knowledge of photography, I have seen explanations that ISO is part of the "exposure triangle," but that doesn't seem correct to me.

Since exposure happens at the sensor, before ISO gain is applicable, it seems that shutter speed, aperture size, and available light are the components that impact exposure. Isn't changing ISO just a way to help compensate for over or under exposure that happens at the sensor?

From what I have learned about ISO, through reading and use, it seems to function similar to an audio amplifier while available light, shutter speed, and aperture work together to produce the music track (image) being made audible by the amplifier.

Am I misunderstanding something about the function of ISO gain?

 

[mostlyboringphotog]

I'd love for the people that seem to have a deep technical understanding of how digital sensors work to talk about practical strategies to apply the knowledge to actual shooting

I would claim that "how sensors work" is of no importance for 99% of digital photography. If it was important, it'd drive us crazy because every few years a new technology starts being widely incorporated. From the photographer's point of view, today's dual-gain sensors function the same as the ancient CCDs; they just have much better noise performance -- hence more dynamic range -- and much faster readout. [Okay, not exactly the same: CCDs have global shutter, whereas no mass-production CMOS sensor yet does.]

As has always been true, for most digital photography, you select the lowest ISO that gives you acceptable exposure settings (shutter speed and aperture). That's all there is to it.

As has always been true, for Raw-only photography where absolute maximum dynamic range is required, the Expose-to-the-Right technique, perhaps combined with UniWB, is probably your best bet.

The latest hoo-hah is that with some modern sensors, Raw-only photographers could purposely use a lower ISO setting for a given exposure, then brighten in post-processing, with little loss of quality. To which I say: yes, but why would you do that? You can still shoot the straightforward way. And as a bonus for not complicating your photography, you'd get marginally better results. - bold added

Thanks for saying what I've been trying to say in so many posts!!

 

[FingerPainter]

RF gain is very much analogous to the "analog gain"; in that "analog gain", which is set by the ISO dial, increase the photographic sensitivity of a camera as the RF gain increases the "sensitivity" of a SW radio.

Brightening in PP does not increase the photographic sensitivity of a camera same way as increasing the audio volume does not increase the "sensitivity" of a SW radio.

RF gain and analogue gain are the same - they are both accomplished by amplifiers. ... an amplifier is an active component and every active component has a noise figure - that is how much noise it adds to the resulting output signal and it is greater than zero. That's right every amplifier even those on CMOS sensor chips add noise. Once you turn up application (in the case of digital cameras with analogue gain) you increase the signal power and the noise power the same and add noise from the amplifier determined by the noise figure (look that up to). You amplify the signal plus noise in the input and add noise from the amplifier. The more amplification the more noise you normally add.

If is important to consider quantities. The amount of noise added by an amplifier in a digital camera is negligibly small relative to the amount of noise in the signal reaching the amplifier. And it seems to me that any noise added by the amplifier is not itself amplified. The net result is that amplifying a signal in a camera increases the SNR, rather than reduces it. This is simple enough to test. Compare the SNR of an image taken at a given exposure and base ISO against the SNR of an image taken at the same exposure with a higher amount of gain applied. I doubt you will find any camera in which the latter will have a measurably lower SNR, but I know you will find many that have a measurably higher SNR.

 

[Iliah Borg]

But how would you do it differently if the camera was ISO invariant and if it was variant?

Do you know what latitude is in photography?

You have a region of ISO settings, within this region the quality of shadows for different ISO settings is acceptably close. I think of these regions, where they start, where they end, not of ISO-variant or invariant cameras.

I do understand the latitude -

My mental block is that:

Not sure you do.

I would increase the ISO to the highest value that does not clip the high light.

Takes time from other tasks. Why try to control something that doesn't matter for the final result? Can make things worse, too, if intermediate ISO settings are implemented via some trick.

Is this less effective if the camera is ISO invariant?

No need to talk about cameras, only about ranges, or better lists of ISO settings - because of how intermediate ISO settings are implemented, some of them might need to be excluded from the ranges.

And if high light is clipped, I would lower the ISO even if the camera is ISO variant...

If that brings too much noise, I don't do that; and again, especially with the proliferation of dual gain sensors speaking of cameras instead of ranges of ISO settings is giving wrong ideas.

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

 

[Ysarex]

I'd love for the people that seem to have a deep technical understanding of how digital sensors work to talk about practical strategies to apply the knowledge to actual shooting

I would claim that "how sensors work" is of no importance for 99% of digital photography. If it was important, it'd drive us crazy because every few years a new technology starts being widely incorporated. From the photographer's point of view, today's dual-gain sensors function the same as the ancient CCDs; they just have much better noise performance -- hence more dynamic range -- and much faster readout. [Okay, not exactly the same: CCDs have global shutter, whereas no mass-production CMOS sensor yet does.]

As has always been true, for most digital photography, you select the lowest ISO that gives you acceptable exposure settings (shutter speed and aperture). That's all there is to it.

As has always been true, for Raw-only photography where absolute maximum dynamic range is required, the Expose-to-the-Right technique, perhaps combined with UniWB, is probably your best bet.

The latest hoo-hah is that with some modern sensors, Raw-only photographers could purposely use a lower ISO setting for a given exposure, then brighten in post-processing, with little loss of quality. To which I say: yes, but why would you do that?

You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."

You can still shoot the straightforward way.

As you noted; "...where absolute maximum dynamic range is required..." you can not still shoot the straightforward way. Raise the ISO and you're losing DR.

And as a bonus for not complicating your photography, you'd get marginally better results.

 

[mostlyboringphotog]

I'd love for the people that seem to have a deep technical understanding of how digital sensors work to talk about practical strategies to apply the knowledge to actual shooting

I would claim that "how sensors work" is of no importance for 99% of digital photography. If it was important, it'd drive us crazy because every few years a new technology starts being widely incorporated. From the photographer's point of view, today's dual-gain sensors function the same as the ancient CCDs; they just have much better noise performance -- hence more dynamic range -- and much faster readout. [Okay, not exactly the same: CCDs have global shutter, whereas no mass-production CMOS sensor yet does.]

As has always been true, for most digital photography, you select the lowest ISO that gives you acceptable exposure settings (shutter speed and aperture). That's all there is to it.

As has always been true, for Raw-only photography where absolute maximum dynamic range is required, the Expose-to-the-Right technique, perhaps combined with UniWB, is probably your best bet.

The latest hoo-hah is that with some modern sensors, Raw-only photographers could purposely use a lower ISO setting for a given exposure, then brighten in post-processing, with little loss of quality. To which I say: yes, but why would you do that?

You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."

You can still shoot the straightforward way.

As you noted; "...where absolute maximum dynamic range is required..." you can not still shoot the straightforward way. Raise the ISO and you're losing DR.

Losing DR does not affect the image DR if the high light is not clipped. The scene did not have enough DR to begin with.

If high light is clipped you should be using lower ISO, regardless of the ISO invariance or variance.

And as a bonus for not complicating your photography, you'd get marginally better results.

 

[mostlyboringphotog]

But how would you do it differently if the camera was ISO invariant and if it was variant?

Do you know what latitude is in photography?

You have a region of ISO settings, within this region the quality of shadows for different ISO settings is acceptably close. I think of these regions, where they start, where they end, not of ISO-variant or invariant cameras.

I do understand the latitude -

My mental block is that:

Not sure you do.

I would increase the ISO to the highest value that does not clip the high light.

Takes time from other tasks. Why try to control something that doesn't matter for the final result? Can make things worse, too, if intermediate ISO settings are implemented via some trick.

Is this less effective if the camera is ISO invariant?

No need to talk about cameras, only about ranges, or better lists of ISO settings - because of how intermediate ISO settings are implemented, some of them might need to be excluded from the ranges.

And if high light is clipped, I would lower the ISO even if the camera is ISO variant...

If that brings too much noise, I don't do that; and again, especially with the proliferation of dual gain sensors speaking of cameras instead of ranges of ISO settings is giving wrong ideas.

"better lists of ISO settings " Got it - thanks

 

[Ysarex]

I'd love for the people that seem to have a deep technical understanding of how digital sensors work to talk about practical strategies to apply the knowledge to actual shooting

I would claim that "how sensors work" is of no importance for 99% of digital photography. If it was important, it'd drive us crazy because every few years a new technology starts being widely incorporated. From the photographer's point of view, today's dual-gain sensors function the same as the ancient CCDs; they just have much better noise performance -- hence more dynamic range -- and much faster readout. [Okay, not exactly the same: CCDs have global shutter, whereas no mass-production CMOS sensor yet does.]

As has always been true, for most digital photography, you select the lowest ISO that gives you acceptable exposure settings (shutter speed and aperture). That's all there is to it.

As has always been true, for Raw-only photography where absolute maximum dynamic range is required, the Expose-to-the-Right technique, perhaps combined with UniWB, is probably your best bet.

The latest hoo-hah is that with some modern sensors, Raw-only photographers could purposely use a lower ISO setting for a given exposure, then brighten in post-processing, with little loss of quality. To which I say: yes, but why would you do that?

You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."

You can still shoot the straightforward way.

As you noted; "...where absolute maximum dynamic range is required..." you can not still shoot the straightforward way. Raise the ISO and you're losing DR.

Losing DR does not affect the image DR if the high light is not clipped. The scene did not have enough DR to begin with.

Then absolute maximum dynamic range wouldn't be required. The assumption is implicit in the requirement for maximum dynamic range that the scene DR is the reason.

If high light is clipped you should be using lower ISO, regardless of the ISO invariance or variance.

The problem with doing that is if the scene DR is very high and the camera is not ISO invariant. Point being that modern ISO invariant cameras make possible today, because of ISO invariance, the successful capture of images that were not possible with yesterday's non-ISO invariant cameras.

Don't think of it as ignoring ISO think of it as extending usable sensor DR (which requires keeping the ISO at base).

The problem with what you say there hinges on the term "regardless." Regarding a near ISO-invariant sensor I can do as you suggest: lower the ISO, not clip the highlight and get a good photo. But regarding a non ISO-invariant sensor I'll get a dramatically different result if I try that and may in fact crash and burn. And there's a practical strategy gleaned from an understanding of ISO-invariance that can lead you to not shooting the straightforward way.

And as a bonus for not complicating your photography, you'd get marginally better results.

 

[JNR]

Worth reading IMO : https://dslrbodies.com/cameras/camera-articles/image-sensors/is-iso-fake.html

That might be the most well-thought-out article I've read from Thom Hogan, and the comment is not intended as back-handed - although it might sound that way.

Thanks, tokumeino and Iliah.

 

[mostlyboringphotog]

I'd love for the people that seem to have a deep technical understanding of how digital sensors work to talk about practical strategies to apply the knowledge to actual shooting

I would claim that "how sensors work" is of no importance for 99% of digital photography. If it was important, it'd drive us crazy because every few years a new technology starts being widely incorporated. From the photographer's point of view, today's dual-gain sensors function the same as the ancient CCDs; they just have much better noise performance -- hence more dynamic range -- and much faster readout. [Okay, not exactly the same: CCDs have global shutter, whereas no mass-production CMOS sensor yet does.]

As has always been true, for most digital photography, you select the lowest ISO that gives you acceptable exposure settings (shutter speed and aperture). That's all there is to it.

As has always been true, for Raw-only photography where absolute maximum dynamic range is required, the Expose-to-the-Right technique, perhaps combined with UniWB, is probably your best bet.

The latest hoo-hah is that with some modern sensors, Raw-only photographers could purposely use a lower ISO setting for a given exposure, then brighten in post-processing, with little loss of quality. To which I say: yes, but why would you do that?

You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."

You can still shoot the straightforward way.

As you noted; "...where absolute maximum dynamic range is required..." you can not still shoot the straightforward way. Raise the ISO and you're losing DR.

Losing DR does not affect the image DR if the high light is not clipped. The scene did not have enough DR to begin with.

Then absolute maximum dynamic range wouldn't be required. The assumption is implicit in the requirement for maximum dynamic range that the scene DR is the reason.

If high light is clipped you should be using lower ISO, regardless of the ISO invariance or variance.

The problem with doing that is if the scene DR is very high and the camera is not ISO invariant. Point being that modern ISO invariant cameras make possible today, because of ISO invariance, the successful capture of images that were not possible with yesterday's non-ISO invariant cameras.

Don't think of it as ignoring ISO think of it as extending usable sensor DR (which requires keeping the ISO at base).

The problem with what you say there hinges on the term "regardless." Regarding a near ISO-invariant sensor I can do as you suggest: lower the ISO, not clip the highlight and get a good photo. But regarding a non ISO-invariant sensor I'll get a dramatically different result if I try that and may in fact crash and burn. And there's a practical strategy gleaned from an understanding of ISO-invariance that can lead you to not shooting the straightforward way.

As best as I can understand, ISO invariant means not increasing "analog gain" for whatever the reason as in X-T1 (tops out at ISO1600) or adding "analog gain" all the way to ISO25600 in X-T3 but read noise is low such that SNR improvement is minimal.

So, if you have a camera that has very low read noise, then one can "under ISO" and keep more high light latitude (headroom) without much image degradation. However, now your camera DR is reduced by the "under ISO"; the highlight or the most bright area does not reach the full scale value of the ADC.

If your camera has lower read noise, you are going to get less noise in deep shadows when it's pulled up in PP. This fact does not affect how the ISO is to be chosen.

If you want to keep more highlight headroom and keep ISO low, even though the image looks dark, I am guessing that when you took the shot, there was some chance that there may be a bright high light, like sun coming out of the cloud. If so, you would do the same even with ISO variant camera; just cannot pull up the shadow as much or as cleanly.

If such time varying scene brightness is not a concern, I simply do not see the benefit of "under ISO".

Moreover, with Auto ISO, the camera DR can be maximized even under fast varying brightness condition. Are we lucky or what ;-)

Regards,

And as a bonus for not complicating your photography, you'd get marginally better results.

 

[bclaff]

... but read noise is low such that SNR improvement is minimal.

...

It's not low read noise per se but the ratio of downstream to upstream noise.

If downstream noise is low relative to upstream noise then it will be "ISO Invariant".

Think of it this way ... do you need to amplify the upstream noise to overcome the downstream? If so, it's not "invariant".

 

[mostlyboringphotog]

Worth reading IMO : https://dslrbodies.com/cameras/camera-articles/image-sensors/is-iso-fake.html

That might be the most well-thought-out article I've read from Thom Hogan, and the comment is not intended as back-handed - although it might sound that way.

Thanks, tokumeino and Iliah.

Add my thanks, too - very readable article.

 

[Iliah Borg]

... but read noise is low such that SNR improvement is minimal.

...

It's not low read noise per se but the ratio of downstream to upstream noise.

If downstream noise is low relative to upstream noise then it will be "ISO Invariant".

Think of it this way ... do you need to amplify the upstream noise to overcome the downstream? If so, it's not "invariant".

 

[mostlyboringphotog]

... but read noise is low such that SNR improvement is minimal.

...

It's not low read noise per se but the ratio of downstream to upstream noise.

If downstream noise is low relative to upstream noise then it will be "ISO Invariant".

Think of it this way ... do you need to amplify the upstream noise to overcome the downstream? If so, it's not "invariant".

Thanks for that clarification - I tend to say "read noise" for all the electronic noise in the image processing chain. Mea culpa :-(

The "analog gain" also matches the signal level to the full scale input range; it seems this aspect of "analog gain" seems not addressed in general. I may have simply missed it or it's not as significant as I think it does.

I've been reviewing SQNR (signal to quantization noise ratio). What I thought to be ADC SNR of 6dB per bit is really SQNR...

 

[Ysarex]

I'd love for the people that seem to have a deep technical understanding of how digital sensors work to talk about practical strategies to apply the knowledge to actual shooting

I would claim that "how sensors work" is of no importance for 99% of digital photography. If it was important, it'd drive us crazy because every few years a new technology starts being widely incorporated. From the photographer's point of view, today's dual-gain sensors function the same as the ancient CCDs; they just have much better noise performance -- hence more dynamic range -- and much faster readout. [Okay, not exactly the same: CCDs have global shutter, whereas no mass-production CMOS sensor yet does.]

As has always been true, for most digital photography, you select the lowest ISO that gives you acceptable exposure settings (shutter speed and aperture). That's all there is to it.

As has always been true, for Raw-only photography where absolute maximum dynamic range is required, the Expose-to-the-Right technique, perhaps combined with UniWB, is probably your best bet.

The latest hoo-hah is that with some modern sensors, Raw-only photographers could purposely use a lower ISO setting for a given exposure, then brighten in post-processing, with little loss of quality. To which I say: yes, but why would you do that?

You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."

You can still shoot the straightforward way.

As you noted; "...where absolute maximum dynamic range is required..." you can not still shoot the straightforward way. Raise the ISO and you're losing DR.

Losing DR does not affect the image DR if the high light is not clipped. The scene did not have enough DR to begin with.

Then absolute maximum dynamic range wouldn't be required. The assumption is implicit in the requirement for maximum dynamic range that the scene DR is the reason.

If high light is clipped you should be using lower ISO, regardless of the ISO invariance or variance.

The problem with doing that is if the scene DR is very high and the camera is not ISO invariant. Point being that modern ISO invariant cameras make possible today, because of ISO invariance, the successful capture of images that were not possible with yesterday's non-ISO invariant cameras.

Don't think of it as ignoring ISO think of it as extending usable sensor DR (which requires keeping the ISO at base).

The problem with what you say there hinges on the term "regardless." Regarding a near ISO-invariant sensor I can do as you suggest: lower the ISO, not clip the highlight and get a good photo. But regarding a non ISO-invariant sensor I'll get a dramatically different result if I try that and may in fact crash and burn. And there's a practical strategy gleaned from an understanding of ISO-invariance that can lead you to not shooting the straightforward way.

As best as I can understand, ISO invariant means not increasing "analog gain" for whatever the reason as in X-T1 (tops out at ISO1600) or adding "analog gain" all the way to ISO25600 in X-T3 but read noise is low such that SNR improvement is minimal.

So, if you have a camera that has very low read noise, then one can "under ISO"

What is "under ISO"? I'm keeping the ISO at base -- no ISO brightening applied.

and keep more high light latitude (headroom) without much image degradation. However, now your camera DR is reduced by the "under ISO";

Camera DR? I'm talking about the DR captured from the sensor and stored in a raw file. And it is in no way reduced by keeping the ISO at base. Maximum DR from the sensor is available at base ISO. If "camera DR" means something else to you please get back on topic.

the highlight or the most bright area does not reach the full scale value of the ADC.

Of course it does. That's the whole point -- set the exposure to take full advantage of the sensor.

If your camera has lower read noise, you are going to get less noise in deep shadows when it's pulled up in PP. This fact does not affect how the ISO is to be chosen.

I'm choosing the ISO to permit maximum DR capture because the DR in the scene is high and because I am going to get less noise in deep shadows when it's pulled up in PP.

If you want to keep more highlight headroom and keep ISO low, even though the image looks dark, I am guessing that when you took the shot, there was some chance that there may be a bright high light, like sun coming out of the cloud.

Yes, very high scene DR -- the single and only thing I've been talking about and the reason I noted for why ISO invariance has value. That's the topic: in my first sentence in this thread in response to Doug: "You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."" Only talking about that condition ..."where absolute maximum dynamic range is required..." please stay on topic.

If so, you would do the same even with ISO variant camera;

Not necessarily because you cannot pull up the shadow as much or as cleanly. And it can be a really big difference -- enough to make option A) ISO-invariant camera a workable solution and option B) non-ISO-invariant camera and unworkable crash and burn.

just cannot pull up the shadow as much or as cleanly.

If such time varying scene brightness is not a concern,

Not talking about such time -- it's off topic. My whole point for entering this discussion is contained in the first sentence I wrote responding to Doug: "You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."" Not talking about any other condition except where maximum dynamic range is required -- don't get off topic.

I simply do not see the benefit of "under ISO".

Moreover, with Auto ISO, the camera DR can be maximized even under fast varying brightness condition. Are we lucky or what ;-)

Meaningless and off topic.

Regards,

And as a bonus for not complicating your photography, you'd get marginally better results.

 

[Banderras]

I think you don’t understand the information and applicable situations that are discussed in this topic, and for some reason get irritated and lash out on other forum members who are trying to break it down for you. If you read carefully the whole thread, it would help you, sir, stay on topic.

 

[Ysarex]

I think you don’t understand the information and applicable situations that are discussed in this topic,

Could you please be specific.

and for some reason get irritated and lash out on other forum members who are trying to break it down for you. If you read carefully the whole thread, it would help you, sir, stay on topic.

 

[Banderras]

I think you don’t understand the information and applicable situations that are discussed in this topic,

Could you please be specific.

and for some reason get irritated and lash out on other forum members who are trying to break it down for you. If you read carefully the whole thread, it would help you, sir, stay on topic.

TL;DR

Not the topic of this thread: If you have enough light to saturate your sensor then of course use the base iso and get maximum DR of meaningful information.

Topic of this thread: If on the other hand, you do not have enough light to saturate the sensor, then choose appropriate iso to saturate the ADC. This will lead to equal USEFUL DR, avoid posterisation of digital numbers in raw and produce less noise than pushing in post software, apart from some exceptions which were mentioned in this thread.

IMHO

 

[Ysarex]

I think you don’t understand the information and applicable situations that are discussed in this topic,

Could you please be specific.

and for some reason get irritated and lash out on other forum members who are trying to break it down for you. If you read carefully the whole thread, it would help you, sir, stay on topic.

TL;DR

Not the topic of this thread: If you have enough light to saturate your sensor then of course use the base iso and get maximum DR of meaningful information.

Topic of this thread: If on the other hand, you do not have enough light to saturate the sensor, then choose appropriate iso to saturate the ADC. This will lead to equal USEFUL DR, avoid posterisation of digital numbers in raw and produce less noise than pushing in post software, apart from some exceptions which were mentioned in this thread.

IMHO

And how did what I say indicate that I don't understand? I entered this thread very late in order to reply with a very specific response to another post.

Doug said:

As has always been true, for Raw-only photography where absolute maximum dynamic range is required, the Expose-to-the-Right technique, perhaps combined with UniWB, is probably your best bet.

The latest hoo-hah is that with some modern sensors, Raw-only photographers could purposely use a lower ISO setting for a given exposure, then brighten in post-processing, with little loss of quality. To which I say: yes, but why would you do that?

I said:

You noted the reason for doing that in your previous paragraph; "...where absolute maximum dynamic range is required..."

Doug said:

You can still shoot the straightforward way.

I said:

As you noted; "...where absolute maximum dynamic range is required..." you can not still shoot the straightforward way. Raise the ISO and you're losing DR.

 

[mostlyboringphotog]

I think you don’t understand the information and applicable situations that are discussed in this topic,

Could you please be specific.

and for some reason get irritated and lash out on other forum members who are trying to break it down for you. If you read carefully the whole thread, it would help you, sir, stay on topic.

Please, accept my apology as I may have addressed my issues and not yours.

Allow me to try if I got it any better...

You are more concerned with maximizing the camera/sensor DR, which is limited by the ADC bit depth (and the noise floor) and is so when the camera ISO is set to base. I agree.

So, if you are saying, if you had an ISO variant camera, you may increase the ISO to get better shadow detail and may let some parts of the high light go but if you have ISO invariant camera, you would keep ISO lower and keep more of the high light and bring up the shadow in PP, I would appreciate such approach fully.

 

[mostlyboringphotog]

I think you don’t understand the information and applicable situations that are discussed in this topic,

Could you please be specific.

and for some reason get irritated and lash out on other forum members who are trying to break it down for you. If you read carefully the whole thread, it would help you, sir, stay on topic.

TL;DR

Not the topic of this thread: If you have enough light to saturate your sensor then of course use the base iso and get maximum DR of meaningful information.

Topic of this thread: If on the other hand, you do not have enough light to saturate the sensor, then choose appropriate iso to saturate the ADC. - bold added

Thanks for clarifying the topic - I'm afraid I can easily lose the track of where I was...

This will lead to equal USEFUL DR, avoid posterisation of digital numbers in raw and produce less noise than pushing in post software, apart from some exceptions which were mentioned in this thread.

IMHO