The ISO argument

[Derrel]

"Exposure" vs. "camera exposure settings/parameters"... maybe we should take some time and define those terms more precisely?

 

[dennyr]

Pardon my ignorance, i only know a little about film and basically nothing about "digital" photography....is the point of all this to say that ASA (numbers) in film mean something different than in digital.?
The size and shape of the grain, as it relates to light sensitivity, is a physical attribute.?
But a digital sensor does not change its sensitivity...like film does. The ASA numbers in the digital world change due to Amplification/Gain of whatever the sensor sees.?

 

[Ysarex]

Pardon my ignorance, i only know a little about film and basically nothing about "digital" photography....is the point of all this to say that ASA (numbers) in film mean something different than in digital.?

The OP in this thread was trying to make a point that ISO is a causal determinant of exposure. On top of that is a general misunderstanding of just what ISO is and does in a digital camera. There's a lot on misunderstanding about both and unfortunately even the camera manufacturers feed into it with some of the beginner-focused material they provide.

ISO in film and digital are different. The standards for both provided by ISO are different. However their use in taking photographs is very similar which of course bolsters the misunderstandings.

The size and shape of the grain, as it relates to light sensitivity, is a physical attribute.?

It is.

But a digital sensor does not change its sensitivity...like film does. The ASA numbers in the digital world change due to Amplification/Gain of whatever the sensor sees.?

You're first sentence above is correct: a digital sensor does not/can not change its light sensitivity, and yes film really does. Your second sentence needs some work.

In digital ISO does not define the engineering methodology used and so we need to be careful with wording so as not to suggest as much. Amplification/gain is involved in some of the engineering options used but so is what you referred to as "whatever." The ISO standard for digital does not concern itself with the "whatever" of how but only with the end result. There was an older thread a week or so ago that likely started all this: ISO is not real In Digital Camra's and in that thread TimT provided an excellent definition of what ISO is in digital cameras. I'll not presume to do better and so to quote TimT: "ISO on a camera is simply a calibration of the middle grey tone to an RGB co-ordinate."

In digital cameras the ISO value specifies the brightness of the output target sRGB JPEG created by the camera processing software relative to the input.

Here's a short description of what ISO on a digital camera does:

1. It biases the camera metering system. When you change the ISO number on a digital camera the meter responds to that change and recalculates the exposure. If set at ISO 100 your camera meter calculated an exposure of 1/250 sec., f/8 then at ISO 400 it would recalculate an exposure two stops less: 1/250 sec., f/16 or 1/1000 sec., f/8 or 1/500 sec., f/11. Exactly like film.
2. Unlike film the digital sensor can't alter it's light sensitivity. Assume the exposure at ISO 100 of 1/250 sec., f/8 was ideal for the sensor. If you raise the ISO and the meter calculates a reduced exposure then you're reducing the exposure to the sensor. The sensor receives a less than ideal exposure -- it is under utilized. Regardless ISO specifies the required brightness level for the output JPEG created by the camera processing software and so the engineers who designed and built the camera have to deal with the reduced sensor exposure so as to render the required output brightness.
3. As noted implementation is not specified in the standard and the engineers have options. The sensor produces a scaled response to light by generating an electrical voltage. Photons in -> electrons out. More photons in -> more electrons out. That voltage signal from the sensor is analog not digital. One option used by many cameras is to amplify the sensor signal (gain) before it is converted to numbers. The sensor signal is on it's way to the camera's ADC (Analog to Digital Converter). Signal amplification works very well but does require hardware to implement. In recent years the sensor manufactures have gotten so good that some of the benefits of hardware amplification have been rendered moot. It's easier (no hardware) to just multiply the numbers in the ADC and now that's becoming a very common implementation method. Many cameras (both of mine) us a combination of methods. Another method is to just flag the EXIF data with a value for the raw file converter that says brighten image by X amount. All are valid and applied methods under the standard.
4. Consequences: It's always a compromise. Under utilizing the sensor means the SNR drops. That's Signal to Noise Ratio. Full use of the sensor produces the strongest signal and a strong signal gives us the best image IQ. As the SNR drops the image IQ suffers. The image becomes noisy because we're relying on the weaker part of the signal. Most ISO implementations that brighten the signal do so either before or during ADC which means they're hard baked into the raw file. Boosting the signal up then reduces overall dynamic range and so each ISO increase typically means DR decrease.

Take aways: Compared with film digital is awesome. The newest crop of sensors are flat out jaw dropping in what they can produce. I can have either 10 usable stops of good clean data at low ISO or amazingly reasonable barely noisy photos at ISO values of 12800 -- that's six full stops of reduced exposure above base ISO for my camera and I can still get an excellent photo!! Damn awesome!

Joe

 

[MTHall720]

There is a lot more to this than I had any idea. As an old film person I had thought getting started in digital that ISO was kind of like ASA.

 

[dennyr]

I get the gist..... Thank You

 

[Ysarex]

There is a lot more to this than I had any idea. As an old film person I had thought getting started in digital that ISO was kind of like ASA.

In practice it can seem very much like film ISO/ASA but under the hood there are some substantial differences. Most notably that you can change ISO on a digital camera with the twist of a dial on a per shot basis.

This thread was started because the OP wanted to provide obvious evidence that ISO in part determines exposure -- a point of common confusion. We live with a "social phenomenon" (not sure what else to call it) in photography that can be a source of confusion. It's a really good thing that photography is practiced by so many people and it is likewise a good and welcome thing that many enthusiasts/amateurs get involved in photo. This large amount of non-professional participation however bolstered by the rapid communication options of the Internet can spread misinformation at an alarming rate (politics?). A term with a specific meaning or a process or phenomenon can become colloquially misunderstood and that misunderstanding can spread so that effectively the term-meaning or phenomenon gets changed. You'll run into the term bokeh at some point if you keep up your interest. What should we do now with bokeh? Should it mean what it originally meant or should we accept what it means now in the minds of the Youtube misinformation hordes?

So "exposure" likewise has a colloquial meaning that deviates from it's formal meaning in our discipline. Colloquially "exposure" means how light or dark your photo appears and when taking a photo with a camera you can alter that light/dark appearance by changing shutter speed, f/stop, or ISO. Therefore the factors effecting exposure must be shutter speed, f/stop and ISO. And then you get this:

(Notice the misuse of the word bokeh in that diagram -- it's a twofer!) The problem with that is ISO doesn't really causally effect exposure. It wouldn't be a big deal if it didn't matter but it does. And it's because it matters that our industry long ago defined exposure differently than the colloquial understanding. Exposure is the amount of light per unit area that strikes the film/sensor. There are three causal factors: the intensity of illumination in the scene, the time of the exposure (shutter speed) and attenuation of the light through the lens (f/stop).

So how does ISO fit in? It's obviously involved. You have to consider it when you make an exposure. Why isn't it a factor then? I explain it in class like this: We're in the lab and I can literally walk over to the sink and do this and I do. Turn on the water and let's call that light coming through the camera. I pick up an 500ml beaker and hold it under the faucet -- that's ISO. Turning the faucet on/off is the shutter speed and how much I turn the faucet is f/stop. Let's assume I can do this with precision. I turn the faucet on for two seconds with a 1/8 turn of the handle. I successfully fill the beaker. How much water do I have? That's the exposure and it's 500ml. Now I switch to a 1 liter beaker and make another exposure. I turn the faucet on for two seconds with a 1/8 turn of the handle. How much water do I have? 500ml just like the last exposure I used the same shutter speed and f/stop. Did changing the volume of the beaker change the exposure? No.

In practice however because ISO does in fact change the light/dark appearance of our final photo and because if we use the camera meter (nearly all of us do) then changing ISO causes the meter to recalculate exposure it certainly seems like ISO is an exposure factor. The triangle diagram above is commonly used and no question it helps beginners get a handle on using the camera. But it muddies the cause and effect of what's going on and that can be bad. Pretty soon you encounter beginners who believe that ISO is the cause of noise in a photo (it's not) and that misconception causes them to alter their behavior using the camera. The primary cause of noise in our photos is reduced exposure. What ISO does in fact tends to suppress noise. So caught up in a misunderstanding (and I see this in class all the time) a beginner resists raising ISO for fear of noise when the ISO function would have in fact made their photo less noisy. Understanding the cause of the noise is the first step in being able to control it.

Joe

 

[dennyr]

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

 

[Ysarex]

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

All else is not equal and it would not be possible to take those two photos while maintaining all else equal. It's very unlikely as it would require an atypical use of the camera. It's possible however because the cause of noise isn't ISO directly. The noise we see in digital photos is primarily a function of exposure.

However ISO is indirectly linked to exposure so that when we raise ISO it's because we typically want to (need to) reduce exposure. So ISO is certainly linked to noise because of how we use it.

Let's look at a less extreme example that is typical of what my students do. Here's three photos (I had ginger snaps with my coffee this morning).

I adjusted them to all appear the same exposure but out of the camera the first was too dark the middle was right and the last was too light. The first is taken at ISO 1600 and in the manner of one of my students it is underexposed. ISO 3200 was indicated by the meter but my students don't want noise so they often resist raising the ISO. They believe the noise comes from the ISO increase. The first two photos are both exposed 1/50th sec. f/4.

Now let's look at the noise in the three photos. This is a 100% enlargement of a section. I have noise filtering in the software turned off for all.

The left photo is taken at ISO 1600 1/50th sec. f/4. The middle photo is taken at ISO 3200 1/50th sec. f/4. The noise is basically the same in both with a very slight improvement showing in the ISO 3200 photo. Typically what ISO does in the camera suppresses noise. My students are making a mistake and they should raise the ISO to the level needed -- the noise is from the exposure. Now the third, the right hand photo has less noise. It's noticeably better and it is also taken at ISO 3200. What's different about the right hand photo is I changed the exposure to 1/20th sec. f/4. It doesn't seem like much but that's more than doubling the amount of light reaching the sensor and it shows. That exposure change however of course highlights the entire dilemma to begin with: and I used a tripod as if I could convince my students to get and use a tripod. 1/20th sec. is not hand holdable -- ISO is often about motion.

So all things not being equal no one is going to reduce exposure for an ISO 200 photo by more than 4 stops to show that it's then noisier than an ISO 3200 photo properly exposed.

Joe

 

[Grandpa Ron]

It has been my observation that most of the discussion around ISO is more of a lesson in semantics then photography.

We all agree that given a higher ISO; the digital picture picks up noise, the film picture picks up grain.

It is also fairly well established that the switching from a low to high digital ISO, while keeping the proper exposure, introduces some degree of change in the picture, because the electronic are not perfect. So to film, a high ISO film or pushing a low ISO negative during development, does not produce the same result as standard film processing, regardless of the exposure. This comes as no surprise to those remember the fickle nature of photochemistry.

My point is, this is all a mind game. We select an ISO that we are happy with, it makes no difference what ISO the other person prefers.

In fact, the choice of ISO is usually dictated by circumstance. We need a high ISO to capture a speeding race car and a slow ISO to blur the image of moving water. However, given the increasing power of post processing, even these old "rules of thumb" are becoming a moot point.

The importance of subtle changes in camera performance is mostly opinion. They have been there forever and will be there forever. I put it on a par with trying to find fly specs in pepper.

 

[eja]

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

All else is not equal and it would not be possible to take those two photos while maintaining all else equal. It's very unlikely as it would require an atypical use of the camera. It's possible however because the cause of noise isn't ISO directly. The noise we see in digital photos is primarily a function of exposure.

However ISO is indirectly linked to exposure so that when we raise ISO it's because we typically want to (need to) reduce exposure. So ISO is certainly linked to noise because of how we use it.

Let's look at a less extreme example that is typical of what my students do. Here's three photos (I had ginger snaps with my coffee this morning).

View attachment 171206 View attachment 171206 View attachment 171208

I adjusted them to all appear the same exposure but out of the camera the first was too dark the middle was right and the last was too light. The first is taken at ISO 1600 and in the manner of one of my students it is underexposed. ISO 3200 was indicated by the meter but my students don't want noise so they often resist raising the ISO. They believe the noise comes from the ISO increase. The first two photos are both exposed 1/50th sec. f/4.

Now let's look at the noise in the three photos. This is a 100% enlargement of a section. I have noise filtering in the software turned off for all.

View attachment 171209

The left photo is taken at ISO 1600 1/50th sec. f/4. The middle photo is taken4 at ISO 3200 1/50th sec. f/4. The noise is basically the same in both with a very slight improvement showing in the ISO 3200 photo. Typically what ISO does in the camera suppresses noise. My students are making a mistake and they should raise the ISO to the level needed -- the noise is from the exposure. Now the third, the right hand photo has less noise. It's noticeably better and it is also taken at ISO 3200. What's different about the right hand photo is I changed the exposure to 1/20th sec. f/4. It doesn't seem like much but that's more than doubling the amount of light reaching the sensor and it shows. That exposure change however of course highlights the entire dilemma to begin with: and I used a tripod as if I could convince my students to get and use a tripod. 1/20th sec. is not hand holdable -- ISO is often about motion.

So all things not being equal no one is going to reduce exposure for an ISO 200 photo by more than 4 stops to show that it's then noisier than an ISO 3200 photo properly exposed.

Joe

So, can a safe take away from this discussion be as follows?

Don't underexpose, since that is the cause of the noise. It doesn't matter what ISO is set, just ensure you set the correct exposure?

 

[Ysarex]

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

All else is not equal and it would not be possible to take those two photos while maintaining all else equal. It's very unlikely as it would require an atypical use of the camera. It's possible however because the cause of noise isn't ISO directly. The noise we see in digital photos is primarily a function of exposure.

However ISO is indirectly linked to exposure so that when we raise ISO it's because we typically want to (need to) reduce exposure. So ISO is certainly linked to noise because of how we use it.

Let's look at a less extreme example that is typical of what my students do. Here's three photos (I had ginger snaps with my coffee this morning).

View attachment 171206 View attachment 171206 View attachment 171208

I adjusted them to all appear the same exposure but out of the camera the first was too dark the middle was right and the last was too light. The first is taken at ISO 1600 and in the manner of one of my students it is underexposed. ISO 3200 was indicated by the meter but my students don't want noise so they often resist raising the ISO. They believe the noise comes from the ISO increase. The first two photos are both exposed 1/50th sec. f/4.

Now let's look at the noise in the three photos. This is a 100% enlargement of a section. I have noise filtering in the software turned off for all.

View attachment 171209

The left photo is taken at ISO 1600 1/50th sec. f/4. The middle photo is taken4 at ISO 3200 1/50th sec. f/4. The noise is basically the same in both with a very slight improvement showing in the ISO 3200 photo. Typically what ISO does in the camera suppresses noise. My students are making a mistake and they should raise the ISO to the level needed -- the noise is from the exposure. Now the third, the right hand photo has less noise. It's noticeably better and it is also taken at ISO 3200. What's different about the right hand photo is I changed the exposure to 1/20th sec. f/4. It doesn't seem like much but that's more than doubling the amount of light reaching the sensor and it shows. That exposure change however of course highlights the entire dilemma to begin with: and I used a tripod as if I could convince my students to get and use a tripod. 1/20th sec. is not hand holdable -- ISO is often about motion.

So all things not being equal no one is going to reduce exposure for an ISO 200 photo by more than 4 stops to show that it's then noisier than an ISO 3200 photo properly exposed.

Joe

So, can a safe take away from this discussion be as follows?

Don't underexpose, since that is the cause of the noise. It doesn't matter what ISO is set, just ensure you set the correct exposure?

Expose the sensor as much as possible. That's the bottom line. Everything about IQ improves with more exposure up to the full capacity of the sensor. There are different causes of noise in a digital camera but the dominant noise source is what we call shot noise and that's a function of exposure -- the common noise we see is caused by reduced exposure.

You have to deal with the pragmatic requirements of taking the photo and you should take the photo even if you have to reduce exposure to do so. ISO plays a role in that raising ISO biases the light meter to calculate a reduced exposure and so the correlation that folks see is real but causation is spurious. So if you raise ISO and take the exposure the meter indicates you'll get more noise because you reduced exposure. If you need a faster shutter speed that's going to reduce exposure and that will cause noise.

Whatever the given circumstance, expose the sensor as much as possible.

Joe

 

[Grandpa Ron]

I think the better take way from this discussion is to remember photography is a personal expression.

It is important to understand the effects of changes to ISO, depth of field, shutter speed etc. and it is interesting to read others opinions on the subject. It adds to your knowledge base and helps build your photographic techniques for various settings.

However, whether the picture has too much noise, or too much or too little depth of field, is too soft a focus or a host of other variables; depends on the person taking the picture. Even "proper exposure" has some latitude.

 

[EventArtist]

In general, people seem to tolerate noisy/muddy shadows more than blown-out highlights; so, shooting digital is more like shooting slide-film than print-film: if in doubt then (slightly!) adjust to underexpose, don't adjust to overexpose.

If that bothers you, then bracket your settings until you become more comfortable with your (meter’s) approach to camera settings in your typical lighting conditions.

• How accurate is your light measurement?

You must decide what’s important in your subject composition, and you should know how (in-)accurate your (meter’s) choice of camera settings may be.

Ideally, you neither under- nor over-expose; rather, you match the dynamic range of (the important parts of) your subject composition to the dynamic range of the camera—using aperture and shutter (and filters, etc.) to control the *AMOUNT* of light hitting the sensor PLUS using “ISO” to control/bias/adjust the *RESPONSE* to that light by the sensor.

• Rarely “ideal”?

Things are rarely “ideal” so you have to decide how to “properly” record the “important” parts of your subject composition—and in a rush without bracketing I’d recommend trying for “correct” and leaning toward “under”—an approach which should bring all the (important) parts of your image within */- 1 f/stop, easily within acceptable adjust/burn/dodge range with a lean toward protecting highlights over shadows.

• Just what works for me.

Your mileage may vary, depending on how your camera electronics’ ISO “bakes into” a raw file “recording” of the sensor’s response to light.

Of course, with JPEG files (but not relevant to raw files) beware of your camera’s “high ISO NoiseReduction” and especially any “auto-dynamic” (“Active D-Lighting” etc.) options which can affect your image if you don’t pay attention to what it’s doing to alter your image.

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

All else is not equal and it would not be possible to take those two photos while maintaining all else equal. It's very unlikely as it would require an atypical use of the camera. It's possible however because the cause of noise isn't ISO directly. The noise we see in digital photos is primarily a function of exposure.

However ISO is indirectly linked to exposure so that when we raise ISO it's because we typically want to (need to) reduce exposure. So ISO is certainly linked to noise because of how we use it.

Let's look at a less extreme example that is typical of what my students do. Here's three photos (I had ginger snaps with my coffee this morning).

View attachment 171206 View attachment 171206 View attachment 171208

I adjusted them to all appear the same exposure but out of the camera the first was too dark the middle was right and the last was too light. The first is taken at ISO 1600 and in the manner of one of my students it is underexposed. ISO 3200 was indicated by the meter but my students don't want noise so they often resist raising the ISO. They believe the noise comes from the ISO increase. The first two photos are both exposed 1/50th sec. f/4.

Now let's look at the noise in the three photos. This is a 100% enlargement of a section. I have noise filtering in the software turned off for all.

View attachment 171209

The left photo is taken at ISO 1600 1/50th sec. f/4. The middle photo is taken4 at ISO 3200 1/50th sec. f/4. The noise is basically the same in both with a very slight improvement showing in the ISO 3200 photo. Typically what ISO does in the camera suppresses noise. My students are making a mistake and they should raise the ISO to the level needed -- the noise is from the exposure. Now the third, the right hand photo has less noise. It's noticeably better and it is also taken at ISO 3200. What's different about the right hand photo is I changed the exposure to 1/20th sec. f/4. It doesn't

So, can a safe take away from this discussion be as follows?

Don't underexpose, since that is the cause of the noise. It doesn't matter what ISO is set, just ensure you set the correct exposure?

Expose the sensor as much as possible. That's the bottom line. Everything about IQ improves with more exposure up to the full capacity of the sensor. There are different causes of noise in a digital camera but the dominant noise source is what we call shot noise and that's a function of exposure -- the common noise we see is caused by reduced exposure.

Joe

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

All else is not equal and it would not be possible to take those two photos while maintaining all else equal. It's very unlikely as it would require an atypical use of the camera. It's possible however because the cause of noise isn't ISO directly. The noise we see in digital photos is primarily a function of exposure.

However ISO is indirectly linked to exposure so that when we raise ISO it's because we typically want to (need to) reduce exposure. So ISO is certainly linked to noise because of how we use it.

Let's look at a less extreme example that is typical of what my students do. Here's three photos (I had ginger snaps with my coffee this morning).

View attachment 171206 View attachment 171206 View attachment 171208

I adjusted them to all appear the same exposure but out of the camera the first was too dark the middle was right and the last was too light. The first is taken at ISO 1600 and in the manner of one of my students it is underexposed. ISO 3200 was indicated by the meter but my students don't want noise so they often resist raising the ISO. They believe the noise comes from the ISO increase. The first two photos are both exposed 1/50th sec. f/4.

Now let's look at the noise in the three photos. This is a 100% enlargement of a section. I have noise filtering in the software turned off for all.

View attachment 171209

The left photo is taken at ISO 1600 1/50th sec. f/4. The middle photo is taken4 at ISO 3200 1/50th sec. f/4. The noise is basically the same in both with a very slight improvement showing in the ISO 3200 photo. Typically what ISO does in the camera suppresses noise. My students are making a mistake and they should raise the ISO to the level needed -- the noise is from the exposure. Now the third, the right hand photo has less noise. It's noticeably better and it is also taken at ISO 3200. What's different about the right hand photo is I changed the exposure to 1/20th sec. f/4. It doesn't seem like much but that's more than doubling the amount of light reaching the sensor and it shows. That exposure change however of course highlights the entire dilemma to begin with: and I used a tripod as if I could convince my students to get and use a tripod. 1/20th sec. is not hand holdable -- ISO is often about motion.

So all things not being equal no one is going to reduce exposure for an ISO 200 photo by more than 4 stops to show that it's then noisier than an ISO 3200 photo properly exposed.

Joe

So, can a safe take away from this discussion be as follows?

Don't underexpose, since that is the cause of the noise. It doesn't matter what ISO is set, just ensure you set the correct exposure?

Expose the sensor as much as possible. That's the bottom line. Everything about IQ improves with more exposure up to the full capacity of the sensor. There are different causes of noise in a digital camera but the dominant noise source is what we call shot noise and that's a function of exposure -- the common noise we see is caused by reduced exposure.

You have to deal with the pragmatic requirements of taking the photo and you should take the photo even if you have to reduce exposure to do so. ISO plays a role in that raising ISO biases the light meter to calculate a reduced exposure and so the correlation that folks see is real but causation is spurious. So if you raise ISO and take the exposure the meter indicates you'll get more noise because you reduced exposure. If you need a faster shutter speed that's going to reduce exposure and that will cause noise.

Whatever the given circumstance, expose the sensor as much as possible.

Joe

 

[Ysarex]

I think the better take way from this discussion is to remember photography is a personal expression. It is important to understand the effects of changes to ISO, depth of field, shutter speed etc. and it is interesting to read others opinions on the subject. It adds to your knowledge base and helps build your photographic techniques for various settings.

However, whether the picture has too much noise, or too much or too little depth of field, is too soft a focus or a host of other variables; depends on the person taking the picture. Even "proper exposure" has some latitude.

This is not a thread about personal expression. It's a thread about "how it works." It's not about how much noise is too much noise it's about what causes the noise. I'm sitting here just now listening to James Galway playing the flute. I'll bet he has spent long hours over the course of his life learning to play the flute and practicing playing the flute during which times he wasn't concentrating on how best to express Mouquet's "La Flute de Pan" -- that goes without saying. In what way does personal expression change "how it works"? Those who understand "how it works" are best prepared to manipulate the process for personal expression. That's the point.

Joe

 

[Ysarex]

In general, people seem to tolerate noisy/muddy shadows more than blown-out highlights; so, shooting digital is more like shooting slide-film than print-film: if in doubt then (slightly!) adjust to underexpose, don't adjust to overexpose.

If that bothers you, then bracket your settings until you become more comfortable with your (meter’s) approach to camera settings in your typical lighting conditions.

• How accurate is your light measurement?

You must decide what’s important in your subject composition, and you should know how (in-)accurate your (meter’s) choice of camera settings may be.

Ideally, you neither under- nor over-expose; rather, you match the dynamic range of (the important parts of) your subject composition to the dynamic range of the camera—using aperture and shutter (and filters, etc.) to control the *AMOUNT* of light hitting the sensor PLUS using “ISO” to control/bias/adjust the *RESPONSE* to that light by the sensor.

What is *RESPONSE* and in what way does ISO control/bias/adjust the sensor's response to light?

Joe

• Rarely “ideal”?

Things are rarely “ideal” so you have to decide how to “properly” record the “important” parts of your subject composition—and in a rush without bracketing I’d recommend trying for “correct” and leaning toward “under”—an approach which should bring all the (important) parts of your image within */- 1 f/stop, easily within acceptable adjust/burn/dodge range with a lean toward protecting highlights over shadows.

• Just what works for me.

Your mileage may vary, depending on how your camera electronics’ ISO “bakes into” a raw file “recording” of the sensor’s response to light.

Of course, with JPEG files (but not relevant to raw files) beware of your camera’s “high ISO NoiseReduction” and especially any “auto-dynamic” (“Active D-Lighting” etc.) options which can affect your image if you don’t pay attention to what it’s doing to alter your image.

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

All else is not equal and it would not be possible to take those two photos while maintaining all else equal. It's very unlikely as it would require an atypical use of the camera. It's possible however because the cause of noise isn't ISO directly. The noise we see in digital photos is primarily a function of exposure.

However ISO is indirectly linked to exposure so that when we raise ISO it's because we typically want to (need to) reduce exposure. So ISO is certainly linked to noise because of how we use it.

Let's look at a less extreme example that is typical of what my students do. Here's three photos (I had ginger snaps with my coffee this morning).

View attachment 171206 View attachment 171206 View attachment 171208

I adjusted them to all appear the same exposure but out of the camera the first was too dark the middle was right and the last was too light. The first is taken at ISO 1600 and in the manner of one of my students it is underexposed. ISO 3200 was indicated by the meter but my students don't want noise so they often resist raising the ISO. They believe the noise comes from the ISO increase. The first two photos are both exposed 1/50th sec. f/4.

Now let's look at the noise in the three photos. This is a 100% enlargement of a section. I have noise filtering in the software turned off for all.

View attachment 171209

The left photo is taken at ISO 1600 1/50th sec. f/4. The middle photo is taken4 at ISO 3200 1/50th sec. f/4. The noise is basically the same in both with a very slight improvement showing in the ISO 3200 photo. Typically what ISO does in the camera suppresses noise. My students are making a mistake and they should raise the ISO to the level needed -- the noise is from the exposure. Now the third, the right hand photo has less noise. It's noticeably better and it is also taken at ISO 3200. What's different about the right hand photo is I changed the exposure to 1/20th sec. f/4. It doesn't

So, can a safe take away from this discussion be as follows?

Don't underexpose, since that is the cause of the noise. It doesn't matter what ISO is set, just ensure you set the correct exposure?

Expose the sensor as much as possible. That's the bottom line. Everything about IQ improves with more exposure up to the full capacity of the sensor. There are different causes of noise in a digital camera but the dominant noise source is what we call shot noise and that's a function of exposure -- the common noise we see is caused by reduced exposure.

Joe

This is all new to me, i do not own a digital camera, so i have no idea.
But are you saying...all else equal... there are pictures that have more noise at Iso-200 than Iso-3200.?

All else is not equal and it would not be possible to take those two photos while maintaining all else equal. It's very unlikely as it would require an atypical use of the camera. It's possible however because the cause of noise isn't ISO directly. The noise we see in digital photos is primarily a function of exposure.

However ISO is indirectly linked to exposure so that when we raise ISO it's because we typically want to (need to) reduce exposure. So ISO is certainly linked to noise because of how we use it.

Let's look at a less extreme example that is typical of what my students do. Here's three photos (I had ginger snaps with my coffee this morning).

View attachment 171206 View attachment 171206 View attachment 171208

I adjusted them to all appear the same exposure but out of the camera the first was too dark the middle was right and the last was too light. The first is taken at ISO 1600 and in the manner of one of my students it is underexposed. ISO 3200 was indicated by the meter but my students don't want noise so they often resist raising the ISO. They believe the noise comes from the ISO increase. The first two photos are both exposed 1/50th sec. f/4.

Now let's look at the noise in the three photos. This is a 100% enlargement of a section. I have noise filtering in the software turned off for all.

View attachment 171209

The left photo is taken at ISO 1600 1/50th sec. f/4. The middle photo is taken4 at ISO 3200 1/50th sec. f/4. The noise is basically the same in both with a very slight improvement showing in the ISO 3200 photo. Typically what ISO does in the camera suppresses noise. My students are making a mistake and they should raise the ISO to the level needed -- the noise is from the exposure. Now the third, the right hand photo has less noise. It's noticeably better and it is also taken at ISO 3200. What's different about the right hand photo is I changed the exposure to 1/20th sec. f/4. It doesn't seem like much but that's more than doubling the amount of light reaching the sensor and it shows. That exposure change however of course highlights the entire dilemma to begin with: and I used a tripod as if I could convince my students to get and use a tripod. 1/20th sec. is not hand holdable -- ISO is often about motion.

So all things not being equal no one is going to reduce exposure for an ISO 200 photo by more than 4 stops to show that it's then noisier than an ISO 3200 photo properly exposed.

Joe

So, can a safe take away from this discussion be as follows?

Don't underexpose, since that is the cause of the noise. It doesn't matter what ISO is set, just ensure you set the correct exposure?

Expose the sensor as much as possible. That's the bottom line. Everything about IQ improves with more exposure up to the full capacity of the sensor. There are different causes of noise in a digital camera but the dominant noise source is what we call shot noise and that's a function of exposure -- the common noise we see is caused by reduced exposure.

You have to deal with the pragmatic requirements of taking the photo and you should take the photo even if you have to reduce exposure to do so. ISO plays a role in that raising ISO biases the light meter to calculate a reduced exposure and so the correlation that folks see is real but causation is spurious. So if you raise ISO and take the exposure the meter indicates you'll get more noise because you reduced exposure. If you need a faster shutter speed that's going to reduce exposure and that will cause noise.

Whatever the given circumstance, expose the sensor as much as possible.

Joe