ISO and signal to noise ratios

[Michael Fryd]

Oh come on! Just what is your agenda?

Helping beginners understand that noise is from the low exposure, not the high ISO.

When beginners think the noise is from high ISO, they are tempted to improve the image by turning down the ISO and underexposing. If the noise was actually from the high ISO, this would improve noise.

Turning down the sensitivity, or ISO as you call it, does invariably improve the image. And shows just what total nonsense you are spouting in order to make yourself look smart. Sensitivity (ISO) is part of the exposure triangle and if you adjust the other two parameters to suit then the exposure does not change. Reducing sensitivity and failing to adjust the other two parameters would indeed result in underexposure and reduced image quality. But what idiot is going to do that?

There's no need to get personal here. We are discussing how cameras work, not the people who use them.

I just did a quick test. I took four pictures with the same exposure, but four different ISO settings (ISO 200, 400, 800, and 1600). I used the same shutter speed, same aperture, and same window light for all four. I processed the raw file with various amounts of brightness adjustments in order to give all four JPEGs a similar brightness.

If the ISO setting is a major factor in the noise, the ISO 1600 image should have significantly more noise than the ISO 200 image. If the exposure is the major factor in image noise, then we would expect to see similar noise levels in all four images.

Here are the images:

View attachment ed875f9dbb3d4f25a36eb3ee8067d8a9.jpg

View attachment 4482d3d283f24c29a661dff863e61e1f.jpg

View attachment 032e848e2b7341a3bc88096eff82e14d.jpg

View attachment 2aec01a647ee4420baf0515f8fd17f79.jpg

By the way, referring to the sensitivity setting as "the ISO" is like referring to the aperture setting as "the F" or referring to the shutter speed as "the seconds".

But the ISO is not a "sensitivity" setting. If there are not enough photons for the sensor at ISO 100, there aren't enough for ISO 64000. The sensitivity of the sensor is the same for both. What the ISO does is change how the processing back end interprets the photon count.

This is why changing the ISO doesn't directly change the noise.

In terms of beginners, the simple (and correct) explanation is that less light on the sensor generally results in more noise in the images. ISO affects metering and processing. It tells the metering system how much light we in our exposure, and informs our backend processing, how much light to expect. High ISO generally results in noisier images because we generally have lower exposures at higher ISO.

Altering the ISO independently of exposure, will not have a major impact on noise.

 

[Michael Fryd]

Oh come on! Just what is your agenda?

Helping beginners understand that noise is from the low exposure, not the high ISO.

When beginners think the noise is from high ISO, they are tempted to improve the image by turning down the ISO and underexposing. If the noise was actually from the high ISO, this would improve noise.

Either one, under exposing or high ISO will increase noise. Do both together, well don't, unless you have no choice.

Actually, if you are shooting raw, the noise will generally be dependent on the exposure. Lower the ISO without changing the exposure, and you don't change the noise.

Lower ISO only provides significant noise reduction when accompanied by the corresponding increase in exposure.

It's that increase in exposure that reduces the noise, not the ISO setting.

 

[FingerPainter]

While I respect your answers, any amount of under exposure from "optimum" is going to result in a lower S/N ratio.

Yes, but if f/2 1/800 ISO 400 is optimum exposure, then f/2 1/800 ISO 200 is no amount of underexposure. The amount of exposure is the same. The amount of light captured is the same, so the SNR WRT shot noise is the same. A change in ISO may or may not result in a change in SNR WRT camera-added noise, depending on how the ISO change is implemented, and/or depending on how and where the camera-added noise is generated.

ISO isn't a parameter of exposure; only aperture, shutter and scene luminance are. Shot noise depends only on the amount of captured light, which for a given camera is a constant function of the exposure.

ISO affects the brightness of the image but not directly its exposure. When an ISO Increase is implemented through analog gain, then the SNR WRT camera-added noise can be increased, because all of the signal but only that noise added before the gain stage are scaled up by the same factor. Note that this effect, which is consistent with what you have been saying, is the opposite of what most explanations of the so-called "exposure triangle" have to say about the effect of an increase in ISO on noise.

The more it's under exposed, the lower that S/N ratio is going to be. Even on a Sony.

Yes, but a f/2 1/800 ISO 200 shot is no more underexposed than a f/2 1/800 ISO 400 shot, if they were taken in the same light.

ISO is still part of the exposure triangle,

Not really. The name "exposure triangle" is a misnomer. The triangle involving aperture, shutter and ISO models the effect of three camera controls on image brightness not on exposure. Its originator didn't call it the exposure triangle, but rather the "photographic triangle" perhaps because he knew it didn't model exposure. So ISO is part of the photographic triangle, and affects image brightness, but not exposure.

From the WIkipeda article on Exposure (photography):

"In photography, exposure is the amount of light per unit area (the image plane illuminance times the exposure time) reaching a photographic film or electronic image sensor, as determined by shutter speed, lens aperture and scene luminance."

and aperture, shutter speed or ISO can be used to change brightness.

Yes, but if you change ISO while leaving aperture, shutter and scene luminance unchanged you change brightness but not exposure. When used properly, "exposure" is not a synonym for brightness, despite its frequent colloquial usage as such a synonym.

 

[bclaff]

With regard to shot noise, you've essentially got it right. The effect of ISO setting on camera-added noise on Sony cameras is slight, so you can afford to neglect it.

Fine point, some cameras, like the A7RM2 have dual conversion gain; so it does matter when you reach ISO 640 (in that case).

Regards,

 

[gollywop]

As noted by several, there are effectively two sources of noise: shot noise, which comes part and parcel of the light, and read noise, which comes from the camera's electronics. Read noise can be safely ignored in the brighter parts of the image since it is swamped by the magnitude of the shot noise, but it can be of consequence in the deeply shadowed parts of the image.

You can do nothing about shot noise: the greater the light, the greater the shot noise. But, since the shot noise increases as the square root of the light, it increases more slowly than the light. That is, the greater the light, the less is the shot noise relative to the signal; the greater the signal to (shot) noise, and, as a consequence, the better the image quality.

Exposure is the amount of light impinging on the sensor per unit area. It is determined by the level of scene luminance, the f-ratio (relative aperture), and the shutter speed. ISO is not an element of exposure; that is, for given scene, f-ratio, and SS, altering ISO does not alter exposure.

For a given exposure, increasing ISO increases the brightness of the resulting OOC JPEG. The effect of the so-called exposure triangle results in images of equal brightness, but not in images of equal image quality. Thus, the same shot taken at f/4, 1/50 secs., ISO 100 and f/4, 1/100 secs, ISO 200 have the same OOC JPEG brightness, but the first shot has twice the exposure and, therefore, 41% better signal-to-(shot)noise.

[It should be noted that this notion of equal comparative brightness applies only to the resulting OOC JPEGs as viewed on the camera's monitor or with otherwise comparable processing, since, clearly, the brightness of any image can be altered at will in processing or by altering the brightness setting of the printer or viewing monitor. No image has an "inherent" brightness. The term brightness applies only to the perceived lightness of the final image as viewed.]

The camera's ISO control merely transforms the exposure information already on the sensor into the final raw data values. Increasing ISO increases both signal and (shot) noise in the same proportion, leaving signal-to-(shot)noise unchanged. The camera's ISO control performs its function after the exposure has taken place (and therefore, as already noted, cannot be part of that exposure). Further, since ISO occurs after exposure, it cannot alter the "sensitivity" of the sensor to light, although it does alter the sensitivity to light of the camera system taken as whole.

There is, however, a slight complication: for ISO-variant cameras, increasing the camera's ISO control actually decreases the read noise. Almost all cameras are ISO-variant up to some point, although now-a-days that point can occur at a fairly low ISO for many cameras (and some of today's cameras have two such points).

Thus, within the camera's ISO-variant region, increasing ISO can actually increase signal-to-(total)noise, since it leaves signal-to-(shot)noise unchanged, but reduces read noise (and therefore total noise). This effect has no real practical consequence except when signal is very low (when read noise is competitive with shot noise); i.e., only in deeply shadowed regions of the image.

--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg

 

[dwalby]

Thus, within the camera's ISO-variant region, increasing ISO can actually increase signal-to-(total)noise, since it leaves signal-to-(shot)noise unchanged, but reduces read noise (and therefore total noise).

can you explain how increasing ISO can reduce the read noise.

My understanding of this topic is any pre-gain-stage read noise is amplified along with the signal (and shot noise), so no read noise reduction there. Then post-gain-stage read noise is added to the amplified signal. Since the amplified signal is about the same amplitude as a low-ISO full well signal without amplification, the post-gain-stage read noise relative to the signal amplitude is about the same as the low-ISO case.

So I'm not seeing how increasing the ISO can reduce the read noise, is there something I'm not understanding correctly?

 

[gollywop]

Thus, within the camera's ISO-variant region, increasing ISO can actually increase signal-to-(total)noise, since it leaves signal-to-(shot)noise unchanged, but reduces read noise (and therefore total noise).

can you explain how increasing ISO can reduce the read noise.

My understanding of this topic is any pre-gain-stage read noise is amplified along with the signal (and shot noise), so no read noise reduction there. Then post-gain-stage read noise is added to the amplified signal. Since the amplified signal is about the same amplitude as a low-ISO full well signal without amplification, the post-gain-stage read noise relative to the signal amplitude is about the same as the low-ISO case.

So I'm not seeing how increasing the ISO can reduce the read noise, is there something I'm not understanding correctly?

You're on the right track, but you're not considering the noise as input referred, which is necessary to get a proper (clear) comparison to the shot noise. See Martinec's classic article

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#ETTR

The reduction in input-referred read noise with increased ISO is clearly less significant with ISO-invariant cameras which have very low downstream noise (R1 in Martinec's notation). For such cameras, the read noise (input referred) is relatively invariant to the ISO setting.

You might want to scroll up and read the preceding section also.

--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg

 

[bclaff]

Thus, within the camera's ISO-variant region, increasing ISO can actually increase signal-to-(total)noise, since it leaves signal-to-(shot)noise unchanged, but reduces read noise (and therefore total noise).

can you explain how increasing ISO can reduce the read noise.

My understanding of this topic is any pre-gain-stage read noise is amplified along with the signal (and shot noise), so no read noise reduction there. Then post-gain-stage read noise is added to the amplified signal. Since the amplified signal is about the same amplitude as a low-ISO full well signal without amplification, the post-gain-stage read noise relative to the signal amplitude is about the same as the low-ISO case.

So I'm not seeing how increasing the ISO can reduce the read noise, is there something I'm not understanding correctly?

You're on the right track, but you're not considering the noise as input referred, which is necessary to get a proper (clear) comparison to the shot noise. See Martinec's classic article

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#ETTR

A better link is www.photonstophotos.net/Emil%20Martinec/noise-p3.html#ETTR

Since there are many broken links in the original article.

The reduction in input-referred read noise with increased ISO is clearly less significant with ISO-invariant cameras which have very low downstream noise (R1 in Martinec's notation). For such cameras, the read noise (input referred) is relatively invariant to the ISO setting.

You might want to scroll up and read the preceding section also.

--
Bill ( Your trusted source for independent sensor data at http://www.photonstophotos.net )

 

[gollywop]

Thus, within the camera's ISO-variant region, increasing ISO can actually increase signal-to-(total)noise, since it leaves signal-to-(shot)noise unchanged, but reduces read noise (and therefore total noise).

can you explain how increasing ISO can reduce the read noise.

My understanding of this topic is any pre-gain-stage read noise is amplified along with the signal (and shot noise), so no read noise reduction there. Then post-gain-stage read noise is added to the amplified signal. Since the amplified signal is about the same amplitude as a low-ISO full well signal without amplification, the post-gain-stage read noise relative to the signal amplitude is about the same as the low-ISO case.

So I'm not seeing how increasing the ISO can reduce the read noise, is there something I'm not understanding correctly?

You're on the right track, but you're not considering the noise as input referred, which is necessary to get a proper (clear) comparison to the shot noise. See Martinec's classic article

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#ETTR

A better link is www.photonstophotos.net/Emil%20Martinec/noise-p3.html#ETTR

Since there are many broken links in the original article.

Yes, thanks Bill.

The reduction in input-referred read noise with increased ISO is clearly less significant with ISO-invariant cameras which have very low downstream noise (R1 in Martinec's notation). For such cameras, the read noise (input referred) is relatively invariant to the ISO setting.

You might want to scroll up and read the preceding section also.

 

[Auf Reisen]

This is precisely the kind of answer I was looking for. Thank you!

 

[Auf Reisen]

Oh come on! Just what is your agenda?

Helping beginners understand that noise is from the low exposure, not the high ISO.

When beginners think the noise is from high ISO, they are tempted to improve the image by turning down the ISO and underexposing. If the noise was actually from the high ISO, this would improve noise.

This is exactly what I did until quite recently. So what you're saying is more or less exactly what I needed to hear.

 

[Auf Reisen]

That is one amazingly thorough answer. Thank you!

The only time noise ever really bothers me is in deep shadows (say, the foreground in astrophotography), or in a blue sky. Astrophotographers usually recommend shooting at very high ISO levels and then to reduce the exposure in post, often with the explicit goal to increase the signal/noise ratio. Do you by any chance know what logic is involved in this reasoning?

 

[gollywop]

That is one amazingly thorough answer. Thank you!

The only time noise ever really bothers me is in deep shadows (say, the foreground in astrophotography), or in a blue sky. Astrophotographers usually recommend shooting at very high ISO levels and then to reduce the exposure in post, often with the explicit goal to increase the signal/noise ratio. Do you by any chance know what logic is involved in this reasoning?

Actually, it is not recommended to use "very high ISO levels." Rather, it is recommended to use the maximal exposure shooting conditions allow (at base ISO) and then, if ETTR is not possible (which is likely the case) and you are able to add ISO without clipping, apply ISO as high as needed up to, but not exceeding, the point where the camera becomes ISO-invariant. This ISO varies greatly for different cameras.*

Because light levels are so low in astrophotography, and the objects are moving (unless using timed trackers), maximal exposures (largest acceptable aperture and longest acceptable SS) are typically below ETTR. Under these conditions, it is not possible to achieve the full DR of the camera, so the fact that adding ISO reduces DR is not an issue. But the fact that read noise is reduced as ISO is added up to the point of ISO-invariance does increase the s/n and, hence, image quality.

Thus, the logic behind this is embodied in my earlier post.

On this issue, you will likely find this article of value:

http://dslr-astrophotography.com/iso-dslr-astrophotography/

* Once the acceptable aperture and SS have been determined for maximal exposure at base ISO, be careful that these settings are not altered when adding ISO. Under many circumstances when using programmed modes, changing ISO can also affect exposure settings. One is best off using manual mode so that the exposure settings (f-ratio and SS) are unaffected by the ISO setting.

--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg

 

[dwalby]

Thus, within the camera's ISO-variant region, increasing ISO can actually increase signal-to-(total)noise, since it leaves signal-to-(shot)noise unchanged, but reduces read noise (and therefore total noise).

can you explain how increasing ISO can reduce the read noise.

My understanding of this topic is any pre-gain-stage read noise is amplified along with the signal (and shot noise), so no read noise reduction there. Then post-gain-stage read noise is added to the amplified signal. Since the amplified signal is about the same amplitude as a low-ISO full well signal without amplification, the post-gain-stage read noise relative to the signal amplitude is about the same as the low-ISO case.

So I'm not seeing how increasing the ISO can reduce the read noise, is there something I'm not understanding correctly?

You're on the right track, but you're not considering the noise as input referred, which is necessary to get a proper (clear) comparison to the shot noise. See Martinec's classic article

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#ETTR

The reduction in input-referred read noise with increased ISO is clearly less significant with ISO-invariant cameras which have very low downstream noise (R1 in Martinec's notation). For such cameras, the read noise (input referred) is relatively invariant to the ISO setting.

OK, I remember seeing that kind of data a few years ago, and I understand the math behind it, but I think its a matter of semantics and how things are being compared.

If your signal is just barely below clipping at ISO 400 and you shoot at ISO100 instead, then I agree that the input referenced read noise is lower at ISO 400 than at 100. But that's just a mathematical convenience, the actual read noise didn't actually get lower. And why would I purposely underexpose by using ISO 100 at all?

To quote the article:

"Read noise at high ISO is much smaller than read noise at low ISO, in terms of the error in photon counting that it represents. Thus, better image quality is obtained for using the highest ISO for which the signal is not clipped."

I agree with that statement 100%.

 

[Michael Fryd]

That is one amazingly thorough answer. Thank you!

The only time noise ever really bothers me is in deep shadows (say, the foreground in astrophotography), or in a blue sky. Astrophotographers usually recommend shooting at very high ISO levels and then to reduce the exposure in post, often with the explicit goal to increase the signal/noise ratio. Do you by any chance know what logic is involved in this reasoning?

Actually, it is not recommended to use "very high ISO levels." Rather, it is recommended to use the maximal exposure shooting conditions allow (at base ISO) and then, if ETTR is not possible (which is likely the case) and you are able to add ISO without clipping, apply ISO as high as needed up to, but not exceeding, the point where the camera becomes ISO-invariant. This ISO varies greatly for different cameras.*

Because light levels are so low in astrophotography, and the objects are moving (unless using timed trackers), maximal exposures (largest acceptable aperture and longest acceptable SS) are typically below ETTR. Under these conditions, it is not possible to achieve the full DR of the camera, so the fact that adding ISO reduces DR is not an issue. But the fact that read noise is reduced as ISO is added up to the point of ISO-invariance does increase the s/n and, hence, image quality.

Thus, the logic behind this is embodied in my earlier post.

On this issue, you will likely find this article of value:

http://dslr-astrophotography.com/iso-dslr-astrophotography/

* Once the acceptable aperture and SS have been determined for maximal exposure at base ISO, be careful that these settings are not altered when adding ISO. Under many circumstances when using programmed modes, changing ISO can also affect exposure settings. One is best off using manual mode so that the exposure settings (f-ratio and SS) are unaffected by the ISO setting.

--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg

I'm trying to come up with a short summary describing ISO on a digital camera. I'm not looking to get technical, or provide every detail. I just want to convey a short overview.

Would you say the following is reasonable, and how would you suggest improving it?



On most modern cameras ISO serves three purposes:

1. It sets the target exposure for the metering system - with higher ISO settings, the metering system will suggest lower exposures (less light)
2. It may alter some of the operating parameters of the camera in order to optimize the camera's performance for the expected exposure
3. It suggests a default mapping to use for converting the sensor data to brightness in the resulting RGB image

I might add:

Generally the biggest factor in visible image noise is the amount of light on the sensor. As you increase the light on the sensor, the visible noise tends to go down.

High ISO setting are generally associated with high image noise because high ISO is generally associated with low exposures (less light on the sensor).

While many think of the high ISO as causing the noise, it is actually the low exposure. In fact, with many cameras, matching the ISO to the exposure will result in better quality than using a lower ISO and brightening the image n post processing.

I talk about "visible image noise" because thats the noise we see with our eyes. This avoids a discussion of absolute noise levels and Signal to Noise ratio.

I've avoided the word "sensitivity" because I don't want to suggest that at high ISO, the sensor is "better" at detecting a small number of photons.

I have mentioned that an ISO setting that is typical for the expected exposure is frequently the way to maximize quality for that exposure. Using lower ISO and changing brightness doesn't help (if it did, the camera would likely do that on its own).

What do you think? Is the above a reasonable overview? Have I left anything important out? Have I included anything that should be left out? Is there a better wording?

 

[bclaff]

Michael Fryd wrote:

...

I talk about "visible image noise" because that;s the noise we see with our eyes. This avoids a discussion of absolute noise levels and Signal to Noise ratio.

...

You can avoid discussing Signal to Noise Ratio (SNR) but that is what determines whether noise is "visible".

 

[Michael Fryd]

Michael Fryd wrote:
...

I talk about "visible image noise" because that;s the noise we see with our eyes. This avoids a discussion of absolute noise levels and Signal to Noise ratio.

...

You can avoid discussing Signal to Noise Ratio (SNR) but that is what determines whether noise is "visible".

 

[gollywop]

Thus, within the camera's ISO-variant region, increasing ISO can actually increase signal-to-(total)noise, since it leaves signal-to-(shot)noise unchanged, but reduces read noise (and therefore total noise).

can you explain how increasing ISO can reduce the read noise.

My understanding of this topic is any pre-gain-stage read noise is amplified along with the signal (and shot noise), so no read noise reduction there. Then post-gain-stage read noise is added to the amplified signal. Since the amplified signal is about the same amplitude as a low-ISO full well signal without amplification, the post-gain-stage read noise relative to the signal amplitude is about the same as the low-ISO case.

So I'm not seeing how increasing the ISO can reduce the read noise, is there something I'm not understanding correctly?

You're on the right track, but you're not considering the noise as input referred, which is necessary to get a proper (clear) comparison to the shot noise. See Martinec's classic article

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#ETTR

The reduction in input-referred read noise with increased ISO is clearly less significant with ISO-invariant cameras which have very low downstream noise (R1 in Martinec's notation). For such cameras, the read noise (input referred) is relatively invariant to the ISO setting.

OK, I remember seeing that kind of data a few years ago, and I understand the math behind it, but I think its a matter of semantics and how things are being compared.

If your signal is just barely below clipping at ISO 400 and you shoot at ISO100 instead, then I agree that the input referenced read noise is lower at ISO 400 than at 100. But that's just a mathematical convenience, the actual read noise didn't actually get lower.

It's not just a "mathematical convenience," whatever that means. The proportion of read noise to total noise has indeed been reduced. That's what's being seen directly in bringing the comparison back to input-referred. You can see the same in output referred, but it's not a direct comparison.

And why would I purposely underexpose by using ISO 100 at all?

I'm not sure exactly what this statement means. You seem to be implying that reducing ISO from 400 to 100 results in underexposure. The exposure is the same in both cases. A more relevant question would be "why would I use ISO 400 if I could acceptably increase my exposure (f-ratio and SS) instead?"

The only reason you would (should) be using ISO 400 "barely below clipping" is because you were unable to increase exposure more closely to ETTR at base ISO. Assuming base ISO 100, that would mean your shooting conditions only allowed a maximal exposure that was 2 stops below ETTR. That can happen if DoF requirements limited your aperture and motion-blur/camera-shake requirements limited your shutter speed.

The general principle is: maximize exposure subject to your shooting requirements at base ISO, and then add ISO only if that maximal exposure is less than ETTR. Never substitute ISO for acceptable exposure (acceptable f-ratio and SS).

To quote the article:

"Read noise at high ISO is much smaller than read noise at low ISO, in terms of the error in photon counting that it represents. Thus, better image quality is obtained for using the highest ISO for which the signal is not clipped."

I agree with that statement 100%.

Well I don't. Taken as it stands, this statement is inappropriately broad and dogmatic. I cited this article because it shows very well how increasing ISO can reduce read noise (which was your initial issue), and it does an excellent job of debunking the original rational for ETTR given by Reichmann in Luminous Landscape. But it doesn't make clear that one doesn't want to substitute the use of ISO for exposure; it doesn't really get into a proper notion of ETTR. The relative reduction in shot noise that accompanies a one-stop increase in actual exposure far outweighs any decrease in read noise that accompanies a one-stop increase in ISO. Thus, as noted above, one first increases exposure as much as possible (up to ETTR at base ISO) subject to shooting conditions (DoF, motion-blur/camera-shake) and only then adds ISO.

Further, there is no need to add ISO beyond the ISO-variant limit of the camera, although there is no cost in doing so as long as one can truly avoid clipping. But, since it is often difficult to know at the time of the shot that one is actually avoiding clipping, it is safer to avoid unnecessarily increasing ISO beyond the ISO-variant limit and, instead, to do any further brightening during processing. I'm assuming here one is shooting raw.

Regarding exposure and ETTR, you (and others) might be interested in the following two articles:

https://www.dpreview.com/articles/8148042898/exposure-vs-brightening

https://www.dpreview.com/articles/6641165460/ettr-exposed

--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg

 

[gollywop]

That is one amazingly thorough answer. Thank you!

The only time noise ever really bothers me is in deep shadows (say, the foreground in astrophotography), or in a blue sky. Astrophotographers usually recommend shooting at very high ISO levels and then to reduce the exposure in post, often with the explicit goal to increase the signal/noise ratio. Do you by any chance know what logic is involved in this reasoning?

Actually, it is not recommended to use "very high ISO levels." Rather, it is recommended to use the maximal exposure shooting conditions allow (at base ISO) and then, if ETTR is not possible (which is likely the case) and you are able to add ISO without clipping, apply ISO as high as needed up to, but not exceeding, the point where the camera becomes ISO-invariant. This ISO varies greatly for different cameras.*

Because light levels are so low in astrophotography, and the objects are moving (unless using timed trackers), maximal exposures (largest acceptable aperture and longest acceptable SS) are typically below ETTR. Under these conditions, it is not possible to achieve the full DR of the camera, so the fact that adding ISO reduces DR is not an issue. But the fact that read noise is reduced as ISO is added up to the point of ISO-invariance does increase the s/n and, hence, image quality.

Thus, the logic behind this is embodied in my earlier post.

On this issue, you will likely find this article of value:

http://dslr-astrophotography.com/iso-dslr-astrophotography/

* Once the acceptable aperture and SS have been determined for maximal exposure at base ISO, be careful that these settings are not altered when adding ISO. Under many circumstances when using programmed modes, changing ISO can also affect exposure settings. One is best off using manual mode so that the exposure settings (f-ratio and SS) are unaffected by the ISO setting.

--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg

I'm trying to come up with a short summary describing ISO on a digital camera. I'm not looking to get technical, or provide every detail. I just want to convey a short overview.

Would you say the following is reasonable, and how would you suggest improving it?

Hello Michael. I really don't want to get into a back-and-forth here. This is not a topic that is going to be readily resolved and formed. I'll make some initial comments but will likely not go further.

On most modern cameras ISO serves three purposes:

1. It sets the target exposure for the metering system - with higher ISO settings, the metering system will suggest lower exposures (less light)
2. It may alter some of the operating parameters of the camera in order to optimize the camera's performance for the expected exposure
3. It suggests a default mapping to use for converting the sensor data to brightness in the resulting RGB image

Item 1 should mention that, in programmed modes, altering ISO alters exposure settings so as to achieve an OOC JPEG of the same brightness. It's effect is entirely different in M mode.

Item 3 is really a part of item 1. The "target" exposure is precisely that which, according to the manufacturer, will result in the average scene luminance (or some such criterion) to produce a middle gray (brightness) in the OOC JPEG when EC is zero. [This "target" can readily miss the mark for many (most) other needs.]

The essence of item 2 is important and not understood by many beginners. But you've already suggested this phenomenon in item 1. Further, the term "expected exposure" is not defined and likely cannot be in any meaningful way. Even further, it's not clear in what sense the alterations to the exposure settings "optimize" the exposure (expected or not). It could be, for example, that, depending on the programming mode, a decrease in ISO is compensated for by a longer SS that is inappropriate for camera shake or motion blur, or an increase in ISO is compensated for by an increase in f-ratio that is not appropriate for desired DoF.

I might add:

Generally the biggest factor in visible image noise is the amount of light on the sensor. As you increase the light on the sensor, the visible noise tends to go down.

High ISO setting are generally associated with high image noise because high ISO is generally associated with low exposures (less light on the sensor).

While many think of the high ISO as causing the noise, it is actually the low exposure. In fact, with many cameras, matching the ISO to the exposure will result in better quality than using a lower ISO and brightening the image n post processing.

The last two items are saying effectively the same thing and can be combined. You do not define the concept of "matching ISO to exposure." In the above, the phrase is empty and needs significant explanation and justification.

I talk about "visible image noise" because thats the noise we see with our eyes. This avoids a discussion of absolute noise levels and Signal to Noise ratio.

Maximizing signal-to-noise is what it's all about. You've left the baby out of the bath water.

I've avoided the word "sensitivity" because I don't want to suggest that at high ISO, the sensor is "better" at detecting a small number of photons.

I have mentioned that an ISO setting that is typical for the expected exposure is frequently the way to maximize quality for that exposure. Using lower ISO and changing brightness doesn't help (if it did, the camera would likely do that on its own).

What do you think? Is the above a reasonable overview? Have I left anything important out? Have I included anything that should be left out? Is there a better wording?

--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg

 

[Iliah Borg]

I really don't want to get into a back-and-forth here. This is not a topic that is going to be readily resolved and formed.

Maybe because phenomenological descriptions are too convoluted Reminds of the old Blind Men Appraising an Elephant fable.

It is much simpler IMHO to start with how ISO setting works. From there, anybody can deduce things like clipping,...