Underexposure DOES NOT PRODUCE NOISE

[digitaldog]

The quote:

"Underexposure DOES NOT PRODUCE NOISE. That is NOT how the physics works. On modern cameras which all these are, the signal is ISO invariant because the amplifiers have gotten so good that noise is caused only by the number of photons hitting the sensor pixel. ISO just scales the number. Doesn't matter whether the entire image is over or under exposed. Doesn't matter what camera.

The test:
http://digitaldog.net/files/ISO200Normal_minus3stops.png

 

Details:

Canon R6 Mark II.

ISO 200 (yes, I can set it lower with a custom setting, no need).

Meter: Minolta Flash Meter III incident reading.

IncidentMeterNormal.dng:

Normal exposure is recommended exposure of the Minolta meter: 1/15th@F8

In-camera meter tells me that would be 1 stop over exposed it recommends 1/30th@F8.

-3stops.dng

125th@F8 or three stops below incident meter recommendations: UNDER EXPOSED.

All settings in Lightroom Classic’s Detail OFF.

Any comments? 

Edited May 4, 2023 by digitaldog
added JPEG

[paddler4]

This quote seems to conflate two things.

One is ISO invariance. If I understand right, ISO invariance has nothing to do with the impact of exposure per se; it is a comparison between amplification in camera (setting a higher ISO) and amplification in post (brightening the image). An ISO-invariant camera is one for which the two types of amplification produce essentially identical results. (A tangent: I don't know about the R6, but many Canon cameras aren't ISO invariant. Mine isn't.)

The second is the impact of exposure, independent of amplification. For this question, the quote--where did it come from?--is simply wrong. The effect of lowering exposure is to lower the amount of signal but not the amount of read or amplification noise. So, underexposing results in noise being a higher percentage of the total output. That seems to be what your comparison shows.

Think about limiting conditions. As one approaches a black frame--no signal at all--the image approaches nothing but noise. As one increases signal by increasing exposure, that noise necessarily becomes a smaller proportion of the whole. That's the physics.

I no longer have them, but some  years ago I created a series of images to illustrate this, using them to illustrate the fact that the way to minimize the appearance of noise in high-ISO shots is to expose to the right, maximizing signal relative to the constant (apart from random variation) amount of noise.

Edited May 4, 2023 by paddler4

[digitaldog]

The quote conflates several things, and I agree. 

My main beef is: "Underexposure DOES NOT PRODUCE NOISE".

My tests seem to dismiss this. FWIW, the formatting was his, not mine. 

ISO has nothing to do with exposure per se. My old copy and paste:

Setting ISO speed does not change the sensitivity of the sensor to light, like volume control does not change the sensitivity of a radio. In both cases the setting (ISO or volume) controls only the signal processing, while the input stage (sensor, antenna) provides the same input signal.

That's why when ISO setting is cranked up, automatic exposure results in more noise - automatic exposure in this case decreases the exposure (that is, the combination of aperture and shutter speed is set to allow less light captured by the sensor). Less exposure, less light, more noise.

As for ISO invariance (and my new Canon), I don't know (nor care) but the person who I'm quoting also stated:

"Doesn't matter what camera".

[digitaldog]

One more test.

1. ISO 800 but using above-recommended exposure 15th@F8.

2. ISO 50 but 125th@F8. 

Detail settings in Lightroom Classic off. 

Low ISO produces less noise: No. 

High ISO produces more noise: No. 

 

Edited May 4, 2023 by digitaldog

[frans_waterlander]

And who is this genius?

[digitaldog]

38 minutes ago, frans_waterlander said:

And who is this genius?

Sorry, I'm not about to unleash you on anyone. 

Edited May 4, 2023 by digitaldog

[paddler4]

On an unrelated topic: how do you like your R6 II? I'm tottering on the edge of ordering one. I would have preferred slightly higher resolution, say 30 MP, but apart from that, it looks like a truly superb camera.

[digitaldog]

I'm pretty pleased with the R6 MII. I have only had it for about 2 months. I was going to wait on the R8 but decided to go this route for the better viewfinder, batteries, etc. This is only my 2nd mirrorless. I decided to try this technology a few years ago with a Sony RX-10 IV and wanted better image quality and a battery that didn't die in an hour. For my uses, 24.2MP is more than enough. It is a bit heavier than the Sony, as I expected but I'm okay with that. The focusing system is amazing. The menu system is vastly superior to Sony. A huge step up, but the Sony was kind of a step down from the DSLR I had prior (5D MII).

One of my issues with Mirrorless was the viewfinder. I just wasn't used to looking at a TV screen so to speak after 50 years of DSLR. But this unit is much better, I can even crop it in a tad as I wear glasses. There are two modes to control the basic visual feedback, one that tries to make it more DSLR-like. 

[paddler4]

I print up to 17 x 22, so in rare cases, if I have to crop substantially, more pixels would be a minor improvement. However, I'm not sure it's worth the costs, both financial and otherwise. I got used to an EVF using a tiny Lumix LX-100, so I suspect that the vastly better EVF on the R6 II would be  just fine for me. I'd be switching from a 5D IV, so I would save a little weight, but not much, and I would lose about 6 MP.

[digitaldog]

The R6 MII is pricey! Which is why the R8 might fly. But I decided at the time I didn't want to wait for the R8 and I'd spend the money since I don't expect to do so for quite some time. 

[paddler4]

But at least cheaper than the R5...

[frans_waterlander]

On 5/4/2023 at 3:57 PM, digitaldog said:

Sorry, I'm not about to unleash you on anyone. 

The Adobe Support Community thread this basterdized quote was taken from can be found here: https://community.adobe.com/t5/camera-raw-bugs/p-denoise-ai-and-pentax-k3-mkiii-dng-files-not-working/idi-p/13769456#comments

Seems to me you did the right thing and showed that ISO invariance is a myth. Good for you and good for all of us.

[digitaldog]

8 hours ago, frans_waterlander said:

Seems to me you did the right thing and showed that ISO invariance is a myth. Good for you and good for all of us.

🤫

https://forum.luminous-landscape.com/index.php?topic=136134.msg1185926#msg1185926

Edited May 7, 2023 by digitaldog

[paddler4]

9 hours ago, frans_waterlander said:

Seems to me you did the right thing and showed that ISO invariance is a myth. Good for you and good for all of us.

Dog did not show that ISO invariance is a myth. The post he debunked confused ISO invariance with the assertion he did debunk, which is that with an ISO-invariant camera, underexposure doesn't increase noise. It does, as it does with all digital cameras.

ISO invariance means only that the impact of amplification in camera (boosting ISO) and of amplification in post are similar, so there is no reason to bump ISO in camera. Most ISO-non-invariant cameras produce less noise boosting ISO in camera (up to several stops) than boosting an underexposed image in post. This doesn't alter that reducing exposure reduces the signal:noise ratio and therefore increases visible noise. This latter is what Dog demonstrated concretely.

[digitaldog]

Indeed: I didn't show it, it's not a myth and again, Frans is misunderstanding then posting. That is no myth!

 

[digitaldog]

On 5/7/2023 at 8:42 AM, paddler4 said:

ISO invariance means only that the impact of amplification in camera (boosting ISO) and of amplification in post are similar, so there is no reason to bump ISO in camera. 

Here's what I'm refining text wise if you have any comments:

 

The term "ISO invariance" refers to a camera's ability to maintain consistent image quality while capturing images at different ISO settings. Suppose a camera is (perfectly) ISO invariant at all ISO settings. In that case, it will produce the same level of image quality whether the ISO setting value that was set at capture or image brightness is adjusted in a raw converter. This ISO setting “doesn’t matter” in that scenario.

A non-ISO invariant camera may introduce more noise or less noise depending on the exposure due to this ISO setting; there is a variant to the outcome of setting the ISO.

In both kinds of systems, ISO will affect the recommendation of exposure (Aperture + Shutter speed), which affects the degree of noise produced. In a non-ISO invariant camera, a higher ISO at optimal exposure can produce less noise than a lower ISO setting using the same exposure! This should not be the case with an ISO-invariant camera system. Hence the name and behavior. This behavior destroys the myth that “higher ISO produces more noise.” Underexposure is the cause of (increased) noise, not ISO.

Altering ISO and, thus, the exposure recommendation can come in handy in low-light situations where you may need to increase the ISO setting and this exposure recommendation to avoid camera shake or stop motion with a higher shutter speed. Or if you wish to alter DOF, that wouldn’t be recommended* at a lower ISO setting. This results in less exposure and more noise. Again, it is worth stressing: underexposure is the cause of (increased) noise, not ISO.

* Lastly, the ISO provides a ‘recommended’ exposure (Aperture + Shutter speed), but the savvy photographer does not need to accept this! Even without the inclusion of ISO, a savvy photographer will often ignore a reflective meter recommendation, for example, a black cat on coal or a white dog on snow. Such a recommendation would produce an overexposed or underexposed capture, despite the ISO setting in those examples.

[Alfred_Pennyworth]

This statement is mostly accurate. In modern digital cameras, underexposure alone does not produce noise. Rather, noise is primarily caused by the number of photons (light) hitting the sensor. This means that if a photo is underexposed, it may appear darker, but it will not necessarily be noisier than a properly exposed photo taken with the same ISO and shutter speed settings.

However, you may require some extra tools to fix the noise such as Denoise or photo repair tools.

[digitaldog]

8 hours ago, Alfred_Pennyworth said:

 In modern digital cameras, underexposure alone does not produce noise. Rather, noise is primarily caused by the number of photons (light) hitting the sensor. This means that if a photo is underexposed, it may appear darker, but it will not necessarily be noisier than a properly exposed photo taken with the same ISO and shutter speed settings.

The example above is from a modern camera released 6 months ago.

Underexposed has less photons yes.

How then is it not necessarily noisier (under-exposure does not produce noise)?

Darker not the case due of increase of brightness in raw converter orthe default rendering for JPEG (not optimal exposure for the raw).

Agreed that ALL noise is from underexpose capture indeed (read noise).

[paddler4]

To resolve Alfred's question: for a given level of final image brightness, lower exposure means more noise.

The base of all of this is that most noise sources are unaffected by exposure. For all practical purposes, you can treat their sum as a constant. Signal is variable: it depends on exposure. So it's unavoidable: if you drop the amount of signal, you decrease the signal to noise ratio. If you leave an underexposed image dark, you may not notice this. But as you brighten the image, you amplify both signal and noise, and as a result, the underexposed image has more apparent noise. There is no avoiding this, and ISO invariance has no relevance except as guidance for how to minimize the issue. On my old Canon bodies, I was better off raising ISO for the first few stops than brightening in post. That's not true of an ISO invariant camera.

One corollary of this, which Dog has stressed, is that to minimize noise under any circumstances, the key is attending to proper exposure. All I would add is that as underexposure gets more severe, this attention to proper exposure becomes even more important because of the simple math of the signal:noise ratio. I had a test series years ago in which I shot with the original 7D, a camera that gets noisy fast. I took repeated shots of a scene with only modest DR, so there was room for variation in exposure without clipping. I showed that even with that camera, one could underexpose by a number of stops without serious noise problems as long as one exposes to the right, that is, as long as one maximizes signal strength.

 

[digitaldog]

Getting back to the original topic (since the ISO invariance myth has been dismissed). 13 conversations with one fellow, I've distilled the following statements, and I'm trying to wrap my head around the various comments as:

1. We're just speaking in different terms (English may not be his native language, I'm not joking or being rude, that's likely true). 

2. He's all over the map in the statements. 

3. I'm missing his points.

So here's copy and paste in one place, comments welcome of course:

Underexposure does not cause noise. It only increases its magnitude in relation to the signal.

Sure underexposure increases noise in digital too but it is far less of a problem.

You let less light into the camera by exposing three stops shorter. So you should get a worse signal to noise (SNR) ratio that is approximately 2.8x worse exactly like you observe. Underexposing only increases it (noise).

You can also correctly say that an 'underexposed' image will have worse noise than a 'correctly' exposed image and an 'overexposed' image (as long as it doesn't blow out too much) will have less noise.

I would say that exposure choices are the primary controlling factor but not the cause of noise. This is a very important distinction as it goes to the heart of how image capture works. Exposing less increases the relative noise. Exposing more decreases it.

It's again simply due to the shorter exposure. You expose at 1/125 vs 1/15 seconds. Exposing less actually decreases the absolute amount of noise. It also doesn't cause the noise. The noise is caused by the photon particle nature of the light. It is not caused by something in the exposure. Exposure however influences it.

At best I would agree with saying that underexposure causes an increase in the noise to signal ratio (or alternatively said, a decrease of the signal to noise ratio).

 

[NAwlins Contrarian]

Quote

The quote:

"Underexposure DOES NOT PRODUCE NOISE. That is NOT how the physics works.... ISO just scales the number. Doesn't matter whether the entire image is over or under exposed. Doesn't matter what camera....

Normal exposure is recommended exposure of the Minolta meter: 1/15th@F8 ...

125th@F8 or three stops below incident meter recommendations: UNDER EXPOSED.

[Coming in late, but interested]

Underexposure doesn't produce noise, or at least, not precisely-speaking; but it does reduce signal, which lowers signal-to-noise ratio, which appears (above some threshold of visibility, and especially insofar as postprocessing equalizes overall image brightness) as relatively more noise. Agree? Disagree?

I think the similar-but-different test that might show very similar noise would be both at f/8 and 1/125 s but one at ISO 200 and the other at ISO 1600, and then equalize brightness in postprocessing. And I stress "might" because dual-gain sensors and non-linear gain ("ISO") settings can substantially change that behavior / result. Agree? Disagree?

Edited May 10, 2023 by NAwlins Contrarian

[digitaldog]

28 minutes ago, NAwlins Contrarian said:

reduce signal, which lowers signal-to-noise ratio

Caused by what action by the photographer? 

[NAwlins Contrarian]

Quote

Caused by what action by the photographer?

I'm not sure I understand the question, but I'll take a crack at it. If the photographer reduces exposure, then that causes reduced signal. The photons coming through the lens and striking the sensor are the signal. The photographer can reduce the signal by shortening the exposure time (i.e. increasing the shutter speed) and/or reducing the size of the aperture (i.e. stopping down).

 

 

[digitaldog]

9 hours ago, NAwlins Contrarian said:

I'm not sure I understand the question, but I'll take a crack at it. If the photographer reduces exposure, then that causes reduced signal. The photons coming through the lens and striking the sensor are the signal. 

No argument.

The photographer reduces (under exposes). The results are more or less noise? That's kind of the OP.

[digitaldog]

This could all boil down to semantics and *some* trying to be precise (which I can't fault).  Three different sentences from three different posters:

1. "Underexposure DOES NOT PRODUCE NOISE. That is NOT how the physics works"
2. "The effect of lowering exposure is to lower the amount of signal but not the amount of read or amplification noise. So, underexposing results in noise being a higher percentage of the total output".
3.  "Less exposure, less light, more noise".

Photographers state far too often: "I shot at a high ISO and the images are noisy".

Most here seem to agree that ISO isn't the cause of the noise. 

Most here seem to agree that less exposure results in more (visible) noise. 

What is a simple sentence or maybe paragraph to provide to those photographers that seem to think, higher ISO produces the noise? Clearly, this is an exposure issue. If underexposure doesn't produce noise, what is the message that can be stated that is both easy for them to understand and makes the scientist happy? Is that even possible?