[bxf]

There would be about the same difference in noise as if you set ISO to one stop lower instead of overexposing 1 stop. But as the images get different processing there might be some differences.

My thinking was that if I overexpose, I am letting in more Total Light, which, as the arguments go, should result in lower noise. Pulling back a stop in PP would likely cause IQ deterioration, but aside from that, if the Total Light argument is valid, then there should be lower noise, no?

[Fogel70]

My thinking was that if I overexpose, I am letting in more Total Light, which, as the arguments go, should result in lower noise. Pulling back a stop in PP would likely cause IQ deterioration, but aside from that, if the Total Light argument is valid, then there should be lower noise, no?

Yes, just as using lower ISO would do, as the exposure time will be twice as long.
By overexposing the image one stop you will collect the same amount of light as using one stop lower ISO with zero exposure compensation.

[Dartmoor Dave]

My thinking was that if I overexpose, I am letting in more Total Light, which, as the arguments go, should result in lower noise. Pulling back a stop in PP would likely cause IQ deterioration, but aside from that, if the Total Light argument is valid, then there should be lower noise, no?

That's really just a variation on the well-understood technique of exposing to the right: exposing so that the brightest highlights are just below clipping. Exposing to the right does indeed give you a low-noise image, but it's nothing to do with "Total Light".

If "Total Light" was real, then an APS-C Sony NEX owner using a full-frame lens with a focal length reducer (such as Lens Turbo) would get the same signal-to-noise ratio as he would using the same lens at the same ISO on a full frame sensor.

He won't.

[Fogel70]

If "Total Light" was real, then an APS-C Sony NEX owner using a full-frame lens with a focal length reducer (such as Lens Turbo) would get the same signal-to-noise ratio as he would using the same lens at the same ISO on a full frame sensor.

He won't.

No, one stop lower ISO is needed on the APS-C camera as the sensor is half the size. You cant capture same amount of light using same ISO if sensor size is different (and using same aperture size).

A 50/1.4 lens on used with a focal reducer on APS-C body will be equivalent of a 33/0.9 APS-C lens.
So using the lens wide open the FF might give exposure of 1/50s ISO400, and the on the APS-C it will give 1/50s ISO200

---

Last edited by Fogel70; 10-16-2014 at 05:53 AM.

[bxf]

Yes, just as using lower ISO would do, as the exposure time will be twice as long.
By overexposing the image one stop you will collect the same amount of light as using one stop lower ISO with zero exposure compensation.

Yes, of course. Understood.

[Dartmoor Dave]

No, one stop lower ISO is needed on the APS-C camera as the sensor is half the size. You cant capture same amount of light using same ISO if sensor size is different (and using same aperture size). A 50/1.4 lens on used with a focal reducer on APS-C body will be equivalent of a 33/0.9 APS-C lens. So using the lens wide open the FF might give exposure of 1/50s ISO400, and the on the APS-C it will give 1/50s ISO200

Why are you changing the ISO to match exposures, rather than aperture or shutter speed?

To use your example: 50mm lens at f/1.4 at 1/60 at ISO400 on full frame equals 50mm lens at f/0.95 at 1/125 at ISO400 on APS-C (using a focal length reducer such as Lens Turbo). Surely that's the same "Total Light" hitting each sensor? So, if it's true that the same "Total Light" results in the same signal-to-noise ratio, then both sensors should now have the same SNR at the same ISO? But of course they don't.

Please understand that I'm not trolling here. I'm genuinely trying to make sense of this notion that there's something called "Total Light" that's different from plain old "exposure". But changing the ISO and then claiming that you've changed the signal-to-noise ratio because of "Total Light" seems like cheating.

[Fogel70]

If using the above exposure the APS-C sensor will capture half amount of light compare to the FF sensor.
The APS-C sensor use half the sensor size, half the exposure time and twice the illumination. The FF sensor will use twice the sensor size, twice the exposure time and half the illumination.

In your example you don't compensate for the difference in sensor size for same "total light". You will need twice the illumination at the same exposure time (or same illumiation with twich the exposure time) for an APS-C sensor to capture the same total amount of light as FF sensor. And to avoid over exposure on the APS-C you need to set one stop lower ISO.

---

Last edited by Fogel70; 10-16-2014 at 07:51 AM.

[Kunzite]

As I said in a previous post, "total light" is just exposure multiplied by area - it should be easy to understand if you keep this relation in mind. Of course, it does not depend on ISO because none of its components does ("exposure" here means the total amount of light per area that reaches the sensor).
This "total light" is influencing photon noise, and it's correct to say that "same total light equals same photon noise". Again, ISO has no influence on this.
But this is just a part of the story and would lead to an idealized, purely theoretical view in which all sensors are "the same".

[Dartmoor Dave]

Thanks to Fogel70 and Kunzite for trying to clarify the concept, but I still don't get it. I regularly shoot with different film and digital formats, mostly using incident metering, and I've never encountered anything out there in the real world that remotely resembles the effects that "Total Light" is supposed to cause. With different formats I just get the differences between angle of view and depth of field and noise/grain that I've been working with for the past thirty-something years. And if my meter indicates an exposure of f/16 at 1/125 at ISO100, then that's the correct exposure no matter what format I'm shooting with.


Anyway, I'll bow out of this now. I've clearly failed to grasp something fundamental about "Total Light", but since it seems to have absolutely zero useful application in real-world photography I've decided to stop worrying about it.

[jsherman999]

Again, if you're fixated by the light part of the signal-noise ratio, but not the equally important noise component, you're going to struggle to explain this:

My dear Clackers, I'm going to forgive you some of this because you barged into the discussion relatively late.

Here:

Overall image noise can be broken down into photon shot noise + read noise.

Shot noise is largely determined by the Total Light. Read noise is largely determined by sensor efficiency, and is seen more in lower light shooting in which image brightening (ie ISO bump) is used.

The Canon 5D vs. K5 (or K3) is a perfect example of an older, less efficient pixel technology (5D) going up against a newer one (Sony Exmor/K5). As DXO shows, even though the Canon has an older, less efficient sensor, it still slightly surpasses the K5 in noise control.

How can it do this if the sensor is worse?

Because the ground it loses in Read noise (caused by lower sensor efficiency) is made up for with less shot noise (caused by more total light.)

Get it?

Now, take a look at something like the D700 vs. K5 - again, an older, less efficient pixel technology in the D700, but it's more efficient than the Canon 5D - so the D700, despite being a worse sensor tech than k5, still has about a stop more noise peformance over the K5. Why? Total light.

The variance here in these sensors is the underlying pixel tech - the constant is the amount of total light they see for the same exposure. So because of this you see, at his point in time AFAIK, *any* FF sensor of any age beating *any* aps-c sensor of any age due to that FF sensor getting more Total Light. How much it beats the aps-c sensor depends on how close in pixel tech the two sensors are. And you also see some of the newer sensors beating the older in DR at base ISO (like K5 vs. D700 or 5D)

Does this clear things up for you? I'm hitting a personal articulation limit where I don't know if I can explain it any more clearly.

.

---------- Post added 10-16-14 at 09:56 AM ----------

So, to summarise the "Total Light" hypothesis:

"Total Light" is the only factor responsible for noise in a digital camera.

Absolutely false.

DD, is that the message you're getting from these threads? Are you actually reading the posts I and others make? Have you followed any external links to source material?

All the other factors that can generate noise in the camera's electronics are irrelevant, because they contradict "Total Light".

Therefore, "Total Light" is the only factor responsible for noise in a digital camera.

Either you're not reading carefully or you're constructing straw-men for your own benefit. C'mon.

.

---------- Post added 10-16-14 at 10:02 AM ----------

Anyway, I'll bow out of this now.

I'd encourage you to follow it if not participate.

I've clearly failed to grasp something fundamental about "Total Light",

You have, but you're not alone.

but since it seems to have absolutely zero useful application in real-world photography I've decided to stop worrying about it.

It establishes the main reason why you might want to shoot with a larger sensor in the first place - larger, as in aps-c vs point and shoot for example - but aside fro that I guess it's unimportant. Truthfully, you don't have to worry about it at all if you only plan to shoot one format anyway. Some of like to understand stuff though, especially if such understanding can help manage expectations and end up saving us money!

[ElJamoquio]

Hence, the sensor size and 'Total Light' cannot be the reason for the difference, EJ ... your example, remember!

My mistake, I was misremembering dynamic range.

Pentax K-5, low light acceptable ISO according to DxOMark: 1162
Nikon D7000, low light acceptable ISO according to DxOMark: 1167
Nikon D800, low light acceptable ISO according to DxOMark: 2853

---------- Post added 10-16-14 at 09:41 AM ----------

???

The FF shot ends up incorrectly two stops underexposed compared to the m43!

I'm not sure what set of assumptions and requirements you're using, so you'll have to spell it out to me for me to be able to explain where your misconception is.

[Dartmoor Dave]

DD, is that the message you're getting from these threads? Are you actually reading the posts I and others make? Have you followed any external links to source material?

You might want to Google the term, "Socratic irony".

[normhead]

I'm guessing everyone here has done their best... I can't think of another angle.. I got nothing.

It was the same with film... too much grain, not enough resolution, go to larger format, some things never change.

[jsherman999]

You might want to Google the term, "Socratic irony".

We do see a lot of that on this subject!

Eventually with enough careful reading and a modicum of research, the "total light is bunk" crowd stops trying to lecture from a weak base of understanding, draining the Socratic Irony they brought to the situation. They come around, individual by individual.

[Dartmoor Dave]

Can we at least agree that Bill & Ted is an awesome movie?