[Dartmoor Dave]

I hope you'll forgive me for taking a couple of crops from the black areas of the example in your original post. This is just two crops of identical pixel dimensions side-by-side with no adjustments of any kind, and I can't see any difference at all in the amount of noise. In one of the crops the noise appears perceptually more saturated because it's against a slightly darker background, but as far as I can see the actual amount of noise is the same.

[GUB]

I hope you'll forgive me for taking a couple of crops from the black areas of the example in your original post. This is just two crops of identical pixel dimensions side-by-side with no adjustments of any kind, and I can't see any difference at all in the amount of noise. In one of the crops the noise appears perceptually more saturated because it's against a slightly darker background, but as far as I can see the actual amount of noise is the same.

Hey it is great to have someone engage in the discussion!. Agree with your call on the dark areas. And in these areas the f8 may well be a little lighter and less contrast because the background highlights have bokehed right out.
But I see a real difference in the petals - there is far more colour noise in the f45.

[UncleVanya]

I would say that without a controlled light source and calibrated shutter speed and aperture it is very hard to definitively say much. Also I wonder if the circle of confusion starts to play a role with very small aperture shots. In any case it is interesting.

[AstroDave]

This article mentions it. Of course it may be BS but it would explain your findings.
More than you wanted to know about noise - Improve Photography

After noting this guy saying: Heat is a form of electricity

I wouldn't believe anything he says! That is just plain wrong.

The right way to do this is to get the actual pixel values (actual raw R, G (two channels), and B data values) for equivalent areas in the comparison frames, calculate the mean, and the standard deviation. Show a plot of the noise distribution around the mean, and then draw some conclusions.

[zapp]

We assume a linear relationship between light collection of the sensor and integration time. The system should be tuned n this way for a wide range of exposure times. F-stop changes integration time, it is not a noise producing factor by itself. You are basically checking whether linearity applies at iso 51200.
F/45 will produce be influenced significantly by diffraction, this may have a small influence on noise distribution, but as said above, you need to do an area based calculation.
Avoid f/45, avoid iso 51200.

[Wheatfield]

I suppose this is why they put an option for reducing noise in long exposures on the camera.

[DonV]

It may be of "academic" interest, but I see no real difference between the two images; I would say both are equally "noisy".

[stevebrot]

How am I missing the point? The aperture starved one has more noise because the shutter is open longer at the same ISO.

In theory, the photon flux is the same as is the sensor gain for both cases.* If heat build up at 1/250s is a concern at ISO 51200, then it appears there is a strong case for short exposures regardless of ISO.

A dark frame exposure for each case might be telling and might prove your point.



Steve

* In case it is not obvious, in both cases the sensor is "starved" in terms of photon flux to the sensor. The tests simply compare the extremes for both lens aperture and shutter speed at the ISO suggested by the meter based on an open aperture reading. The difference in apparent brightness means either shutter of aperture mechanism is not accurate for one or both exposures.

[stevebrot]

Also I wonder if the circle of confusion starts to play a role with very small aperture shots.

I wondered the same thing. What happened to the specular reflections in the f/45 shot? Did the light change or are they reduced to veiling flare in the wide-open shot?


Steve

[GUB]

I wondered the same thing. What happened to the specular reflections in the f/45 shot? Did the light change or are they reduced to veiling flare in the wide-open shot?


Steve

If you mean the highlights in the background then they were moving around gently in a breeze and also the shots were handheld. Having said that it is reasonable to presume they have dissolved into the background in the open shot as I said in post 17.
If the result had shown more difference between the two I would be interested in doing a more calibrated setup. Perhaps with a faster lens say f1.4 to f8 . It's reasonably large dimensions would beless prone to minor sticking of the iris. Controlled lighting and a control exposure shot at lower iso.
I am going with there is a difference in noise in the lighter area .
Does anyone disagree with that.?
Is there some reason we need to be looking only at the dark area.?

[stevebrot]

Does anyone disagree with that.?

I don't know what to say. The recorded noise is definitely different, but it hard to say if it is more pronounced. Provide a dark frame for comparison. Paucity of light, by itself, does not cause noise.


Steve

[GUB]

I don't know what to say. The recorded noise is definitely different, but it hard to say if it is more pronounced. Provide a dark frame for comparison. Paucity of light, by itself, does not cause noise.


Steve

Here ya go - not done one of these before, is this what you mean? they are the same size crops as the OP shot. They appear identical. In a lighter area is it taken that this same noise is overlaid on a light surface (rather than a black)?

[stevebrot]

Here ya go - not done one of these before, is this what you mean? they are the same size crops as the OP shot. They appear identical. In a lighter area is it taken that this same noise is overlaid on a light surface (rather than a black)?

Thanks, that is what I expected, though I would not have been surprised had it been a tad noisier with the long shutter opening. (I assume the sensor turns off between exposures?)


Steve

[Bob 256]

If you do a Gaussian blur of each photo (which average adjacent pixels) it appears that granularity disappears with the same setting (blur circle) for each so from that, I would assume the noise in each is very nearly equal. Any small differences are probably due to errors in aperture and/or shutter speed (as is probably the case for overall exposure).

[GUB]

Thanks, that is what I expected, though I would not have been surprised had it been a tad noisier with the long shutter opening. (I assume the sensor turns off between exposures?)


Steve

It wasn't on liveview .