Impact of shutter speed on noise

[Stagecoach]

A recent post on another forum discussed the reasoning for coloured artifacts observed on an image that had been ppd from a very severely underexposed shot taken at 1/50s, f9, ISO 1250 for which it 'appears' the speedlights had not fired. Simply, data was non existent in much of the area.

Anyway ............................

During the discussion someone suggested "at 1/50s the camera will generate a lot of noise whereas a higher ISO and a corresponding higher shutter speed reduces the time to generate noise and so the grain".

To demonstrate this they posted two images, shot with a D500. The image was of the rear wheel/tyre of a white truck/van/SUV that had a black/grey protector around the arch and black/grey bumper. The images were exposed/brightness well, exp being only 0.3EV difference between the two. What they demonstrated is as follows ...........

Image 1 - ISO 6400 f/5 1/3200. minimal noise (posters info).

This image showed no noticeable noise anywhere.


Image 2 - ISO 6400 f/14 1/320 massive noise (posters info).

This image showed very significant noise/grain on the tyre, inside wheel arch, wheel arch protector and bumper.

So being my normal inquisitive self I undertook numerous testing using the D800 at ISO 6400 for all shots and can not replicate any noticeable difference in noise between equal exposure shots taken at 1/2.5s through to 1/3200s.


So my question is, has anyone else come across a case of a reduction of noise by using a higher shutter speed than lets say 1/50s, increasing the aperture whilst maintaining the same ISO.

[xpatUSA]

So my question is, has anyone else come across a case of a reduction of noise by using a higher shutter speed than lets say 1/50s, increasing the aperture whilst maintaining the same ISO.

Yes, although my experience (Sigma) would be of little interest to most of us.

On a 95+ degree day, I shot my SD10 at about 1/30 sec and also at 30 sec (camera max) with the ISO set to 1600 (extended!). With the 30 sec shot, the color blotching almost totally obscured any detail in the shot. At 1/30 sec, the noise was kinda tolerable. Sensor got quite hot during the long exposure.

I did another series from 0.6 sec to 25 sec at 800 ISO here:

http://kronometric.org/phot/sensor/f...nd_temp800ISO/

The difference in noise is quite plainly visible.

[Stagecoach]

Yes, although my experience (Sigma) would be of little interest to most of us.

On a 95+ degree day, I shot my SD10 at about 1/30 sec and also at 30 sec (camera max) with the ISO set to 1600 (extended!). With the 30 sec shot, the color blotching almost totally obscured any detail in the shot. At 1/30 sec, the noise was kinda tolerable. Sensor got quite hot during the long exposure.

I did another series from 0.6 sec to 25 sec at 800 ISO here:

http://kronometric.org/phot/sensor/f...nd_temp800ISO/
.

I'm aware that this can be the situation with 'long' exposures Ted and can be increased with temperature but I had never come across any mention with higher speeds. Perhaps my mention of speeds higher than 1/50s in the question was a bit unclear.

But, just a few minutes ago I trawled through the Exif (using Jeffreys Friedls) of one of the posted images and have noted a suspicious value of +2.26 with the tag of Exposure 2012?

I'll delve deeper ......................................

[xpatUSA]

... Perhaps my mention of speeds higher than 1/50s in the question was a bit unclear.

Yes, I did miss that, duh.

[Stagecoach]

I'll delve deeper ......................................

Continuation,

I have found that the two images posted to 'demonstrate/support' that noise in an image shot at 1/320s, f/14, ISO 6400 would be significantly reduced when shot at 1/3200s, f/5, ISO 6400 appear to have had their exposure altered in PP. (although the Exif given by the poster shows identical ISO and exposure (speed/aperture).

The image with the significant noise had +2.26 in the Exif tag 'Exposure 2012'. The image that showed no noise had -1.43 in the Exif tag 'Exposure 2012'.

The Exif tag 'Exposure 2012' from a test with an image of my own is the value of 'exposure correction' entered in PS ACR during PP.

The only way I can think that the poster could have achieved this was by using equivalent exposure settings (S & A) with the same ISO on each image but with one shot taken with a ND fitted, or, a significant lighting difference

[DanK]

I can't see the images. The links don't work for me in either Firefox or Vivaldi.

So my question is, has anyone else come across a case of a reduction of noise by using a higher shutter speed than lets say 1/50s, increasing the aperture whilst maintaining the same ISO.

Essentially, no.

Very long exposures are another matter. I had one very long exposure, I think it was 20 or 30 minutes, taken on a hot night with a Canon 50D that was ruined by noise from overheating. However, I can go to 10 minutes with no problem in most cases. For example, the following is a 7 -minute exposure at ISO 100 taken with a 5D III, with no noise reduction in postprocessing:



You can see that it has very little noise. If you want to see the full-size JPEG, you can see it here.

I think the original poster you are quoting had several things mixed up. First, what matters isn't the quantity of noise; it's the signal/noise ratio. Second, s/he may be ignoring that fact that there is more than one kind of noise: photon (shot) noise, several kinds of front-end read noise, and back-end read noise. Increasing ISO and decreasing shutter speed increases back-end read noise. In every instance where I have looked, the result of increasing ISO, if noticeable, has been to increase the overall noise in the image.

[Manfred M]

Without seeing the original images, I would be very hesitant to comment.

The D500 has the same sensor and image processor as my wife's D7500, which has remarkably low noise at that ISO so long as it is a "good exposure". Take a shot of a dark subject, like in the example, and underexpose it, there will be noticeable noise after the image has been post-processed to brighten things up. That would occur irrespective of shutter speeds in this range unless the camera has been kept on in LiveView for some long period (measured in many minutes) to get the sensor and other circuitry quite warm.

I used a black background and largely black subject matter just to ensure the character of the images described in #1' black tire in a shaded wheel well. The images are SOOC JPEG: I lit the scene with the halogen modeling light from my studio light and used my incident light meter to ensure that I had the same ISO, shutter speed and aperture as in your example. This ensured "correct exposure" without having to worry about the camera's metering system getting fooled by the predominantly dark / low key shot. Shot with Nikon D7500 which has the same sensor and processing engine as the D500:


1. ISO 6400, 1/3200 sec shutter speed at f/5






2. ISO 6400, 1/320th sec at f/14




I don't see a lot of noise and definitely no difference in noise between the two shots. I have posted full-sized images to keep the pixel peepers happy. These results are in line with what I would have expected; the only difference I see in the two images are that the light meter moved between the shots...

Now just for fun, I underexposed the shot by 2-stops: ISO 6400, 1/1250th sec at f/14 and then brightened in post to roughly the same levels as the two "properly" exposed shots, I am seeing more noise, as expected. I used the green bubble level as my reference point for the exposure adjustment based on a sampler tool I dropped there to get the readings.

3.





The person who made the comments either has a faulty methodology or is writing something to get a reaction?


P.S. - Just for the sake of completeness, this is the shot used in point 3 without any work in PP.

4.

[Stagecoach]

I can't see the images. The links don't work for me in either Firefox or Vivaldi.



Essentially, no.

Thanks Dan. I'm not sure what he had mixed up but I'm convinced his examples and explanation were not showing the full story

[xpatUSA]

I think the original poster you are quoting had several things mixed up. First, what matters isn't the quantity of noise; it's the signal/noise ratio.

By coincidence, I just measured that in Manfred's examples, using ImageJ histograms set to luminosity (.3R+.6G+.1B).

Used SNR = mean / standard deviation.

In a darkish area 2.339 v 2.791: In a medium-gray area 8.696 v 10.291.

Won't insult you with which is which.

[Manfred M]

By coincidence, I just measured that in Manfred's examples, using ImageJ histograms set to luminosity (.3R+.6G+.1B).

Used SNR = mean / standard deviation.

In a darkish area 2.339 v 2.791: In a medium-gray area 8.696 v 10.291.

Won't insult you with which is which.

I posted four images. Which ones did you do the comparisons of?

The differences in #1 and #2 would be of interest as this is where the only variable was the shutter speed (and the light level). I had to turn down the modeling light a bit to match the exposure in those two images. 2/3 ev.

[xpatUSA]

Continuation,

I have found that the two images posted to 'demonstrate/support' that noise in an image shot at 1/320s, f/14, ISO 6400 would be significantly reduced when shot at 1/3200s, f/5, ISO 6400 appear to have had their exposure altered in PP. (although the Exif given by the poster shows identical ISO and exposure (speed/aperture).

The image with the significant noise had +2.26 in the Exif tag 'Exposure 2012'. The image that showed no noise had -1.43 in the Exif tag 'Exposure 2012'.

The Exif tag 'Exposure 2012' from a test with an image of my own is the value of 'exposure correction' entered in PS ACR during PP.

The only way I can think that the poster could have achieved this was by using equivalent exposure settings (S & A) with the same ISO on each image but with one shot taken with a ND fitted, or, a significant lighting difference

Well done, Sherlock!

[Stagecoach]

The person who made the comments either has a faulty methodology or is writing something to get a reaction?

Thanks for the test examples Manfred. Your test, settings and results are identical to what I had done and confirm the same as my observations that there's no noticeable difference in noise between shots taken at 1/3200s and 1/320s with the same ISO.

I have gone back to him with my findings and will see what transpires, if anything.

[xpatUSA]

I posted four images. Which ones did you do the comparisons of?

#1 and #2.

[DanK]

I'm now confused. The original post was this:

at 1/50s the camera will generate a lot of noise whereas a higher ISO and a corresponding higher shutter speed reduces the time to generate noise and so the grain

That's what I was responding to. But the thread is also discussing this:

there's no noticeable difference in noise between shots taken at 1/3200s and 1/320s with the same ISO.

[xpatUSA]

... But the thread is also discussing this:

there's no noticeable difference in noise between shots taken at 1/3200s and 1/320s with the same ISO.

That's Manfred's shots 1 and 2. A slight difference - See post #9.

As Bill Clinton might say: "it all depends on what the meaning of 'noticeable' is ..."

[Stagecoach]

Ok Dan I'll try and make things clearer .....................

I'm now confused. The original post was this:

"at 1/50s the camera will generate a lot of noise whereas a higher ISO and a corresponding higher shutter speed reduces the time to generate noise and so the grain".

My original post 1 'contained' the sentence above which was a statement made by the poster and the speed of 1/50s mentioned was that of which the original poster with the severely underexposed image had used.

This poster to support this statement of his had posted two images, shot within a minute of each other, same high ISO on each but with different speeds/apertures (they had a difference of 0.3EV equiv EV as I mentioned and for my testing I retained this difference as close as you can), the one with the higher speed showing no noise but excessive noise on the shot with the lower speed of 1/320s.

That's what I was responding to.

Yes I understand, but the question was the last paragraph of post 1 in bold italics. Perhaps I should have mentioned that my concern/interest was not with long exposure images of which we know can generate noise but if 'significant' noise differences can be noticeable between images (of the same exposure) shot at the same high ISOs but with different speeds. I had mentioned speeds higher than 1/50s.

But the thread is also discussing this:

there's no noticeable difference in noise between shots taken at 1/3200s and 1/320s with the same ISO.

Which is the conclusion that I and Manfred came to using the same parameters of the posters examples. I understand that the word 'noticeable' is subjective but my own interpretation of "not noticeable" was opposed to viewing someones tests of which my interpretation was "WOW"

I hope this explains.

[DanK]

Grahame,

Thanks for the explanation. Sorry to put you through that. I just hadn't read the first post carefully enough because I was about to leave the house. My bad, as people here now say.

Dan