Effect of EV on Picture Noise

[Marc Hildebrant]

Group,

Another Hobby of mine is audio restoration, where I remove the noise from old records. I wrote a Handbook on the subject, and I have attached a section where the Signal to Noise ratio of the A to D process is calculated due to first the fact that an analog signal is converted to limited digital values, and second how the use of part of the A to D converter increases the noise. This component is called "Load Factor" or LF. With our digital camaras, this is similar to how much of the histogram is used.

Comments?

Marc

 

[fabiankeller9090]

no comments...

 

[Soocom1]

So with respect, the point is?

 

[Marc Hildebrant]

Soocom1

Is your question for me?

If so, the point is that the highest signal to noise ratio (within just the A to D process) occurs when the full range of the input is used, or the full range of the histogram.

Interesting connection to audio low noise work (where I have spent a lot of time).

Marc

 

[RAZKY]

Soocom1

Is your question for me?

If so, the point is that the highest signal to noise ratio (within just the A to D process) occurs when the full range of the input is used, or the full range of the histogram.

Interesting connection to audio low noise work (where I have spent a lot of time).

Marc

I fail to see how this would (should?) affect how I determine digital photography exposure. Clue me in please.

 

[Ysarex]

I fail to see how this would (should?) affect how I determine digital photography exposure. Clue me in please.

It shouldn't.
The rule is very simple: Expose as much as possible.*

There's different types of noise that can turn up in our digital images. Most are inconsequential to most of us, like very long exposure noise that concerns for example folks photographing the night sky.

For most of us it boils down to read noise and shot noise. Quantization error noted here is read noise. Read noise is introduced in the hardware at various stages of processing. Today in 2022 our camera makers have all but engineered read noise off the table. Read noise may be an issue if you have an older camera. That leaves us with shot noise. Shot noise is randomness in the signal itself and increases as the signal weakens and decreases as the signal strengthens. So there it is; noise depends on exposure. More exposure = stronger signal = less noise. It is just that simple.

Noise is not caused by ISO. Noise correlates with ISO because the primary reason we raise ISO is to force an exposure reduction. Noise is caused by reducing exposure. Here's some fun: I take some variation of this photo regularly for my students when they need to better understand ISO. It's a parlor trick in that what I did was use a tripod and expose more -- so I really didn't need to set the camera at 25600 ISO. This is from an APS-C sensor camera with the ISO set to 25600. No fancy Deep Prime or Topaz denoise, etc. just the normal noise processing in C1.

*Don't overexpose diffuse highlights.

 

[Marc Hildebrant]

Razky,

A previous posting on the forum (do not remember when) suggested that the best exposure setting for a picture uses as much as possible the full range of the sensor, or the histogram shows energy over the full range.

My article gives some mathematical insight into why using the full range of the sensor reduces the quantization noise and increases the overall signal to noise value.

Set the EV value to use as much of the camaras sensor range gives the lowest noise was my conclusion.

Foot note...posting by Ysarex is an excellent explanation.

Marc

 

[RAZKY]

Marc and Ysarex, this is all good stuff for the tech minded, and I enjoy reading it.

 

[RAZKY]

Marc and Ysarex, this is all good stuff for the tech minded, and I enjoy reading it.

I got interrupted and accidently submitted this before I was finished, so to continue - though I do enjoy this technical talk about exposure, determining exposure latitude, film or digital, has always been the simplest part of photography for me.