What Does ISO Mean in Digital Photography?

[Neil S.]

My questions are not motivated by a desire to become a better photographer, but by a desire to understand photo theory better out of pure scientific curiosity.
-TC

OK...

I think I understand now lol.

It was hard for me to tell where you were coming from there for a while.

To try and answer your question for Mike...

I think that you need to be looking into image noise in general, as that is the root of what you want to know.

Wikipedia: Image Noise
Image noise - Wikipedia, the free encyclopedia

Also this is good info

Wikipedia: Film Speed
Film speed - Wikipedia, the free encyclopedia

Digital camera ISO speed and exposure index (about 1/2 way down)

In digital camera systems, an arbitrary relationship between exposure and sensor data values can be achieved by setting the signal gain of the sensor. The relationship between the sensor data values and the lightness of the finished image is also arbitrary, depending on the parameters chosen for the interpretation of the sensor data into an image color space such as sRGB.
For digital photo cameras ("digital still cameras"), an exposure index (EI) rating—commonly called ISO setting—is specified by the manufacturer such that the sRGB image files produced by the camera will have a lightness similar to what would be obtained with film of the same EI rating at the same exposure. The usual design is that the camera's parameters for interpreting the sensor data values into sRGB values are fixed, and a number of different EI choices are accommodated by varying the sensor's signal gain in the analog realm, prior to conversion to digital. Some camera designs provide at least some EI choices by adjusting the sensor's signal gain in the digital realm. A few camera designs also provide EI adjustment through a choice of lightness parameters for the interpretation of sensor data values into sRGB; this variation allows different tradeoffs between the range of highlights that can be captured and the amount of noise introduced into the shadow areas of the photo.
Digital cameras have far surpassed film in terms of sensitivity to light, with ISO equivalent speeds of up to 102,400, a number that is unfathomable in the realm of conventional film photography. Faster processors, as well as advances in software noise reduction techniques allow this type of processing to be executed the moment the photo is captured, allowing photographers to store images that have a higher level of refinement and would have been prohibitively time consuming to process with earlier generations of digital camera hardware.

-----------------------------------------------------------------------

Maybe "signal gain" would been a better way to say it.

- Neil

 

[Neil S.]

Wikipedia "Image Noise":

"Amplifier noise (Gaussian noise)

The standard model of amplifier noise is additive, Gaussian, independent at each pixel and independent of the signal intensity, caused primarily by Johnson–Nyquist noise (thermal noise), including that which comes from the reset noise of capacitors ("kTC noise").[2] In color cameras where more amplification is used in the blue color channel than in the green or red channel, there can be more noise in the blue channel.[3]
Amplifier noise is a major part of the "read noise" of an image sensor, that is, of the constant noise level in dark areas of the image."

-----------------------------------------------------------------------

This is about the best that I can do with some quick research. (I am learning stuff here too!)

It seems to me that it may be caused by heat, and as the signal gain (and probably voltage) is increased so is the noise.

It may be similar to how overclocking your CPU too far can lead to errors and an unstable system.

This is likely because there is increased electrical noise as a result of the higher temps from increasing the CPU's voltage.

If you can manage to lower the CPU's temps (with extreme cooling) it often will stabilize it as well.

This is why liquid nitrogen cooling is often used for world record overclocks, and allows the CPU to run at sub-zero temps.

This is making me want to use extreme cooling on my DSLR sensor to see if it lowers noise lol.

Something tells me that it may void the camera's warranty though...

- Neil

Edit: Can anyone say if they have heard of or experienced lower image noise at high Iso when shooting in a very cold environment?

If the noise was based on heat, then it would seem logical that there would be lower noise when the camera was very cold, and more if it was very hot.

I may be totally wrong on the thermal thing by the way, it just seemed logical based on the wikipedia article...

 

[supraman215]

The noise in high ISO is the same as the noise in low light. It's electronic noise produced by the capture device. The stereo example given before is a good one. If you have a stereo turn it all the way up. You're not actually turning up the amplification properties of the amp, these are constant, You're allowing more and more input signal into the amp (gain). Same with increasing ISO (gain). Even if nothing is hooked up to your amp and you turn the volume (gain) up you will hear noise. This is just the imperfections of the equipment being amplified. most of the time at normal volumes they are undetectable by the human ear but turn up the gain and they are aparent. Same with noise. Take a picture of complete darkness at iso 200 and iso 3200 asuming you have achieved complete darkness these pictures should be the same with the exception of the added noise to the iso 3200 shot. As with audio equipment there are various 'noise gates' you can add to pictures, in the camera with hardware or in PP with software. Of course noise gates aren't perfect and as with audio they reduce the quality somewhat. As the quality of the sensors increase so will the level of noise at high iso's and low light increase. It's a simple question of the quality of the equipment. A more sensitive chip will be more precise and more expensive.

If camera's had started as electronic devices then I'll bet it wouldn't be called iso at all it would be gain, or volume.

 

[Garbz]

Edit: Can anyone say if they have heard of or experienced lower image noise at high Iso when shooting in a very cold environment?

If the noise was based on heat, then it would seem logical that there would be lower noise when the camera was very cold, and more if it was very hot.

I may be totally wrong on the thermal thing by the way, it just seemed logical based on the wikipedia article...

You are totally right. Have a look at sensors designed for astophotography. They typically have 2 things in common, long exposures and heatsinks. Or the CCDs sitting in one of the cameras in the Spitzer Space Telescope which is cooled by liquid Helium to a very chilly 4 degrees Kevlin (-452F).

The noise in high ISO is the same as the noise in low light. It's electronic noise produced by the capture device. The stereo example given before is a good one. If you have a stereo turn it all the way up. You're not actually turning up the amplification properties of the amp, these are constant, You're allowing more and more input signal into the amp (gain). Same with increasing ISO (gain). Even if nothing is hooked up to your amp and you turn the volume (gain) up you will hear noise. This is just the imperfections of the equipment being amplified. most of the time at normal volumes they are undetectable by the human ear but turn up the gain and they are aparent. Same with noise. Take a picture of complete darkness at iso 200 and iso 3200 asuming you have achieved complete darkness these pictures should be the same with the exception of the added noise to the iso 3200 shot. As with audio equipment there are various 'noise gates' you can add to pictures, in the camera with hardware or in PP with software. Of course noise gates aren't perfect and as with audio they reduce the quality somewhat. As the quality of the sensors increase so will the level of noise at high iso's and low light increase. It's a simple question of the quality of the equipment. A more sensitive chip will be more precise and more expensive.

If camera's had started as electronic devices then I'll bet it wouldn't be called iso at all it would be gain, or volume.

Actually the analogy is wrong. The the methods of operation for the signal amplifier on the sensor and the volume control on your amp are distinct and separate. One is a variable gain device.

You're right in your explanation of the hifi amplifier. The gain is set and constant and the overall volume is varied by varying the signal you put into it, however this is not the only way the amplifier can work, think that gain of your amplifier (Vout / Vin after the volume control) is set by something, and that something can often be adjusted.

The amplifier for a photosensor isn't actually a voltage amplifier at all, it's a transimpedance amplifier. It's gain is Vout / Iin, produces a specific voltage for a given current, and that voltage is often generated across a resistor in it's own feedback loop.




From the beginning we have:

Sensor -> I/V Converter -> analogue to digital converter -> pretty picture.

Noise is caused in the sensor in the form of shot noise due to quantum noise, it's caused in the I/V converter (amplifier) in the form of shot noise or johnson noise, it is caused in the analogue to digital converter in the form of quantisation noise, it can be brought in via imperfect electronics (power supply noise), or even it can come from the sun or the powerlines or your mobile phone in the form of coupling from external radiation.

The formula for Johnson Nyquist noise includes a term "R". Remember that a transimpedance amplifier has a gain of Vout/Iout? Well good old electronics 101 says V=IR so the amplifier naturally is a source of Johnson noise. These fundamental limits of physics is why it's so important to match each and every component in the system so that not even the slightest bit of the signal is lost on it's way through.

Now we have limited control over photon -> electron conversion at the sensor by adjusting the bias. Moving the bias will affect the charge in the device. It's possible to get a boost in linearity at the expensive of sensitivity. We have a lot of control over the transimpedance gain and these are typically variable gain devices, where really they do generate larger and smaller signals but in doing so naturally introduce more and less noise. We also have control over the analogue to digital conversion (an 8bit number could represent 0-5V, it could also represent 0-1V)


Typically the gain of the system as a whole, and thus the ISO of the camera is adjusted by playing with the last two elements, the dynamic range of the analogue to digital converter, and the gain of the current to voltage converter.

And if you thought that is complicated, there's whole books written about designs of photo diodes and how to bias them alone, and there's whole libraries written about the designs of amplifiers too

 

[supraman215]

Great explanation! though it's a little over my head, I only had a very terse understanding. Was I correct in my estimate that low light noise is the same as high iso noise? Sounds like not.

 

[Neil S.]

And if you thought that is complicated, there's whole books written about designs of photo diodes and how to bias them alone, and there's whole libraries written about the designs of amplifiers too

WOW this is really good info!

What do you do for a living? Are you an engineer or an electrician?

Ty for the in depth explanation, and I didn’t think anyone here understood this so well.

The most useful thing here for me is knowing that lower temps result in lower noise. This is very interesting…

I really want to overclock+extreme cool my 7D's sensor now lol...

- Neil

 

[Garbz]

Great explanation! though it's a little over my head, I only had a very terse understanding. Was I correct in my estimate that low light noise is the same as high iso noise? Sounds like not.

I'm not sure what you mean by low light noise, can you elaborate? If you mean when you have a darker picture you can see more noise, the noise is the same and the sources are from all the above however the key differences in that case is a low signal to noise ratio. And that's ultimately what everything is about. High ISO amplifies both the signal and the noise, but if the signal is crap (low light) to begin with... I think we're talking about the same thing.

The most useful thing here for me is knowing that lower temps result in lower noise. This is very interesting…
I really want to overclock+extreme cool my 7D's sensor now lol...

Like this guy: Peltier Cooling of Modified Canon Digital Rebel XSi (450D) - Version III -by Gary Honis

Check out the 5 minute long ISO1600 exposures towards the bottom with and without the peltier cooler applied to this camera.

 

[supraman215]

Yes signal to noise! That's the piece I was missing, if you have low signal noise becomes more obvious. Gotcha.

 

[Big Mike]

And when talking about S/N ratio, there is this...
Expose Right