ISO adjustments can happen in any of the following places:
1) in the conversion of charge to voltage
2) In programmable gain amplifiers after the source followers and multiplexing
3) in the camera's firmware
4) in the raw developer, if the image is a raw one
In theory, you could do ISO adjustments by programming the ramps fed to the comparators in the single-slope ADCs, but I don't know of any sensors that work that way. I'd have a hard time figuring out if that's what they're doing by treating the camera as a black box.
Sensors are not rated in ISO units these days, what Jim lists as 1. & 2. are more of the noise counter-measures. I would say that in-camera JPEG engine and raw converters account for 1. & 2. when establishing default midtone brightness according to ISO standard (but ignoring that standard is not out of question too).
I can't argue with that. I was talking about gain introduced by the control labeled ISO, concentrating on the raw data and ignoring the details of the JPEG conversion.
pretty much, but extended iso happens after the ADC from what ive read. I have also done some extensive reading into how valid the exposure triangle is today using digital cameras. and have come to the conclusion that there is absolutely nothing wrong with it. in fact there were never ever exposure meters in the days of film only light meters, the term exposure meters was born because the light meters had printed tables on them that you had to align. asa/iso had to be set as the default value before you could even read off shutter and aperture values. so to say that iso has no usable implementation even today is just plain silly.
I wish that was true, even when parents pay up to $4000 on dance lessons per year and $1000s of dollars on costumes that get worn for 10min, squeezing $10 for a 5x7 print is like extracting blood from a stone. but i work on mass sales 1000 images sold per school is why i keep doing it. work is work.
Many people have accepted the same misconception but that is their choice to make.
An example of the thing that the exposure triangle teaches that is wrong is it teaches that
f/8, 1/400s, ISO 200
and
f/8, 1/200s, ISO 100
are the same exposure - amount of light that struck the sensor per unit area while the shutter was open - when in fact they will output the same image lightness with the ISO 200 shot having received only half the exposure of the ISO 100 shot and so the ISO 200 shot will have more visible noise.
Image lightness and exposure, although related, are two different things which the exposure triangle does not teach.
With the exposure triangle you could argue that for a constant image lightness, not exposure as defined above, there is an inverse relationship between any two of aperture, shutter speed and iso.
But I don't need a triangle or any other shape to understand that.
Nobody will catch me arguing thus ... my argument has usually been that the ISO knob should be re-labeled "noise" or, if there's room, "degree of under-exposure", LOL.
That is because it wasn't the intention behind my op.
For anyone needing advice/help on how to set exposure correctly with the aim to minimise visible noise then my post "Why are my photos noisy?" might help or at least steer you in the right direction.