chiron
Offline
Upload & Sell: On
|
 The main (theoretical) differences between APS-C and Full-Frame? | |
ruthenium wrote:
chiron wrote:
ruthenium wrote:
chiron wrote:
ruthenium wrote:
jaygould wrote:
gunmetal wrote:
That's ISO performance, not 'color depth'. And as many have already pointed out software has mitigated much of the deficiencies of smaller sensors in that area. I think 'color depth' is either something people want to see or just manufacturer color science.
They are interlinked (ISO and color). ISO affects color.
No, software cannot correct something that isn't captured by the sensor. Well, maybe you can now with AI, but then we are not discussing photography anymore.
ISO is a measure of the amount of light per unit sensor surface. In low light (at high ISO), the sensor signal to noise ratio is low; thus, the signal from many pixels becomes random.
I believe that ISO is only indirectly a measure of the amount of light on the sensor. When you change the ISO, you do not change the amount of light hitting the sensor, which is what does happen when you change the aperture or shutter speed. And you can choose to set ISO to a higher or lower value at the same level of light and with the same aperture and shutter speed in order to achieve a particular effect, i.e., overexposure or underexposure.
In digital, I believe that ISO is a measure of the amplification that is applied to the electrical outputs coming from the sensor. Base ISO is without amplification. As ISO is increased, the amplification of the sensor's electrical outputs is increased. Amplification of the sensor's outputs increases noise. In film, ISO is a measure of the sensitivity of the film's surface emulsion to light energy.
You are correct, ISO is the signal gain. When decided by the camera, ISO does correlate with the light density; a high auto ISO is typically caused by low light. Indeed, this relationship is not simple, and one can manually set a lower ISO and collect more light by increasing the exposure time. Note, however, that it is literally incorrect to say that "Amplification of the sensor's outputs increases noise." Changing the ISO to a higher value doesn't work to increase noise; it doesn't add noise.
Actually, I believe that the amplification does add noise. The amount of amplification is determined by the ISO setting. This amplification takes place after the light has already struck the sensor and has been converted to an initial electrical signal, i.e., after the "picture" has been taken.
The most common and troublesome type of noise in our images is read noise. Read noise is produced by the camera's electronics and is combined with the image signal as part of the read-out of the sensor. The greater the amplifcation during the readout, the more read noise is produced by the electronics and then combined with the image signal. More amplification equals more noise.
Read noise can produce both chroma and luminance noise and adversely affects dynamic range. This is why in Bill Claff's charts the dynamic range of any given camera falls as the ISO setting increases.
My knowledge of the subject is not from photography but from spectroscopy, where we are interested in the signal to noise ratio, rather than the noise alone. When the signal is weak, the gain is increased to better digitize the low-intensity signal(s). To the best of my knowledge the increased gain does not add noise. I cannot think of a reason it should add noise in a camera. I cannot claim this because of my limited expertise, but there are experts among FMers who hopefully should be able share their knowledge and understanding of the way how ISO gain works in tge modern cameras.
[EDIT: Parts of what follows are clearly wrong.]
Yes, I understand, and my own knowledge of these issues is also limited--I am not an engineer.
When the incoming photons hit the sensor, they are subject primarily to three types of noise: fixed pattern noise, dark or thermal noise, and photon or shot noise. There is a fourth type of noise that is added when the signal in the sensor is read out and amplified--it is called read noise. Read noise is usually the main source of the troublesome noise in our images. [EDITIt is produced by the camera's own electrical process and the greater the amplification, the more of it is produced. At higher ISO settings more read noise is being generated and added to the signal, which is the primary reason we have more noise at higher ISO settings. Different sensor and electrical designs can make this better or worse.
It is more like an audio amplifier than it is like the "noise" in film. The more one turns up the amplification of an audio signal, the more distortion one hears in the music.
I'd be happy to have this account corrected by an electrical engineer if it is wrong or needs tweaking (in which case I would then have the advantage of understanding it better). [
|
