I keep hearing people having issues with noise in their images. I also see people post images with what I consider too much noise at base ISO.
I never have this issue. I use a GX80 but I don't think that the camera model matters that much unless it was the older 12mp models.
Anyhow, below are two photos. They are from the DPR 150-400mm lens review. I noticed the usual "pile on" about the image noise from other format users in the comments. I tend to agree, there is too much noise. The first image is cropped from their JPEG. The 2nd image, I downloaded the raw and opened it using RawTherapee, a free but powerful editor. The point I'm making here is that I did NOT use any noise reduction. It was turned off, yet the noise grain is finer than in their image despite using sharpening. I processed the image to be similar but there's minor differences. There may be better ways to process, certainly worse ways, but for me, it has got to be budget minded.
My personal opinion
Noise in MFT sensors are due to the following vicious circle:
1. User believes small sensor has noise
2. User sets a low ISO as starting point
3. User underexposes the image as ISO was too low
4. User tries to correct the image using a raw editor
5. Optional - User applies some noise reduction also to low ISO shots with tools like topaz etc
ISO maps different level of analogue gain in the camera this cannot be recovered in post processing with digital gain. Set your aperture and shutter based on what you need to accomplish and maximise light then if you need to worry about ISO
This is exactly right in overview, wrong in the details., The problem is exactly caused by people wrongly thinking that high ISO settings cause noise, and thus setting too low ISO for the exposure that they will use. The truth is that the electronic noise added by the camera actually
reduces as the ISO is raised, so noisewise you're best with the higher ISO setting, so long as it doesn't clip the highlights at the exposure you use. There is nothing whatsoever to be gained by using a low ISO and low exposure, it's double jeopardy.
But, it's not a case of substituting 'analogue gain' with 'digital gain'. 'Gain' is not needed for producing a higher ISO. The reason that the voltage gain is there is exactly to reduce the electronic noise at high ISOs, when exposures will usually be low. Digital processing functions do not add noise, so the problem with low ISIO is that the analogue circuitry adds more noise at low ISO settings, not that 'digital gain' is an inadequate substitute for 'analogue gain'.
My statement is entirely correct in so far as the editing correction cannot recover an incorrect exposure as exposure was set at the time of the shot manipulating a RAW files does not correct the exposure by me referred as digital gain.
In terms of your statement is it not true that camera add more noise at low ISO setting clearly each amplifier has a base noise but as gain increases this does not stay flat it also goes up because noise gets amplified too.
The point is that if the level of light is low it makes sense to apply gain as that will amplify also the signal and improve SNR this will in turn reduce the maximum DR but as long as it does not clip this is not an issue. This is the reason when you model input referred noise you see a pattern going downwards it does not mean the level of noise goes down but the level of gain goes up the total maximum SNR therefore still drops as consequence with increase in gain despite read noise is actually going up (but less than gain)
So while it is true that when the image is correctly exposed and you want to maximise DR you should keep gain low the reality is that a lot of situation either don't fill the DR of the camera and therefore you can pump up the gain without issues and improve SNR
OR would clip anyway and what is important is to expose correctly what matters in the image.
Either way you start with exposure and artistic intent that means aperture and shutter and worry about ISO later not the other way around
