when I take two pictures, one at ISO 1000 and the other at 2000 and open them in RAWtherapee, they don’t appear to have the same brightness when turning the profile to neutral. This means that before the actual profile you create in RAWtherapee, there has to be a preliminary ISO profile at work.
I have just performed the experiment you described, and
here, both exposures have the same brightness in RT. Stupid question:
do you have your camera set to Automatic, Program och Manual?
Yes, ISO affects the raw data. It is more than just metadata.
You may wonder whether shooting an underexposed photo at ISO800 and then increasing the exposure compensation in RawTherapee by say 3 stops leads to the same result as if you had shot the photo at ISO6400, all other parameters being equal. Well, that depends on the camera, but generally the result won’t be identical. More info here: http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/
Hm… Since I performed the same experiment and did not notice any difference in brightness: might there be a significant difference between how a Bayer and an X-trans equipped camera is treated? (I am using a Fuji X-T1.)
I think ``ISO Invariance’’ is the term commonly used for this. FUJI XT1 is one of the most ISO Invariant camera. Canon EOS 6D (what I have) is one of the least ISO Invariant camera.
Both camera and RT need to amplify the signal with an increase in ISO or EV, but the amplification happens at vastly different places (the camera might be able to do it at a higher bit depth and in analog) which can lead to visible differences.