[this post makes the possibly dubious assumption that I have more or less figured it out]
The IMX571 sensor features "Low-Conversion Gain" (LCG) and "High-Conversion Gain" (HCG). Some manufacturers have their camera automatically switch from LCG to HCG at a specific gain value (I think for ZWO LCG is 0 - 200, HCG kicks in automatically above 200).
The Risingcam (and, I presume, Omegon, Altair, etc that are Touptek cameras) uses gain values from 100 and up.
People on CN have performed sensor analysis on the Risingcam ATR3CMOS26000KPA, using the native driver. Those analyses consistently show that the best gains to use for maximum dynamic range are LCG/100 and HCG/100. The dynamic range is very similar between those two settings.
The INDI driver has a 'checkbox' style entry that allows you to set LCG or HCG. I decided to use LCG/100 for unfiltered OSC imaging and HCG/100 for imaging with the Antlia AlpT duoband filter. But, during one recent narrowband session I happened to notice that the LCG / HCG switch setting had reverted to LCG, even though I had looked at the start of the session to make sure it was on HCG. Some investigation led me to realize that the INDI driver uses a 'switch at a certain gain' mechanism, but the gain to switch at is adjustable. I always knew the "HCG Threshold" setting was there, but I assumed incorrectly that it was for the 'HDR' mode.
So, if you leave HCG Threshold at the default 900 and think you have manually set Dual Mode to HCG, when you tell the camera to take an image at Gain 100 it switches the mode to LCG. Going forward I will try setting the Threshold to 102, and using HCG/102 for narrowband imaging.
Does anyone know what happens if you change the HCG/LCG gain ratio setting?