When the autopilot detects severe noise in an image, why isn’t the RAW Strong noise model used instead of the RAW Normal?
