Native ISO
Also known as: base ISO, native sensitivity
The sensitivity a sensor is designed around, where it delivers its cleanest image and widest range before any amplification is applied.
Updated
What it means
Native sensitivity is the setting the sensor was built for. Everything above it amplifies the signal and the noise together, and everything below it, where it exists, throws away highlight range.
It is a property of the hardware rather than a preference. An ISO value can be set anywhere the menu allows, and only one or two points on that scale are where the sensor is actually at its best.
Some sensors have two. A second native point higher up the range exists on bodies designed for low light, which is why a particular higher value can be cleaner than the one below it.
How it works in practice
Staying at base is the default for daylight. Nothing about bright conditions requires amplification, and the extra light is handled with a Variable ND filter rather than by raising sensitivity.
Knowing the number for a body is worth five minutes. It is published for most cameras, and the difference between the base value and one stop above it is visible in any shadow.
Dusk is where the decision gets made deliberately. One or two stops up still grades, and anything beyond that produces a clip that needs Denoise and loses fine detail to it.
What to watch out for
Automatic sensitivity ignores it entirely. A camera left to choose moves wherever the exposure needs, which is why a locked value matters more on a small sensor than almost any other setting.
Intermediate values can be worse than the ones around them. On some bodies the in-between settings are produced in software, so a value chosen between two real ones is noisier than either.
Noise interacts with compression. A high-sensitivity clip at a given Bitrate loses detail everywhere rather than only in the shadows, which makes the penalty larger than it first appears.
How RawClip handles it
Noisier clips are still analysed and used, and cleaner coverage of the same moment tends to win where both exist. Noise reduction is applied during assembly, so shooting near base sensitivity shows up as detail that survives the whole process.