>>3893180>lol, the entire point is that if the scene is lower contrast than the sensor can handle, you get better SNR with higher ISO levels for a given exposure value. You do not. Max SNR comes with ETTR at base ISO.
>In your example sun and clouds will clip when ISO is raised but the foreground will become much cleaner. The sun will clip no matter what you do. As for the foreground, if the scene DR is less than the sensor's DR then the shadow detail will be well above the lowest zones where read noise is relevant. If the scene DR is equal to or greater than the sensor DR, then you have shadow detail that deep. But if you raise ISO those details will have worse total noise than before because the shot noise increase will be greater than the read noise decrease.
Again, this is intuitive to people who actually shoot such scenes. They can see it in their RAWs.
>Expose for something that sits at 12dB SNR with base ISO, then up the exposure one stop without changing the SS and aperture so that you lift that metered thing. You'll get better SNR for it. No, you will not. Your total noise will be worse because your shot noise increase will be greater than your read noise decrease.
>Unless of course your sensor is nearly isoless, like it's for the Pentax. In that case you lose 1 stop worth of highlights without much gain in the shadows.Here are two Canon bodies known for high read noise and therefore poor DR (i.e. limited ability to push shadows in post). That initial jumpiness where ISO 160 is slightly better than ISO 100 (for example) is the closest any FF ILC is likely to get to your theory. Peak DR still occurs at a very low ISO (2/3ev above base). Nothing shipping today has read noise as bad as these cameras.
You're only correct on decade old cameras with terrible read noise, and even then only in that 2/3ev above base ISO might have slightly better DR than base. Any modern sensor, including from Canon, has read noise so low that psn limits DR.
