ISO Invariance Explained With Real Test Shots

ISO invariance is the idea that, for a sensor that is truly invariant, raising the ISO in camera produces the same noise as brightening the image later in software, which has real consequences for how you expose. A truly invariant sensor lets you underexpose to protect highlights and recover the shadows in post without a noise penalty, and testing how close a camera comes to that ideal changes the way you shoot.The test is straightforward. Take one shot at a low ISO and lift it several stops in software, then take the same shot at a higher ISO with the same exposure settings, and compare the two files. On a perfectly invariant sensor the noise is identical, because the sensor adds the same read noise regardless of the ISO setting, and the only difference is whether the gain was applied in the camera or in the editor.Real sensors sit somewhere short of perfect invariance. Most modern sensors behave nearly invariantly above a certain ISO threshold, often around 800 or 1600, which means that once you pass that point there is little benefit to raising ISO further, and you are better off