ISO invariance is a sensor property describing how raising ISO in the camera produces essentially the same final image as raising the exposure of a lower-ISO raw file in post. On a fully ISO-invariant sensor, an image shot at ISO 1600 looks identical to an image shot at ISO 100 and pushed four stops in editing, because the camera’s analog gain stage adds no additional noise above what was already present. The concept reshapes how photographers should think about exposure in low light.
Most modern Sony, Fujifilm, and Nikon sensors are largely ISO-invariant above a threshold (often ISO 320 or ISO 640), thanks to dual-gain architectures and very low read noise. Canon’s sensors have closed the historical gap but were less invariant in the 5D Mark III and 6D era, when boosting shadows in post produced visibly more noise than raising ISO at capture. Photonstophotos.net publishes read-noise charts that show invariance behavior camera by camera and is the standard reference.
The practical implication is that on an invariant body, photographers can shoot at a lower ISO than their meter recommends and recover exposure in post, gaining headroom for highlight recovery without paying a noise penalty in the shadows. For a concert with bright stage lights and a dark audience, this means setting exposure for the stage and accepting underexposed crowd shadows, knowing they can be lifted later with the same quality as if ISO had been raised at capture.
Below the invariance threshold, raising ISO in camera still produces cleaner files because the gain happens before the noisy read-out stage. This is why even an invariant camera’s base ISO is usually optimal for clean files when exposure can be controlled by shutter and aperture alone, and high ISO is preferable when those have already been pushed to their limits. Cameras with dual-gain or dual-conversion sensors have a second invariance threshold (often around ISO 640 to 800) where the gain switches and noise drops again.
The concept does not apply to JPEG workflows, since the tone curve has already been baked at the moment of capture. It also does not extend to expanded ISO settings (ISO 50, ISO 102400, and similar pull or push values), which are simulated through digital manipulation rather than analog gain and typically lose dynamic range in both directions. The invariant region is the native range of the sensor, where the underlying signal is unmodified beyond gain.
Common misconceptions include treating invariance as a license to underexpose recklessly. Even on a perfectly invariant camera, exposing closer to the right of the histogram captures more total signal and yields cleaner output. Invariance only guarantees that lifting shadows costs no more than raising ISO would have; it does not eliminate the benefit of more light. The right mental model is that ISO invariance gives the photographer freedom to choose where to make the exposure decision, not freedom to ignore it.