ISO invariance. What is it? Since I’m just barely beginning to scratch the surface of the whole concept of ISO invariance, I don’t feel qualified to tell go into details as to what it is. To begin to get an understanding I recommend going here. In a nutshell, ISO invariance is the concept that the quality of today’s camera sensors are beginning to approach the point in which ISO is not as important as it once was. Basically, in many cameras, you can keep it at the base ISO, under expose, and then in post bump up the exposure or shadows as needed and you won’t take a hit with noise any more than had you used the “proper” ISO to begin with.
OK, so then why does this make ISO invariance something cool? There are a couple of reasons that come to my mind: Preserving highlights and selectively brightening photos in post to effectively increase dynamic range far beyond what your camera’s sensor can do on its own while still maintaining relative control over the noise that’s introduced which pushing shadows and/or exposure. Again, I’m far, far from being an expert in ISO invariance. Hell, at this point I’ve barely got a rudimentary understanding. I’d recommend checking out the link above for sure.
What I can do is show you an example of what I’m talking about. Keep in mind that my current camera, a Nikon D5200 is only somewhat ISO invariant. If you’d like to find out how ISO invariant your camera is, this is a good place to look at. Just go there and select your camera from the list on the right. The more level the line the more ISO invariant the camera is. For example, the Nikon D5200 is fairly ISO invariant; closely on par with the D610, not as much as the D7200, and it completely smokes the Canon EOS 5D Mark III. But the Mark III smokes it in native ISO performance for sure. But that’s another topic.
Anyway, in this photo, it was taken substantially underexposed. -2EV to be precise. I’m not sure why I did that, but I did. It makes a good photo to test ISO invariance, though:
You can see that this photo is quite underexposed. For the sky it’s not so bad because those clouds are awesome and it’s nice to keep the details without blowing out the highlights. But, if you brought up the exposure to get the sky just right, you’re still going to be pretty off in the rest of the photo; especially the shadows. So, what I did with this in Lightroom was brought up the exposure by 1 stop. That was perfect for the sky and clouds. But the rest of the pic. To address that I used the Mask tool in Lightroom to selectively bump up the exposure of the foreground by another stop; painting it in. After that, I pushed the shadows a bit–actually quite a bit. +71. The highlights I pulled -70. I then tweaked the blacks and whites to just below the clipping level.
I then moved it over to Photoshop and did some color grading with curve adjustment layers, did some noise suppression with Dfine 2 and added some contrast and detail with Topaz Clarity because, well, clouds. All of this a pretty light though. There wasn’t much noise to contend with. I also brought down the global saturation a bit as well.
This is what I came up with:
There was a ton of detail hidden in those shadows that was completely recoverable. Granted, I’m dealing with the D5200. It ain’t no Sony ar7ii or Nikon D750, but for this little crop sensor entry level camera, it’s not bad at all.
One thing that I’ll point out, though, is that even with as little noise as there is here, I think that there would be even less had I exposed ETTR, peaking the highlights and bringing it down. Sort of an opposite approach. But I suspect that there is a real risk of clipping some highlights in the clouds to the point of not being recoverable. Whereas this approach has no risk of clipped highlights. The slight noise is an acceptable trade off in my opinion.