Originally posted by GUB Iso invariance is not a concept - it is a description of the performance of certain sensors.
Utilising Iso invariance is a concept - a subtle change in technique. Many Pentaxians are instinctly using it anyway.
It does not improve noise -it merely improves your control over the dynamic range.
The horse on the beach can serve as an example.
I presume it is at Iso 100. Lets say it would be good to have some more detail in the dark areas.
So we use the curves tool in pp to bring up the shadows two stops and hold the highlights back where they are now.
Quote: In a non Iso invariant camera the shadow area would be degraded.
In an Iso invariant camera the shadow area would effectively be at Iso 400 with the usual noise associated with 400 and the highlight area will have the tonal range and noise of the original 100 Iso
It is simple as that.
Quote: It does not improve noise -it merely improves your control over the dynamic range.
Example? I mean I know you say it does but, you present no evidence. Iignore ISO altogether. The base ISO of the sensor doesn't ever change, it's physical characteristic.bAll that matters to the sensor is how long the exposure is. What happens after that is all about boosting the signal in camera, or if you choose, boosting it in post processing software. Shooting with higher ISO shortens the exposure time. This invariably degrades the image, increases noise etc. Less signal increase the signal to noise ratio and the noise is increased as well as the signal.
You say it increases your control over dynamic range. It could. However there are great number of images where that's not true. The majority. Well over 50% of my images the camera's ability to capture the dynamic range at 200, 400 or even 3200 ISO is not exceeded. There is not improvement in dynamic range to be had. That makes this point irrelevant. In fact that's just part of knowing how to use your camera. There are a lot of times when a faster shutter speed is more important than the possibility of maximum IQ because if you don't use a fast shutter speed you get nothing. So your choice is clipped dynamic range or no worthwhile image at all. You speak of invariance as if you can just shoot base ISO. In you example of the horse, you shoot 100 ISO but your shutter speed is slow enough the horse moves slightly and you get a blurred image. You have nothing. The guy shooting at 3200 ISO has an image. Maybe he has restricted DR, maybe not.
What is ridiculous is that there is invariance. There isn't. Even the author of the quoted article says that. It's a concept which as far as I can tell in only applicable to images shot in less than Ideal lighting conditions. The only possible use would be in selecting a camera that is relatively ISO invariant, but no one tells you which cameos those are so even that is nonsense.
Shot at 1600 ISO
The highlights are blown and the shadows are black where I want them to be. There is virtually no noise. What would be the advantage of shooting this at 100 ISO? I'd have to set up a tripod. The dynamic range was very limited, low light, no dark shadows, probably an EV of 5 or less in the original scene.
You whole argument seems to be that you should shoot at base ISO and then increase your exposure values in post.
That depends entirely on the post processing editor being better than the in camera software.
So to make the case for invariance being a thing with considering, I'd have to see some examples of images I owl actually use, Whaat's posted in the thread displayed above is crap. I wouldn't use any of them, base ISO or 1600 ISO or 3200 ISO. If the noise is that bad, they all suck. As for dynamic range, the opinion seems to be you can't rescue highlights or shadows at higher ISOs.
Two comments there, one, the whole point of using a higher ISO is so you can properly expose your shadows and have an image come off the camera ready to use. So you are intentionally discarding some of the shadow detail if present, and usually it isn't even present.
My argument against the concept of invariance beyond that it doesn't actually exist, is that it doesn't really add anything to standard shooting practice. Perhaps only of interest to the inexperienced who don't have a lot of understanding of how cameras work.
My advice would be forget it. learn to ue your camera to best effect for what you are doing. If that means shooting everything at base ISO go ahead. Everyone should. However Everyone should also understand that given a base ISO of 100 you have to increase you shutter speed for many images. So, you can just shoot at base ISO and create the image you want in post, or you can shoot at a higher ISO and let the camera do some of the worth. Based on the images posted, I'd rather have the feed back of my histogram.
I guess you could take the image at a high ISO, say 2 stops, note the exposure, set the same exposure manually a base ISO then work in post.
Based on the images provided, or any I've seen, that would be a total waste of my time. I'm going to have to see a whole image that is better because this technique is employed, and it's going to have to be one where one image is useable, and the other isn't. Comparing different levels of bad, which is what I've seen to date is useless. It's not a thing until it can be used to produce significantly better phots. You simply can't use pixel peeping to do that. As happened in McLosky's examples. He focussed on the arts where the "invariant" section was better. But even in his own examples, parts of the "invariant photo" were demonstrably worse. With all the other variables, the difficulty of keeping the focus on the same point, minor variations in light, etc. everything I've seen falls into the category of "inconclusive to the point of being irrelevant to actual shooting." Even in the best examples, they would probably have been un-noticable in a final print.
It's theory with no practical application, and it's an extrapolation of properties that don't theoretically exist.
In the world of signal to noise, invariance is noise.
There is nothing in the whole concept of invariance that isn't explained by the superior DR of the cameras considered to be "invariant". If you have more DR, you have more to play with. When I set my camera to 16 ISO, I know all I'm doing is cutting off the lower 4 stops, but 11 stops of EV is still more than I need for many images, so I'm cool with that. Saying you can do better in post, depends completely on having a way to calculate your exposure. I've never seen a non pixel peeping image where me with a 2 second change in ISO isn't better than some guy messing around for minutes to hours calculating exposures and adjusting in post processing.
Unless you plan to sell your images for millions, it's not worth it.