The thing that no one seems to be asking in the whole "ISO is fake" discussion is why you'd really care.
I had one reader write to me with a brilliant line: "what problem are photographers trying to solve by focusing on ISO invariance?"
That's a really good question.
If you're a JPEG shooter, there's no way you want to shoot at base ISO and boost in editing. That's because you've thrown away at least 80% of the data and then stuffed it into an eight bit container. If you try to "resurrect" that image with post processing, you're asking for pain. Let's hope it isn't an image you really needed.
If you're a raw shooter, at least you're not (usually) throwing away any data, plus you're keeping it in a larger (12 or 14 bit) container from which you can probably pull data upwards without distorting it much. But why?
In essence, you're replacing ETTR with ETTL. (That's Expose To The Right and Left). And you're believing—as I wrote last time—that the data accuracy is better in your raw converter with processing than in your camera's electronics using ISO changes.
What is it we really want?
- JPEGs that are as close to "perfect" coming out of the camera. Now, "perfect" is in quotes because your definition and mine probably don't match. So maybe a better thing to ask for here is a JPEG that needs no post processing, because you were able to get it the way you want with camera settings. Which includes ISO ;~).
- RAW files that are optimal data captures of what the sensor saw. This one gets tricky for all kinds of reasons. Nikon, for instance, does color temperature pre-balancing of the color channel data before saving it to a file. I wish they didn't do that, but they do. As I noted before, the actual DNs (digital numbers) that get written to the file are accurate only in so far as the camera did its math correctly. We have issues with some cameras about that, too.
- Both at once! Yep, some of us still shoot for clients who want a JPEG right away—it doesn't have to be full size, either, as it probably is being squirted to the mothership via wireless communication—but may request the raw file later on. The JPEG should be perfect, and the raw data should be optimal!
What's missing in camera are actual tools for raw users. Tools that we have to use after the fact with third-party software products to investigate what's happening in our files (e.g. RawDigger).
First up, we have the missing raw histogram. Sometimes I wonder if the camera makers don't provide this because (a) they're lazy; (b) they haven't heard anything customers have been saying (I've been asking for this for nearly 20 years); or (c) they don't want people to see what actually is in the raw data.
Don't laugh at the last one: raw files often contain what seem to be inexplicable gaps or consolidation peaks of data in one or more channels. Canon lately has been doing some form of noise reduction at low ISO values. If customers could see that, some poor marketing guy in Tokyo would have to go to the engineering department and get them to explain that in a way they could understand, then they'd have to come up with a technical note to send to the subsidiaries and make sure it gets translated correctly, and then some lower level marketing guy would have to post something in each subsidiaries knowledge base.
But I really want more than just a raw histogram. Just stick RawDigger Lite (doesn't exist, but hey, why not?) into the camera so I can see much more than just a correct histogram. I want to check saturation levels, black point, see clear saturation and went-to-black markers, maybe even see where the bit division falls.
Worse still, I want to set my exposure "correctly" for both that JPEG and for optimal raw capture, and those may be different!
