Fun stuff with ETTR and ISOs!

Hey gang!

We've talked a lot over time about ETTR (Expose to the Right) and the place ISO plays, specifically with Canon lenses.

For example, we've seen that if you are shooting Raw and want to ETTR, you can get a bit more "mileage" out of shooting at the "native" ISO that gives you the best "To The Right" exposure, and then pull things in post processing, rather than shooting at the intermediate +2/3 ISO that pulls the native ISO lower in post -- it results in less "noticeable noise" out of the camera, but you lose some control and the lower noise is not "innate" -- you can get the same effect by lowering the higher ISO levels in post.

Also, with the lower intermediate ISO, you run into a problem, in that the signal is decreased to achieve the proper "exposure". This means that shadows are lowered, and this can lead to early shadow "clipping", and therefore a decrease of the Dynamic Range of the signal.

This is all "known" stuff to us ETTR "geeks", but some of us were playing with things in another thread recently and stumbled on a surprise!

You see, one could make a logical assumption that when shooting at ISO 50, the ISO 100 signal is "linearly" lowered by a stop. So, it would follow that if an ISO 50 shot was exposed "to the right", just short of clipping highlights, that an ISO 100 shot with the same aperture and shutter speed would have all the levels boosted by a stop, leading to the highlights clipping, although the shadows and midtones would be brighter. Am I right, or am I missing something?

Anyway, we stumbled on the fact that really, the ISO 50 exposure is not in fact linearly lowered by one stop. In fact, highlights are only lowered by a very small amount, midtones, maybe by 1/2 a stop, and then the shadows by a full stop.

And, as it turns out, if you avoid clipping of the ISO 50 shot and just switch to ISO 100, then your highlights, which may show "some" clipping, are very recoverable! And, you have better shadows that, if the ISO 50 shot has the shadows boosted, well the ISO 50 shots in my tests show more noise in the shadows!

OK, I've given things "in a nutshell". If you are interested, check out this thread:

https://photography-on-the.net …/showthread.php​?t=1152256

You will note that it starts on a slightly different topic, but it does lay out a good background, so give it a read!

Very interesting....thanks for sharing

Tony, et al.,
Since it was released I have been using the Raw Digger application to examine 5D2 files. More about that below.

After reading your post above I decided to take a look at ISO 50 which until now I have ignored (I don't even shoot ISO 100, no lower than 200.) The results I found are much the same as Iliah Borg got from 5D2 RAW files he downloaded from Imaging Resources http://www.rawdigger.c​om …pacity-and-amplifier-gain , but since he used only a small evenly toned selection from the test chart, I think it might be of interest to see full range histograms.
This is my test photo which shows a fairly wide range of subject tones:

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

So Eli, I did install RawDigger, but am not yet up to speed with it -- are you saying then that you see the Raw data the same between ISO 100 and ISO 50? The histograms you show do look a bit different, but I just don't know how to read them at this time.

But, since I was working with DPP and Lightroom, yes, I saw a "curve" applied, and the White Point, the brightest highlights, were pretty well preserved at ISO 100 -- at least they were very "recoverable". Nothing like a stop brighter.

Well, do you agree in practice that you have more Raw latitude with the ISO 100 shot?

For an average shooter, how important is it to actually know this information?

For a pro shooter, how important is it to actually know this information?

It seems to me that there are specialists in the software/technical side of this industry who are constantly coming up with additional programs and technical data that are seemingly suffocating.

I can see this type of data being used to improve software/equipment.

Sometimes I think it is easier to be ignorant and shoot to style/looks than to be so technical. I know there are a bunch of technical people and I do not disparage them in this post. If not for technical people we would still be in the dark ages.

Well, for many people it's "no big deal". Shooting at ISO 50 and other intermediate ISOs can be a convenience. The question is whether for those looking to maximize IQ and especially those shooting "To The Right" and shooting Raw, can it be shown that sticking with the higher native ISO (100, 200, 400, etc) is demonstrably superior?

My tests say "Yes" because with the higher native ISO you not only get better shadow exposure but, because the highlights are not actually amplified by a full stop you get better latitude with the highlights.

What Elie has presented goes farther -- he apparently concludes that an ISO 50 signal is still ISO 100 in the Raw data, if I'm reading him correctly. If that's the case, that would explain why I see good latitude in the highlights. However, I do see more noise in the shadows if I try to boost the shadow noise of the ISO 50 shot. So, not sure what all the facts are! But at least in DPP and Lightroom, the effect of the ISO 50 setting is to apply a "curve" so that shadows are pulled back the most, then midtones, and then the bright highlights are the least affected.

Important? No. Interesting? For me, definitely. I admit to being an addicted measure-bator. I am endlessly curious about how digital imaging in general is done and about how my cameras in particular work. Does it have any practical value? Probably very little, but I still want to know. But then, I've never been very practical; when my parents wanted me to go to college to learn a highly paid profession, I studied philosophy and became a farmer in a commune.

Tony, look at the second pair of histograms. (I should have taken the time to put the camera on a tripod. I can't be sure the differences in the first pair aren't because of differences in framing, but for the second pair I selected the same area of medium tone.) The ISO 50 shot is about 100 units darker than the ISO 100, which in 14 bits represents about 1/10 of a stop. Hardly significant, certainly not the 1/2 stop difference DPP shows. The test wasn't scientific, just shooting out the window and there could have been variations in the light. Still, they are close enough for me to think that the curves are in DPP and LR, who react to the ISO 50 tag. I had thought previously that perhaps the RAW was doctored in-camera, but with RAW Digger's data that does not seem to be the case.
I have Rawnalyze on my old computer. It won't support the 5D2 but it will work with my 40D. I'll resurrect it and compare the two apps.

I do appreciate the effort you guys went through, and the knowledge you have. In hashing it out between yourselves I did actually pick up some good information. I am just glad I am not going to have to figure any of this stuff out without really smart people making it mostly understandable for myself.

Well, to me it just boils down to this: ISO 50 can be convenient, and if you're shooting jpeg use it if the scene calls for it, but with Raw there is not a real gain but your Raw processor can kill the shadows...

Found a version of Rawnalyze that supports the 5D2,
http://www.netzwelt.de​/download/11699-rawnalyze.html
so here are the screen shots. Pretty much the same, the ISO 50 shot is around 0.17 of a stop darker.

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.