The Optimum Digital Exposure

[BobD]

I think I found a minor mistake in the article.  Under method #2: "The exposure that shows no red clipping in highlight detail areas is the optimum exposure of the set. In this case EV+1.3."  No, I think that should be the EV+.6 exposure.  It goes on to say that either (+1.3 or +.6) would work in this case.  But clearly the +.6 has the least clipping.

Duane, I'm not sure it is a mistake... just choices. I thought this might be confusing but I wanted to bring up the concept that "Highlight Clipping" is acceptable when there is specular reflection or bright sunlit snow (a.k.a. Zone "X").  Notice the "Note" under the example:
"Note: There can be some red highlight warning in an optimum exposure if the photo includes sunlit bright snow, specular reflections, etc.. In the example above you might choose the middle +0.6 EV exposure, however, the red “Highlight Clipping” on the roof in the first exposure, +1.3 EV, could also be used. The bright snow covered roofs in bright sunlight would look natural being 100% white with no texture. Remember the expression of the Artist’s Vision is what matters."

Although "you might choose the middle +0.6 EV exposure"... "the first exposure, +1.3 EV, could also be used."  When ever I have choices like this example I usually choose the brighter exposure. Hope that helps.

[BobD]

I think a lot of people are getting lost in the trees here.  Seems to me the premise is pretty simple.
...8. As they say, the proof is in the results. Jpegman

Jpegman,
Thanks for your clarification and perspective.  I am an empirical kind of guy, so when I read an article (like the one I wrote here)... I try to think it through but I always need to try it.

As you say the proof is in the doing.  I am not looking for people to adopt my method buy adapt the concept with their photography.

I can only hope that those that try it have the same quality results that I have experienced.

Thanks Again

[deejjjaaaa]

but this benefit maxes out at ISO 1600 or thereabouts (see Emil's Fig 12a for the Canon 1D3).

Emil's tested cameras long time ago... nowadays Aptina's approach ( http://www.aptina.com/products/technology/DR-Pix_WhitePaper.pdf ) in some cameras can kick in @ higher "ISO" (= Sony A7s for example)... so ISO1600 or thereabouts is not set in stone, is it ?

[trichardlin]

...

Why don't our cameras, especially any over $500, have an ETTR exposure mode built in? Surely it is just a software addition to the metering reading?
...

Imagine a family on vacation.  With hundreds or even thousands of pictures taken in this 'optimal' mode.  The poor dad has to individually adjust all of them in post to get usable pictures to show friends and families.  It's just too much work to get arguably marginal improvement in picture quality (depends on the scene and what's important in the picture).  In addition, you could suddenly lose a couple of stops, resulting in more blurry photos due to longer shutter speed.  So, if you are not careful, ETTR can give you more bad photos than not using ETTR.

[Guillermo Luijk]

Imagine a family on vacation.  With hundreds or even thousands of pictures taken in this 'optimal' mode.  The poor dad has to individually adjust all of them in post to get usable pictures to show friends and families.  It's just too much work to get arguably marginal improvement in picture quality (depends on the scene and what's important in the picture).  In addition, you could suddenly lose a couple of stops, resulting in more blurry photos due to longer shutter speed.  So, if you are not careful, ETTR can give you more bad photos than not using ETTR.

Having the AETTR feature doesn't mean you have to use it all times, as simple as that. That family could spend their vacation happily producing tons of nice JPEGs without worrying about ETTR or even RAW.

[ErikKaffehr]

Hi,

Just an example:

This is an image that in all probability was exposed using both blinkies and histogram on my P45+. Histogram in Lightroom indicates significant underexposure:



Below is the same image in RawDigger, note that green channel is very close to clipping. Any additional exposure of original raw image would cause reconstruction of skym as green channel would be clipped.This is what the sensor sees:


Developed image below. More exposure would be beneficial for shadows, but may have reconstruction artefacts in sky.


Best regards
Erik

[laughingbear]

Greetings,

if there was one single article that nailed it for me, then it was this one:

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#ETTR

in conjunction with the Niagara falls example from Jeff Schewe here:

http://schewephoto.com/ETTR/

Best
Georg

[Jonathan Cross]

What a fascinating article and thread.  I have been conscious of ETTR that Michael espoused and has rightly promoted for a long time and I try to put it into practice most of the time.  I am also vastly impressed by the efforts contributors to this thread have made to get the very best quality images.  There are times, however,  e.g. photographing wild life or 'grabbing the moment' when ETTR is difficult to employ.  I am also aware, rightly or wrongly, of images, particularly b&w, where there are areas of deep black that add to the visual impact.

All this is a preamble to my view that, 'It all depends.'.  Yes, of course ETTR is a good idea as it gives flexibility in post processing, BUT...  To me part of the joy of photography is to try to get it right in camera and not have to do huge amounts of post processing.  Getting it right in camera, involves having an idea of how I want the final image to look.  If I am visualizing a b&w image or a graphic one with use of black as an important component, then my exposure considerations will be different to when trying to get an image of a scene with a wide range of subtle tonal and colour variation.  To me ETTR is an important consideration, but not the only one.

Thanks to all LULA contributors for making me think!

Jonathan

[BobD]

There are times, however,  e.g. photographing wild life or 'grabbing the moment' when ETTR is difficult to employ.

To me part of the joy of photography is to try to get it right in camera and not have to do huge amounts of post processing.

"There are times, however,  e.g. photographing wild life or 'grabbing the moment' when ETTR is difficult to employ."
This is the beauty the OneZone method #2. You set your exposure for EV +1-1/3 then bracket +/-2/3 of the stop!. If you put these settings in one of your camera presets, when the time comes for "grabbing the moment" you just simply hold your finger on the shutter button (capturing 3 exposures) then go on to the next scene.

I also understand "...the joy of photography is to try to get it right in camera".
However, the optimum exposure IS getting it right in the camera (placing the brightest part of the scene at 99+% brightness in your software). Think of the "huge amount of post processing" being a commitment to your craft.

I know many photographers think getting the "final" exposure in the camera it is a badge of honor... that badge can get heavy.

A more correct statement may be ...to (you) part of the joy of photography is to get "the final exposure" in the camera. But you do realize that this is not the optimum and the results  are that "digital signature" that we have unfortunately come to accept.

Thanks for your thoughts.
Bob

[BobD]

Imagine a family on vacation.  With hundreds or even thousands of pictures taken in this 'optimal' mode.  The poor dad has to individually adjust all of them in post to get usable pictures to show friends and families.  It's just too much work to get arguably marginal improvement in picture quality (depends on the scene and what's important in the picture).  In addition, you could suddenly lose a couple of stops, resulting in more blurry photos due to longer shutter speed.  So, if you are not careful, ETTR can give you more bad photos than not using ETTR.

I agree... this is the time for Jpegs or iPhones.  As I say in my book:
Where the OneZone May Not Be Useful
There are shooting situations where the OneZone may not be practical - such as event photography. If you are “Wi-Fi-ing” hundreds of images to the sports photo editor or shooting hundreds of photos at a wedding ceremony the OneZone method will be unwieldy...
...you can take advantage of having a camera with custom presets. Simply move off the  Custom setting that has the OneZone settings, shoot and upload the  JPGs as normal, then return to the custom setting containing the OneZone settings.

If you place you OZ Method #2 in one of your camera presets... then when "poor dad"comes across a wonderful scenic, can just go into the custom setting mode take the 3 OneZone brackets set then quickly go back to shooting JPEG.

Thanks for your input
Bob

[Ray]

And I absolutely disagree. ETTR is about shifting the histogram to the right (maximise RAW levels before clipping). To achieve that on a digital camera there are three variables: aperture, shutter AND ISO, the last one has no connection to captured photons. As simple as that

I have to agree with Guillermo's logic here. The expression ETTR clearly refers to exposures which shift the histogram to the right without clipping wanted highlights. What else could it mean other than 'Expose To The Right' of the histogram? It certainly doesn't mean 'expose to the right of my left ear lobe', or 'expose to the right of my little toe'. 

Although it's true that an ETTR at base ISO will result in the maximum number of photons captured without clipping relevant parts of the scene, the fact is, when using a higher-than-base ISO, the same principles of ETTR apply, but apply with a significant reduction in the number of captured photons, compared with an ETTR at base ISO.

[deejjjaaaa]

Imagine a family on vacation.  With hundreds or even thousands of pictures taken in this 'optimal' mode.  The poor dad has to individually adjust all of them in post to get usable pictures to show friends and families.

the poor dad uses OOC JPG

[deejjjaaaa]

Having the AETTR feature doesn't mean you have to use it all times, as simple as that. That family could spend their vacation happily producing tons of nice JPEGs without worrying about ETTR or even RAW.

and not only that... if a manufacturer now suggests poor dad some exposure that prev. poster thinks is good for a raw conversion unaltered, then that manufacturer can simply write a tag in raw file hinting to raw converters how to dial it back, during raw conversion, exposure-wise, from ETTR exposure back to the one camera is using today... then poor dad just need a raw converter support to use that tag automatically - so it is a solvable problem @ code level (but certainly as usual w/ all issues related to automated batch raw conversion w/o manual attention to each shot).

[Bart_van_der_Wolf]

I have to agree with Guillermo's logic here. The expression ETTR clearly refers to exposures which shift the histogram to the right without clipping wanted highlights. What else could it mean other than 'Expose To The Right' of the histogram? It certainly doesn't mean 'expose to the right of my left ear lobe', or 'expose to the right of my little toe'. 

Although it's true that an ETTR at base ISO will result in the maximum number of photons captured without clipping relevant parts of the scene, the fact is, when using a higher-than-base ISO, the same principles of ETTR apply, but apply with a significant reduction in the number of captured photons, compared with an ETTR at base ISO.

It depends on what one's Expose to the right goal is. If the goal is a right aligned histogram, then there are several routes possible, even postprocessing. If one's goal is to optimize overall image quality/dynamic range, then increasing  the number of Photons is the path.

Boosting ISO / gain has some merits, but will not produce the same high level of Dynamic range. If the subject contrast is moderate, it might be a good alternative, especially on certain Canon camera models. In some fields of photography, where relatively short exposure time is important, it offers another trade-off to choose from. But that's IMHO not about ETTR, it's rather about optimizing noise characteristics of mostly shadows (and the more limited DR dictates shorter exposures to avoid clipping).

Cheers,
Bart

[deejjjaaaa]

But that's IMHO not about ETTR, it's rather about optimizing noise characteristics of mostly shadows

from a practical standpoint optimizing exposure (first of all) and - if possible - gain (after exposure consideration) shall be always considered together ...

PS: probably some different abbr. shall be used - like Ettr&GO ( exposure ttr & gain optimization ) or whatever

[williamchutton]

When the single goal is shifting the histogram to the right, it is possible to do more harm than good. Simply increasing ISO alone will shift the histogram. But the analog exposure of the sensor when the shutter is open is not increased. The SNR and DR range may not be optimal compared to using a different shutter time and, or aperture with a lower ISO parameter.

This is how come LuLA member ejmartin's advice from 2011 to maximize exposure (the analog signal recorded by the sensor when the shutter is open) is less ambiguous than the ETTR acronym. Member ejmartin suggested the new "mantra" should be ME. ME often produces a histogram that coincidentally resembles an ETTR histogram.

[bjanes]

1. If you are 2 stops underexposed (90%) you are using only 25% of your sensor data and effectively have thrown away 75% of your data; If you want to stop at the 97% level and don't want to go to the 99% level because it too risky you have discarded 50% of your sensor data - your choice! George Jardine documented this in his article referenced in the DiNatale's LuLa post (http://mulita.com/blog/?p=3358#Tonal-Compression) -it's real no matter what camera or sensor or raw converter one chooses - it's the nature of current digital capture technology!

This business about the number of possible tonal levels in the brightest f/stop or the brightest 2 f/stops is what Emil debunked in the referenced article. The important thing here is the signal to noise ratio (SNR), which varies with the square root of the number of photons collected. Two stops underexposed means the SNR is cut in half. The number of possible perceived tones will be limited by noise, not the reduced number of encodable tones using only 25% of the range of the ADC.

Bill

[MarkL]

ErikKaffehr's exaple is interesting. I know I have had shots where the histogram in LR shows no clipping but colour information has clearly been lost.

A similar example is here: http://diglloyd.com/blog/2014/20140812_2112-NikonD810-MiningShack.html

[trichardlin]

Having the AETTR feature doesn't mean you have to use it all times, as simple as that. That family could spend their vacation happily producing tons of nice JPEGs without worrying about ETTR or even RAW.

Of course.  Remember we were just guessing why camera manufactures don't build this function into the camera.  If only 1% of the skilled customers need this function, it probably won't make it into the final product, especially if the function is tricky to implement or confusing.

[jpegman]

This business about the number of possible tonal levels in the brightest f/stop or the brightest 2 f/stops is what Emil debunked in the referenced article. The important thing here is the signal to noise ratio (SNR), which varies with the square root of the number of photons collected. Two stops underexposed means the SNR is cut in half. The number of possible perceived tones will be limited by noise, not the reduced number of encodable tones using only 25% of the range of the ADC.

Bill

Specific numbers aside -(I spent quite some time going over Emil's paper and was more confused when I was finished about what he was (or was not) trying to say. Glad I didn't have him as an prof at Tech!

However, academics aside, when you collect as many photons via ETTR as you can on the sensor, your data is increased - common sense.
The impact is easy to understand - More input data = More RAW converter data for image processing. The result I SEE is less shadow noise!

Empirical testing shows what Bob demonstrated in his "optimum ISO Comparison" figure where the same sensor/subject/light when "Optimally" underexposed by 2 stops has better s/n than the low ISO (200) image and much better than the "2 stop non-ETTR" image.
What is there to argue about specific number like 25%, 50% or just better images if you want them. I rather be in the company of Bruce Frasier or George Jardine with their imperfect math rather than Emil's analytical corner.

The important thing her is NOT what actually drives the SNR, but, the proposed process will improve your images visually - which is how I evaluate my images. My simplistic conclusion is - if it makes my images better, than I use it until a better paradigm comes along to replace it!

Jpegman