ETTR & post processing

[Eric Brody]

I think I understand ETTR, basically push the histogram as far to the right as possible without blowing the highlights. I've read many of the LULA discussions about it going back to Michael's early article and the commentary. So... when I import my images into Lightroom CC, they look "over exposed," e.g. the histogram tends to peak on the right. After setting the white balance, I then go through LR's sliders, one by one, starting with exposure, contrast, etc, making the image "look right" and also fitting within the histogram, taking into account my "artistic" plans for the image.

Michael's advice, way back on April 13, 2008 was "Just "normalize" the exposure so that it looks correct using the Exposure slider. What you'll have done though is to move the data-rich portion of what has been recorded down into the data poor mid-tone and quarter tone area, thus giving yourself a more robust image to work with."

When I compare the finished ETTR image with a less exposed bracket of the same image, there seems to be minimal difference after all the manipulation. Am I really getting more data into the ETTR image? I do understand that most of the information is in the highlights and that the noise resides in the shadows but I've been struggling with this for a while and just decided it's time to ask those likely more knowledgeable than me. Do the ETTR "experts" use a different workflow? Am I missing something or is it really that simple?

Thanks for any comments or suggestions.

[AlterEgo]

when I import my images into Lightroom CC, they look "over exposed,"

if you want to understand what is the situation with your raw file you need to use products like rawdigger ( www.rawdigger.com ) or FRV ( http://www.fastrawviewer.com )

if you are using ACR/LR then

1) depending on the "process" version their exposure correction in UI might not be linear
2) ACR/LR might be using hidden (when your exposure slider is zeroed in UI it does not meant that theres is actually no correction behind the scenes) exposure correction for certain camera models/certain nominal ISO values hardcoded in their code
3) Adobe's own DCP camera profiles might have further hidden exposure correction inside for some camera models that can either compensate, somewhat compensate or further increase the exposure correction done in the code
4) Adobe's own DCP profiles can have brightness correction with tone curve and/or LUT done

so do not use Adobe's own raw converters to understand what is in your raw __unless__ you select the proper process version (like 2010 for example) and use a specifically tuned dcp profile that is designed to compensate any corrections done in code for a given camera model and nominal ISO and also otherwise make no brightness corrections (linear tone curve prescribed) and possibly avoid WB and clamping to a smaller gamut during color transform

so basically don't use Adobe to evaluate raw files, it is much easier to use the tools designed for this purpose

[AlterEgo]

Am I really getting more data into the ETTR image?

ETTR is about signal to noise ratio with the intelligent decision about what you can allow yourself to clip (or even put into potentially non linear areas - for some camera models) in raw (that also includes the decision about "ISO" too - which is your decision about post exposure /as "ISO" is not part of exposure itself/ signal/data processing that you actually do before the exposure), that's it... skip the crap - read Emil Martinec article (but if you read the discussion here, @ LuLa forum, after the first LuLa article then you shall know) = for example https://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html

PS: the original article has to be removed from LuLa website and replaced with permission with Emil's one - long overdue

[AlterEgo]

that the noise resides in the shadows

you simply have worse signal to noise ratio in shadows (more so in deep shadows where relative contribution of so called "read" /actually there are many components there of different nature related to the operation of the chip/electronics itself, but for simplicity you can aggregate them by one word/ is high vs shot noise contribution when signal is so low/small)

[AlterEgo]

When I compare the finished ETTR image with a less exposed bracket of the same image, there seems to be minimal difference after all the manipulation.

depending on the nature of the scene (if what you pay attention to lies above deep shadows in a regular shot anyways the difference might not be as "visible" as if you pay attention to the details in deep shadows which might be moved from that area in ETTR'd shot), noise reduction applied in raw converter, magnification (are you pixel peeping), difference in exposure (1/3 EV difference for example is not as noticeable as 2 EV difference), illumination (consider tungsten light, lack of blues), particular camera model (some modern sensors have good performance in deep shadows in terms of the noise) and your perception you may or may not notice a difference or call it "small", etc, etc... but 2x2 = 4 still

[Guillermo Luijk]

ETTR is about improving noise. If noise is not a problem in your application you are wasting your time with ETTR.

Older articles about ETTR (like Michael's) focused too much on the "number of levels" subject, when this is pretty irrelevant since no RAW file lacks in number of levels for its SNR, no matter how you expose. RAW posterization is a myth.

[mouse]

Guillermo has put it all in a nutshell and I can only echo what he said.

ETTR is not a magic formula for improving every photo.  I can improve definition and reduce noise in the shadows when such problems are anticipated.  Otherwise it's a waste of effort.

[ErikKaffehr]

Hi,

The major benefit of ETTR is that signal noise ratio (SNR)  is maximised. Now, SNR is seldom problematic in the highlights, but can be problematic in the shadows. So, shooting ETTR maximises signal noise ratio over the whole luminance range, but it is most noticeable in the shadows.

Forget about levels and that kind of stuff, it is just irrelevant information. Signal noise ratio is the issue. You want as much useful information (signal) over meaningsless information (noise) as possible, and the way to achieve that is maximising exposure.

Best regards
Erik

I think I understand ETTR, basically push the histogram as far to the right as possible without blowing the highlights. I've read many of the LULA discussions about it going back to Michael's early article and the commentary. So... when I import my images into Lightroom CC, they look "over exposed," e.g. the histogram tends to peak on the right. After setting the white balance, I then go through LR's sliders, one by one, starting with exposure, contrast, etc, making the image "look right" and also fitting within the histogram, taking into account my "artistic" plans for the image.

Michael's advice, way back on April 13, 2008 was "Just "normalize" the exposure so that it looks correct using the Exposure slider. What you'll have done though is to move the data-rich portion of what has been recorded down into the data poor mid-tone and quarter tone area, thus giving yourself a more robust image to work with."

When I compare the finished ETTR image with a less exposed bracket of the same image, there seems to be minimal difference after all the manipulation. Am I really getting more data into the ETTR image? I do understand that most of the information is in the highlights and that the noise resides in the shadows but I've been struggling with this for a while and just decided it's time to ask those likely more knowledgeable than me. Do the ETTR "experts" use a different workflow? Am I missing something or is it really that simple?

Thanks for any comments or suggestions.

[mouse]

Hi,

Forget about levels and that kind of stuff, it is just irrelevant information. Signal noise ratio is the issue. You want as much useful information (signal) over meaningsless information (noise) as possible, and the way to achieve that is maximising exposure.

Best regards
Erik

The operative word in the last sentence is "useful".  Noise in the shadows becomes a problem when one tries to boost the shadows in post-processing.  If one is content to let the deep shadows go to black then the extra information is not useful.

Please don't misunderstand; I am not one of the camp who rail against ETTR.  It's a great tool when you need it, but you don't need it in every shot.

[Eric Brody]

Many thanks to all who responded. I can now relax. My photographic OCD, trying to squeeze the maximum out of every pixel was taking some of the fun away. As all of us should, I'll concern myself with more with subject and composition and also do my best to make good exposures. I use high quality equipment, D800E and Fuji X T-1 with excellent sensors and dynamic range. I do this for fun, so now it will be even more fun. I'll still work on ETTR, just not quite so compulsively, to the possible detriment of the more important aspects of image making.

[Garnick]

Hello Eric,

I'm surprised that this article on LuLa hasn't been brought to light in this discussion.  "Light" being the operative word.  The article(The Optimum Digital Exposure) can be found at LuLa Home/Articles/Techniques, but here's the URL to get you there immediately - https://luminous-landscape.com/the-optimum-digital-exposure/.  It's actually a sort of continuation of the aforementioned article on ETTR by Michael.  Bob DiNitale has done a lot of experimentation on this subject, and at first glance his findings seem to be somewhat unbelievable and counter intuitive.  However, if you find that you are in any way intrigued by the introduction, do as I did and purchase his eBook.  The initial testing procedure might seem somewhat rigorous, and even Bob admits to using the less complicated method in most situations.  To put it in the proverbial nutshell, if your RAW image doesn't look like it has been dipped in milk, you have underexposed by at least 1 to 1.5 EV.  Bob also states that this method goes beyond the notion that ETTR's "only" advantage is noise reduction.  He states, as I have proven to myself, that when properly exposed and processed, the image exhibits a much more film-like appearance, with smoother gradations and tonal qualities.  I would seriously suggest that you at least give it a read.  I think you might find Bob's approach quite interesting, whether or not you decide to adopt it for your own use. Take a peek.

Gary      

[Eric Brody]

Thanks Gary, I did read Mr DiNatale's article but as best as I can tell from re-reading it briefly, he does not really answer the question of what to DO with this "optimally exposed" image, e.g. the best way to process it. That is really the essence of my question. I'm not sure I am concerned about just why ETTR is a superior method, though I do like to understand what I'm doing and why. I have been convinced that optimum exposure does produce a higher quality file containing more digital information. Initially it seemed to me that just moving the exposure slider in LR to the left, e.g. reducing the exposure of an "overexposed" image would be the same as making an exposure at the reduced amount but it appears that is not the case. Thanks for your comments, they are appreciated.

[mouse]

Thanks Gary, I did read Mr DiNatale's article but as best as I can tell from re-reading it briefly, he does not really answer the question of what to DO with this "optimally exposed" image, e.g. the best way to process it....
Initially it seemed to me that just moving the exposure slider in LR to the left, e.g. reducing the exposure of an "overexposed" image would be the same as making an exposure at the reduced amount but it appears that is not the case. Thanks for your comments, they are appreciated.

Hi Eric-

If, in fact, the exposure slider in your raw processing software operated in a strictly linear fashion, then this would indeed be the case.  The advantage of the "overexposed" image (I dislike the term because it is only overexposed based on the camera meter's reading.) is that, after moving the exposure slider in LR to the left, one can then boost the shadows without accentuating noise.
 
However the exposure slider in Adobe, Process 2012 (ACR and LR) does not operate in a linear fashion.  Thus reducing the exposure with this slider will result in compressing the tonal range of midtones and below while expanding the tonal range above the midtones.  This means that you will have to apply some boost to the shadows if you need to increase detail in that area.  Other raw converter software may differ; I don't know.

The general message of Mr. DiNatali's article (and most other ETTR discussions) is that, if you rely on your camera's meter to set exposure, you have significantly more headroom to increase exposure before you risk blowing out the highlights.  (A story so old it has grown whiskers while we wait for the camera makers to give us on-camera access to the raw data.)  This leads to his concept of an "Optimum Exposure", which will be somewhere between +0.6 and 2.0 EV above the camera meter's recommendation.

If you are really interested in how this behavior relates to your own camera I strongly recommend the article:
http://www.rawdigger.com/howtouse/calibrate-exposure-meter-to-improve-dynamic-range
and carry out the procedure described therein.  It requires you purchase a copy of RawDigger software (very inexpensive).   I have recently carried out this test with my D800; it is easy to do and to understand.

After examining my results of this test and later reading Mr. Dinatale's article, I believe this article contains at least one serious flaw.  "Method #2, Using in camera multizone metering".   He recommends using the raw conversion software's (LR or ACR) highlight clipping indicator to determine the Optimum Exposure.  This would be the highest exposure at which no (or very minor or specular) highlight clipping occurs.  (Choosing from a series of 3 bracketed shots at 0.6 EV intervals.)

The highlight clipping indicator in LR & ACR will be very dependent on which Process is used.  In my examination, using Process 2012, highlight clipping does not becomes apparent until one exceeds +3.7 EV (above camera meter).  With Process 2010 highlight clipping becomes pronounced at +2.7 EV.  Using other raw conversion software one might well expect still different values.  Using the RawDigger analysis I find that (with my D800) there is a headroom of +3.0 EV above the (spotmetered) brightest highlight.  YMMV.

Bottom line, to answer your question: " what to DO with this "optimally exposed" image, e.g. the best way to process it."
It's really simple: process it the same as you would any other image, by eye-ball!.  Begin by backing off the exposure slider until it looks right, and go from there.  Your only advantage is that, if you need to boost the shadows to extract more detail, you can do so without risking increased noise.

[digitaldog]

When I compare the finished ETTR image with a less exposed bracket of the same image, there seems to be minimal difference after all the manipulation. Am I really getting more data into the ETTR image?

ETTR is just about optimal exposure for raw data, you end up with less noise. It's rather simple, it's about exposure for the data you intend to use (raw) and how you decide to develop after exposure. It's been done in photography for over 100 years (optimal exposure and development). What's new is digital, a camera with an exposure system and guide for a JPEG, not a raw. ETTR was a way to educate new photographers about optimal exposure and development (what Michael calls normalization) for raw data. IF the capture is such you can optimally expose for the raw, do it. If you can't, you can't. You'll end up with more noise, mostly observed in shadows.

[kirkt]

The article is an interesting take on ETTR, but really is of little practical value.  If you adopt a metering strategy that is based on your raw converter's default rendering (or user's tweaked rendering) of the data once imported, then your exposure strategy is tied to that software and whatever is going on under the hood.  As mentioned previously, there are tools to inspect the raw data and make more converter-agnostic exposure decisions based on raw data analysis that does not contain secret sauce.  Otherwise, you might shoot with one raw converter in mind and then decide to try others and find that the latitude you had with converter A is not really present with other converters.

Most of the time, you will find that some experimentation and field shooting will make it apparent that your ETTR strategy will be some metering mode + some positive exposure compensation.  It does not need to be just to the brink of clipping, where you are constantly walking some highlight clipping tightrope - just enough to get the signal you want, especially at higher ISOs.  Of course, this is still a balancing act of sorts, as the exposure increase needs to come from somewhere - if you are already shooting relatively high ISO, then you need to open the aperture or let the shutter stay open longer, with the situation and your artistic vision hopefully not compromised by the choice.

With Canon cameras supported by the Magic Lantern project, you can tweak this strategy even further with the raw histogram and raw channel clipping indicators, or the auto-ETTR functionality.  Why camera manufacturers do not include such tools for cameras that are capable of raw file acquisition (or at least some "pro" or "advanced" raw-capable cameras) is a mystery.

kirk

[AlterEgo]

If you adopt a metering strategy that is based on your raw converter's default rendering (or user's tweaked rendering) of the data once imported, then your exposure strategy is tied to that software

but if he is tied to his software anyways then he by all means has to take into account how his software deals with data clipped or near clipping... that is of course unless he never clips, but otherwise in some situation you do make a decision to allow clipping and then it makes sense to understand how your raw converter will invent the data... some do nicely some do not

[kirkt]

but if he is tied to his software anyways then he by all means has to take into account how his software deals with data clipped or near clipping... that is of course unless he never clips, but otherwise in some situation you do make a decision to allow clipping and then it makes sense to understand how your raw converter will invent the data... some do nicely some do not

Right - I'd rather know when my raw data are clipping versus when my raw converter tells me there is clipping.  Then I know when my raw converter is lying to me and making it up.  And it is not just about the exact point of clipping, but what is happening to the stop or so of data below clipping that is getting compressed to make it seem like little clipping is taking place.

kirk

[Tim Lookingbill]

Right - I'd rather know when my raw data are clipping versus when my raw converter tells me there is clipping.  Then I know when my raw converter is lying to me and making it up.  And it is not just about the exact point of clipping, but what is happening to the stop or so of data below clipping that is getting compressed to make it seem like little clipping is taking place.

kirk

I just happen to experience just what you're talking about comparing tonality and clipping with a cloud/sky image I've been discouraged from editing in LR's PV2012 vs ACR6.7 PV2010 which maps highlight definition quite differently and I thought mistakenly much better. I also found that "My Prints Are Too Dark" solution in PV2012.

The one on the left is ACR6.7 PV2010 edit with ProPhotoRGB output histogram showing just clipping where if I increased exposure or any other brightening slider I'ld get pink tinged blowouts (ACR4.4 profile). The one on the right is in LR4's PV2012 with some tweaking to get a match. No clipping in ProPhotoRGB output set in LR's Soft Proof. I can actually make this image brighter and still retain definition but with the use of a tweaked curve. Highlight Recovery slider is maxed out to -100. Clarity is at +76. Note no dark halos. I had to take them out in ACR6.7 with adjustment brush which I had to delete in LR4.

[texshooter]

Can I expose to the right by increasing the ISO, or is it always better to ETTR by pushing the shutter speed or aperture?

[stamper]

Can I expose to the right by increasing the ISO, or is it always better to ETTR by pushing the shutter speed or aperture?

Manual or aperture priority? In aperture priority I use EV. In manual increase the exposure past the centre point.