A practical approach to ETTR

[nathaniel_alpert]

<p >A practical approach to ETTR: Just so you know where I am coming from, my work includes landscape and nature photography, but no BIFs or portraits. I am contributing my observations in the hope of helping those who would like to use this method. Please don’t tell me about special test targets, or your pet theories of exposure. This is supposed to be a discussion of the practical use of ETTR.</p>

<p > </p>

<p >In my applications I am seldom under the pressure of relying SOLEY on the in-camera exposure meter. I pay little attention to whether I am using averaging, evaluative or spot metering. Black cats in coal bins, white on white, it does not matter, for I am relying mainly on the in-camera histogram. When I achieve an exposure with no clipping of any channel in the histogram that is close to the right, I am done.</p>

<p > </p>

<p >For the record, though I doubt the brand of camera is a determining factor, I am shooting in raw mode with either a Canon 5d i or 5d ii. The jpeg settings are turned down (saturation, contrast, etc.) to produce the most neutral, low contrast in-camera jpeg. I adjust exposure till the in-camera histogram is close to clipping. No channel is allowed to clip, unless I am bracketing. As a practical matter, the histogram obtained from the corresponding raw image is usually similar to the in-camera histogram when the image is opened in ACR or LR. I don’t claim they are identical. Sometimes the ACR- or LR-rendered histogram looks like it is clipped but it is (almost) always possible to decrease the exposure slider and recover a histogram that is not clipped. The observant reader will notice the “almost” in parentheses. There is the deviant image with a few specular highlights, the histogram with a long tail on the right, etc. These require judgment, experience, and bracketing when in the field. The blinking over exposure warning can also be helpful, as it shows WHERE the clipping is taking place. However, the clipping warning is computed from the luminance information, so even if there are no warnings you may still have clipping, a proper RGB histogram is required. A special tip for users with “real time histogram (live view)”: This histogram seems more accurate than the one determined from the in-camera jpeg. This is just my observation, what do you think?</p>

<p > </p>

<p >I have never noticed any color shifts or distortions from this approach. The best digital exposure is obtained by exposing so that the raw image histogram is as far to the right as possible (even if it looks clipped), so that (reducing) the exposure slider can bring it down to the point where there is no clipping. The brightness slider is then used to refine the intensity level to suit your photographic intent. With the method as outlined, not every exposure will be as far to the right as possible, but many will. That is what I mean by practical.</p>

<p > </p>

<p >Let me know what you think</p>

[tonybynum]

I think you might have a point or two in there someplace . . . but basically you said expose for the

highlights, thanks for the tips. . .

 

Tony

[digitaldog]

<p>Sounds very reasonable to me. It would all do us good if we had actual feedback of the exposure for raw rather than the JPEG produced in camera from the raw. I do think each sensor will have its own, unique properties, hence more support for providing us with a histogram based on the raw. If in doubt (and if possible) bracket in + EV over what the meter suggests. And I think those few specular highlights you might clip is just fine (assuming you want them to and which will keep the image from looking muddy). </p>

<blockquote>

<p>I have never noticed any color shifts or distortions from this approach. The best digital exposure is obtained by exposing so that the raw image histogram is as far to the right as possible (even if it looks clipped), so that (reducing) the exposure slider can bring it down to the point where there is no clipping.</p>

</blockquote>

<p>Again, I can’t think of anything in the above paragraph I haven’t observed and it seems spot on to me. </p>

[ben_goren]

<p>Nathaniel,</p>

 

<p>Earlier this morning, Josh closed down two threads devoted to ETTR, and was pretty emphatic about

wanting to give the topic a rest.</p>

 

<p>By that I mean, don’t be surprised or offended if this thread gets shut down (or even deleted?)

very quickly. The horse, if not dead, is bleeding profusely.</p>

 

<p>Cheers,</p>

 

<p>b&</p>

[digitaldog]

<p>It will not get shut down if everyone behaves! FWIW, I applaud Josh for closing down those threads.</p>

[nathaniel_alpert]

<p>Ben,</p>

<p>I read most of the responses in those threads. It seemed to me that there were some readers who were looking for guidance about how or whether to use ETTR techniques. That is the purpose of my post. I do not claim to have the most detailed knowledge, I am not looking for an argument. Most of us do not need perfection but a practical approach that can be used in the field. If others have a different approach to practical ETTR or wish to correct me, they should speak up. I don't need to win any arguments to succeed in my self-appointed goal :-)</p>

[ben_goren]

<p>[sigh]</p><p>Andrew, in your article, you described observing ±3% shifts in neutral patches; this is

consistent with my own observations. In the closed-down thread, you dismissed those shifts as

insignificant. Just now you concurred with Nathaniel’s statement about never noticing any shifts

or distortions.</p><p>Perhaps you’d care to clarify your position on the matter?</p><p>For reference, I’m attaching a picture with nine neutral patches. The bottom three are L* =

10, L* = 50, and L* = 90. The top six are L* ± 3.</p><p>Cheers,</p><p>b&</p><div></div>

[kelly_flanigan1]

<p>In movie/cine work lighting seems to be valued more than most still work. The bulk of still shooters seem to think that some magical single exposure will fix a poor lighting setting that has a range WAY beyond a sensor or film can grasp.<br>

<br /> With exposure you have to consider the specific film or plates response; or sensors response.<br>

<br /> Without considering each specific film/plate/sensors actual characteristics; detailed rules are not going to work well.<br>

You do not even what the sensor is doing; or know he actual code used to deconvolve a raw file; if it the camera has a bias either. Assuming that all cameras behave the same from an Engineering standpoint is like assuming all software is all the same; or everybody wears the same brand of shoes<br>

<br /> LONG AGO in landscape work and in days movie cine work folks use spot and incident meters; and talked about lighting ratios.<br>

I mention all this because in printing for the public; what ever trendy method of exposing there is still some folks do not get it at all. If ETTR makes you not blow the highlights and you use too big a margin; you might push valuable details into a more noisy shadow area; or totally loose them.</p>

 

[Ray House]

I often apply ETTR in my reading of the histogram and I find that reducing saturation, contrast and even sharpening with my in camera settings will give me a histogram closer to the RAW/NEF in Photoshop. Not exact, but closer. I might add that I use ETTR very conservatively and do bracket when possible using 1/2 stops. I keep thinking that an incident meter would really help, anybody have thoughts on that?

[jeff_schewe]

<p>

<blockquote>

<p>Nathaniel wrote" </p>

I have never noticed any color shifts or distortions from this approach. The best digital exposure is obtained by exposing so that the raw image histogram is as far to the right as possible (even if it looks clipped), so that (reducing) the exposure slider can bring it down to the point where there is no clipping.</blockquote>

<p>Correct, as long as you don't blow your highlight textures, the color rendering isn't an issue (despite some of the contradictory, radical and dramatic claims to the contrary). </p>

<p>However, it's not just the Exposure slider of ACR/LR that needs to be adjusted if you do an ETTR exposure, you do need to adjust the Blacks slider (black clipping) because of course with the increased exposure you have indeed moved all the tones of the image to the right including your black points...but really that will have very little impact on the color rendering of a lower contrast scene.</p>

<p>It will punch the saturation a bit (low contrast vs higher contrast tone curve) but that really shouldn't have any color shifts–the contrast slider and the parametric curves function of ACR/LR is designed not to impact hue, only saturation which you can dampen if you wish. So unless you don't really understand how to use the basic color controls of ACR/LR, ETTR will do nothing to radically alter your color rendering–and if it does, it's user error and not the fault of either ETTR nor ACR/LR.</p>

</p>

[michael_young3]

<p>The contrast and color shifts at +1.5 EV are not too horrible. By +2 EV, there's a distinct yellow'ish cast that's difficult to remove. It's as likely to be from drastically expanding the tonal scale during conversion -- only a very flat scene has room to move +2 EV -- as it is from the overexposure. For all that, none of the post processing headaches are worthwhile, even for the relatively mild shifts at more "normal" overexposures. The commonsense dictum of simply selecting lower ISO is still the best advice. It works for JPGs, and it works for RAW, while introducing no new post processing needs. I have not found anywhere on the web or in my own testing that overexposure captures more or better detail than simply selecting lower ISO.</p>

<p>If we're only talking about in-camera metering, my wife is very fair skinned. During the depths of our sun-less winter months, I need to dial in +1 1/3 EV compensation with the metering dot on her eye. Where I'm concerned, that's simple normal everyday metering. It doesn't matter what blows out, or where the histogram sits. My main subject is properly exposed. Select ISO to suit the shutter speed, and adjust lighting to tame the background highlights and flatten unflattering shadows. Nowhere is there anything nearly complicated enough to warrant an exotic sounding four letter acronym.</p>

<p>For Jeff's sea port example, the white warehouse wall is Zone VIII. Roll in +3 EV compensation, put the metering dot on the wall, and let Zone III take care of itself. Select the lowest ISO that suits the conditions, and click the shutter. Bracket it if you're still not sure. If that's still not enough for the pier wall shadow details, or some specular highlight detail beyond Zone X, you're now talking HDR, not ETTR.</p>

<p>ETTR is a wonderful sounding theory, but when I look inside the bag, it's empty but for post processing headaches. I wish there really was something more in there, but I find nothing that normal, naive metering and camera settings don't already address.</p>

 

[ben_goren]

<p>Michael,</p>

 

<p>Thank you for that clear, concise summary. In four paragraphs you expressed what I’ve been

trying to all along.</p>

 

<p>Cheers,</p>

 

<p>b&</p>

[garrison_k.]

<p><em>ETTR is a wonderful sounding theory, but when I look inside the bag, it's empty but for post processing headaches. I wish there really was something more in there, but I find nothing that normal, naive metering and camera settings don't already address.</em></p>

<p>bravo, Michael.</p>