Thanks for all the interesting replies, although some missed the point :frowning
I get the idea that not all cameras underexpose images by default like my Canon 80D, and that some cameras actually overexpose images.
So some of us have to adjust the exposure down (ie to the left) to get the optimum exposure, while those like me have to constantly move the histogram to the right.
I will assume the author is from the first group whose cameras are overexposing images, and that my understanding of the topic was correct after all
So, I think in a “Getting Started” sort of section, such a statement is valid as an indicator of what the majority of images would require for an approximation of an ETTR exposure. Reality is, that rule of thumb won’t apply for all scenes. If your goal is to ETTR as a strict nailing of the highest highlight to a place just under where the sensor saturates, you are going to find wild variation in how “well-exposed” is the rest of the scene.
So many factors dictate how light bathes a scene that it is nigh well impossible to come up with a single aperture and shutter speed that accommodates all parts of a scene.
Regarding the so-called over- and under-exposure tendencies of particular cameras, you won’t know that until you conduct a controlled test, comparing two cameras’ equivalent metering modes against the same scene. And, don’t let that metering mode be matrix; even if both vendors use machine learning in their matrix mode, they are sure to have used different training data. So really, spot-metering a 18% gray card bathed with the same light source is probably the only sure way to know what you assert. And, based on what I wrote in my previous post, I think you should realize this would not be an endeavor I would consider worth pursuing…
So, back to the minus-EV rule of thumb, it’s based on how straight-metering for a middle gray will sometimes put highlights above sensor saturation. So, the author recommends a -EV to avoid blowing highlights, knowing that it’s safer to pull up shadows than to let some highlights blow into oblivion, Depending On The Scene…
1 Like
As said: ETTR is just a question of clipping avoiding… If that’s what you want. Some sunsets are not possible without clipping sun disk, for example. And sometime, working with low dynamic range pictures, that theoretically would be better to overexpose in ETTR to use the best area of sensor (the upper one in lightness), and then compensate underexposing when processing, I feel more comfortable having them with more room in the right area of histogram, so that I can more easily get the proper colors and exposition.
Question of experimenting and finding what your camera offers, and what do you really like.
I do use Zebras in my Sony a6400 as a starting point (at 100+), and then, depending on the scene, I do adjust the EV to make them go out completely or partially.
My old camera (a Canon 300D) gave me a nice option with the possibility of install the non official MagicLantern, which was soooo useful… I miss it in some questions (even if I’m more than happy with my Sony), but mainly in the possibility to add a real RAW histogram, and the very easy way of making it auto adjust to a perfect ETTR exposition, and then to work from there if that was not OK for me. But to be true, the bigger dynamic range of my present camera makes easy to just be a bit underexposed (in the ETTR sense, not in what I do try to get) without big problems.
Just try and find what works for you.
Al rules are made to ignore them once you what what you want.
I understood the misunderstanding not long ago.
For studio photographers, working in artificial light with reduced contrast (4 to 6 EV), ETTR implies “overexposing”, as in “exposing brighter than what the lightmeter would rather have you to do”.
For outdoors photographers and anybody working with primary light sources in the frame, that implies “underexposing”, as in “exposing much darker than what the lightmeter would rather have you to do”.
This is because of a legacy of digital photographers still working under the assumption that camera sensors have very small dynamic range (smaller than color film). Which is not true anymore since 2013-2015, at base ISO. With cameras closing to 14 EV of dynamic range and very low noise level, ETTR rarely ever means overexposing anymore.
Regarding over/under-exposure, by the way, those terms are bullshit and mean nothing. They are to be taken solely in the meaning of “in comparison to what the camera lightmeter would expose”.
Exposing is not about highlights or noise, it’s about managing both ends of the dynamic range to salvage details everywhere.
10 Likes
Salut, mon père!
But is it possible to salvage details at the low end as well as at the high end?
Have fun!
Claes à Lund, La Suède
1 Like
Yes, as long as your scene DR is smaller than your sensor DR… which happens most of the time, when you don’t shoot sunsets or Christmas lights.
3 Likes
If the scene DR is smaller than the sensor DR, nothing has been lost (certainly not at both the low end as well as the high end), so nothing needs to be salvaged.
The terms “under-exposed” and “over-exposed” are overloaded, they have incompatible definitions, so beware. If they mean “less or more exposure than the meter says”, well, that is one definition.
But who is in charge, the meter or the photographer? If the ideal exposure is less than the meter says, then that (to me) is the correct exposure, not an under-exposure.
1 Like
For a lot of people the meter is in charge 
I’m mostly shooting in aperture priority with highlight metering mode and exposure compensation set to +1.7 EV on my Sony A7 III which Fred Miranda wrote about. He calls it automated ETTR, but that isn’t exactly true.
He’s using the A7R III, but mostly works with A7 III too.
There’s a feature on Fuji X cameras called Natural Live View that turns off the jpeg processing so the highlight alert (blinkies) more closely match the raw data. I haven’t done any tests with it yet to see how close it shows true clipping, but I’ll give it a try soon. Usually I can get away with about an extra stop from the blinkies with the usual jpeg histogram / highlight alert.
Thanks again to all who replied. Let’s assume my camera can handle a DR of x, and the DR of my scene is x - 2 (please excuse my elementary maths). My camera will want to expose the image so that the histogram is smack in the middle. I have the choice of lowering (by 1) or increasing (by 1) the exposure without blowing anything.
But increasing the exposure (i.e. moving the histogram to the right) will capture more data in the highlights and the shadows, and that is what I understand as ETTR.
Seeing ETTR and lowering the exposure in one sentence had me doubting my understanding of the subject for a moment.
Having said that, I want to add that I appreciate the effort that goes into creating and maintaining user manuals and I do not want to shoot down the efforts of the person who wrote it (without receiving compensation)
Highlight metering is already ETTR. That advice is bogus, let exposure bias at zero.
Not sure I can agree with that, for his particular camera. As I said before, compared to the in-camera jpegs, Sony raws have at least 1EV margin in the highlights, due to Sony’s processing choices. Highlight metering mode (a camera feature in this camera model) will still leave that margin available, unless the in-camera processing changes when using that metering mode.
Early on in my journey, I came to a similar perspective. But it doesn’t really work that way. Exposure compensation is a essentially a multiplication, so the lower data moves less in magnitude than the upper data for a given multiplier. You’d have to be doing adding and subtracting to get the behavior you’re describing.
Indeed, Expose To The Right isn’t always a positive EC. Depends on the dynamic range of the scene; think about an interior shot with a window on the daylit exterior; if you’re going to not blow the highlights of the exterior, you’re going to have to use an exposure with a -EC, and the interior will be sunk in the shadows. Drawing the curtains on the windows now probably makes the scene uniformly lit, with a radically different EC to preserve the highlights that are now in the interior.
ETTR really means “Anchor Your Exposure To The Right”, not “Move Your Exposure To The Right”…
1 Like
… Which translates in: switch to spot metering mode, measure the brightest part of the scene, lock, recompose, shoot. (EDIT: That’s what I do with an old Nikon D7000)
Right?
1 Like
No, in general the metering assumes the Spot is 18% grey 
Ah, ok, forgot to mention: … Increase by some ammount the exposure. In my case, usually +0.7 EV works well
Yeah, I picked the word “Anchor” more as a reference than a determinant; geesh, does that make sense?
@gadolf, @MStraeten is right, if you measure the brightest part of the scene with the spot meter mode, you’ll be leaving two stops of headroom. It’d be more like, “Set EC to +2 (or so), then measure the brightest part of the scene…”
Edit: beat me to the punch… 
1 Like
Ok, forgot to mention that I add a positive exposure compensation to shift the histogram to the right, but I’m conservative and usually add just 0.7 EV. I have to review that.
I won’t beat you, I’m a peaceful guy 
But dynamic range is expressed in stops/EV values, which are logarithmic. So multiplying the exposure means a constant shift in EV value over the whole range?