Technical Question Is “Expose to the Right” relevant with current digital cameras

[Asher Kelman]

The late original editor and founder of Luminous Landscape, Michael Reichman, preached a rule the became a mantra for all of us: ETTR i.e., “expose to the right”.

The idea was to reduce shot noise.

We all thought that was something that Moses would have added to the 10 commandments had he been gifted with a digital camera of 10 years back!

Now arrived, Steve Perry, a new truth-teller who says otherwise! He questions the whole concept of expose to the right with modern sensors “shot noise only depends on number of photons collected by the sensor during exposure.

He claims that doesn’t matter how the aperture, shutter speed or even ISO is set! The only things that matters is how many photons were collected!

See the video here


Asher

 

[Asher Kelman]

A great description of ETTR can be found in Wikipedia, here!

 

[Asher Kelman]

So who will still use ETTR?

I am sobered up now and just care about getting sufficient light into the camera to minimize shot noise!

But can we do do better?

Remember that pixels, or more better “senses” which represent shadows, need to have low shot noise too!

Also the accuracy of the count is related to the total number of photons captured as electrons. Accuracy is determined by root n/n x 100.

So I suggest the following:

“bracketing a scene should get all the pixels in the exposure scale sufficiently counted accurately and also the shot noise should be optimized.”

Do you have any argument with that contention?

Asher

 

[Doug Kerr]

I would first want to be sure I know what you mean by "bracketing a scene".

Doug

 

[Asher Kelman]

I would first want to be sure I know what you mean by "bracketing a scene".

Doug

Exposure bracketing. You may not be familiar with modern bracketing options.

One can have focus bracketing whereby a series of shots is taken with some front focused and others back focused and then the shots are stacked and the focused areas are fused in an optimally focused entire picture.

The same is achieved in exposure bracketing, again built in to modern cameras, say a sequence of shots + 0.5, 2.0, 1.5 and2.0 EV and then normal exposure and a similar series of decreased exposure.

Asher

 

[Jerome Marot]

The late original editor and founder of Luminous Landscape, Michael Reichman, preached a rule the became a mantra for all of us: ETTR i.e., “expose to the right”.

The idea was to reduce shot noise.

A quick look to the link above gives the reason why the late Michael Reichman imagined that rule: at the time dynamic range of digital cameras was 5 to 6 stops. Therefore, fitting the dynamic range of the scene within that limited range needed compromises. Interestingly, the dynamic range of slide film (the kind landscape photographers used) was about as limited, so compromises were also needed for slide film.

What Michael Reichman noted is that the compromises were not the same. With slide film, one loses the shadows. With digital sensors of the time, shadows could yield some information, provided one exposed "to the right".

Today's cameras have 10 to 12 stops of dynamic range (the sensors may have more, but lens flare limits the end dynamic range). That compromise is no longer needed. One can "lift the shadows" of any picture without noticing negative effects.

As a side note, the dynamic range of displays can be even more limited. This is especially true for prints, in particular on mate paper. There is a reason why museums display photographs as glossy prints under direct light.

 

[Asher Kelman]

Once again, Jérôme,

Your contributions provide the most valuable factual reminders of what’s relevant here: dynamic range, which, as you point out so succinctly, has released us from the previous compromises we were forced to endure!

I should have thought of that myself!

Thank goodness you are far more awake today than I am!

Asher

 

[Robert Watcher]

I’m a bit of a hack as a photographer, because I’m never really too concerned with trying to get my exposure “technically” right.

I have never used a histogram on my camera screen, when taking a picture —— not have I ever used a histogram in my editing software. I am totally familiar with how it works and what it means. I am always shooting in Auto modes and constantly dialing my exposure compensation up and down based on what I see on my camera screen or viewfinder (using mirrorless cameras where it is always in live view). If I want some confirmation in extreme settings, I can always use the blinking blue and orange shadow / highlight setting on my screen, to confirm what parts of the photo are out of range.

Anyway, because I have never taken note of the usefulness of exposing to the right on a histogram while shooting —— I grabbed my camera with the files still on from shooting extreme lighting situation the other night at a music gig. Surprisingly the use of the histogram would have been useless. In fact if I based my exposures on exposing to the right, I would have totally lost the mood of the setting.

I’m not saying that expose to the right isn’t useful or valid in balanced lighting conditions, if that is what someone wants to use, but it isn’t the end-all.

I’m sorry that these pics taken with my phone of the back of my camera, are such poor quality —— but they reflect my findings.


Here is using the shadow / highlight blinkies which shows the dark crowd and foreground out of range and crushed blacks. But that is fine with my artistic vision for the photo. I dnt notice any orange areas of overexposure - especially on the main singer —- and when I zoom in on the main part of the scene, I have an exposure that I am happy with. I’ll let everything else land where it may

Now the interesting part is 5ge Histogram reading of this scene. Good thing I didn’t rely on adjusting exposure comp to make the highlights reach the right side

And then there is this pic, where the orange overexposure blinkies are clearly visible on a spotlight and some other streaks of light. Adjusting to the right on the histogram for this, would have been traumatic.

I don’t know that this proves anything or is useful, but it just validates my use of the Shadow / Highlight blinkies for verification of extremes. Where I found this useful was in a Portrait job that I had last fall, where the large family had to be positioned in the open midday sun overhead and slightly behind. I filled in light the best I could with two portable flashes. I was able to benefit from the Shadow Highlight information to determine how much overexposure was on different parts of the scene, some areas I let clip as long as it didn’t affect the people. The one thing with this setting, is that the blinkies only show once you take a photo and review it. That worked perfectly for something like the portrait session.

 

[Asher Kelman]

Robert,

Today, we are saying, “expose to the right is not of any value”, as in post, one then has to decrease brightness! What’s the point of that extra step?

There is no point as, Jérôme points out, we now have at least 12 stops of dynamic range instead of 5 for film and early digital.

But your point of using the blue and red indications of under and over exposure is necessary, IMHO, always.

We are fools to ignore such simple and easy to use aids on our cameras and smart phones!

I am with you!

Check the area of interest is illuminate to your wishes on the screen or viewfinder and mostly the rest doesn’t matter.

If it does, then you need a studio setup or lights and the club or concert hall would boot you out in one Hollywood second!

Asher

 

[Doug Kerr]

An "accepted" way of defining the dynamic range of a digital camera is the ratio between (a) the scene luminance that falls at the "saturation" level of the imaging system and (b) the scene luminance at which the signal-to-noise ratio of the imaging system is some arbitrary value (often 1:1, which is of course a terrible noise result).

Of course we only fully exploit whatever range this if we expose so that the brightest part of the scene is essentially at the saturation of the imaging system.

That is, we expose to the . . .

Well, you know.

But what the relatively-great dynamic range of modern cameras, as to the photon noise issue, does lead to is that we get whatever signal-to-noise ratio is our bogey for scene luminances that are much less, compared to the saturation luminance, than for cameras with lesser dynamic range.

Looking at it in another way, for the noise performance we "require", we can afford to squander more of that greater dynamic range by not exposing to the . . .

Well, you know.

In any case, what we have start with is that (for a given sensor system) the photon noise performance for any scene luminance is related to the photometric exposure resulting from that scene luminance (given the aperture and exposure time in use).

Doug

 

[Doug Kerr]

Rob makes a very good point about the value of the under- and over-exposure "blinkies". (I assume that the under-exposure blinkies come from a photometric exposure that would produce a signal-to-noise ratio less than some bogey.)

If the scene has a relatively small luminance range, then there is quite a range of photographic exposure that will eliminate any "blinkies". It would be hard to give a general rule as to which photographic exposure within that range would be the best, and best in what regard? Except that the greater phtographic exposure will probably be "better" from a photon noise standpoint.

If the scene has a quite large luminance range, there may be no photographic exposure that will eliminate essentially all blinkies of both directions.

Doug

 

[Asher Kelman]

Rob makes a very good point about the value of the under- and over-exposure "blinkies". (I assume that the under-exposure blinkies come from a photometric exposure that would produce a signal-to-noise ratio less than some bogey.)

If the scene has a relatively small luminance range, then there is quite a range of photographic exposure that will eliminate any "blinkies". It would be hard to give a general rule as to which photographic exposure within that range would be the best, and best in what regard? Except that the greater phtographic exposure will probably be "better" from a photon noise standpoint.

If the scene has a quite large luminance range, there may be no photographic exposure that will eliminate essentially all blinkies of both directions.

Doug

We don’t care, in this circumstance, (a concert, night club or assassination) should 3/4 of the photograph be bright white or black!

What counts is that the small zone of interest is adequately exposed.

Asher

P.S. [For an assassination, this is your money shot!

You can set your price at $500,000, or perhaps more, if you are the sole photographer present at the sudden demise of a Prime Minister, President or the CEO of an electric car company!]

 

[Jerome Marot]

Robert's latest concert pictures are very nice and I don't think they should be exposed any differently, but let me cite the first one he posted as an example:

View attachment 12576

In that picture, the white shirt is overexposed. This is what "expose to the right" seeks to avoid. Whether it would make sense here or not is not in question, the theory was invented for landscape pictures, not concert pictures. In landscape, overexposure of the highlights often results in less than ideal skies and clouds.

Another point is that "expose to the right" does not make much sense when shooting jpeg. The process requires post-treatment, so as to map the recorded dynamic range of the picture to the more limited dynamic range of the output. If the input and output are both jpegs, the two ranges are the same.

 

[Asher Kelman]

Jérôme,

You just wrote,

“In that picture, the white shirt is overexposed. This is what "expose to the right" seeks to avoid.”

Did you mean that or there can error?

ETTR, doesn’t avoid overbright whites, it causes them: demolishing delicate multiple gradations of bright areas in the clouds.

Asher

 

[Jerome Marot]

ETTR seeks to avoir overbright whites. One exposes to the maximum allowed, yet avoid saturating the values. At least that is how I remember the original article.

 

[Asher Kelman]

ETTR seeks to avoir overbright whites. One exposes to the maximum allowed, yet avoid saturating the values. At least that is how I remember the original article.

Ideally, yes!

However it’s dangerous, LOL!

One might not always realize one is crushing delicate transitions of the bright side of thr grey scale!

Asher