« Two New Canons | Main | Flooding Affects Camera Manufacturers »

Wednesday, 19 October 2011


Finally some common sense written about ETTR. The thing about rules like ETTR is that they do not necessarily apply in all situations. A good understanding of any general rule and your equipment is mandatory to choosing how you are going to expose and process an image – shock, horror!

Personally, I never understood the obsession with always retaining shadow and highlight detail. Why can't shadows be black or highlights be pure white? Because it's "common practice"? How about exercising some aesthetic and technical judgement?

A good image is not always dependent on having full tonal range. Conversely, a bad image is not necessarily one that lacks tonal separation. The aesthetic and emotional quality of an image is made up of more than just tonal range, noise and/or grain. Daido Moriyama's work is a perfect example of this.

ETTR has progressed from being a useful rule of thumb to dogma.

"The darkest stop = 32 tonal values"

That would be correct if the accounting started at zero. However, because there is a noise floor, and for other reasons, in essentially all digital cameras there is always an "offset" at the A/D converter*, such that there are a few bits that fall below the noise floor. The offset is introduced in part to prevent the under-quantization that Mr. Pedde and others on this thread have shown some concern about. Noise (read noise, and under some circumstances shot noise) is a concern. Under-quantization is not. People who say that do not understand that the EE's who design sensors are smarter than they realize. The EE's recognized these problems decades ago, and solved them.

*Note that except for highly specialized scientific cameras, "digital" cameras are actually analog devices, accumulating charge (potential) that is first amplified by one or more analog amplifiers. Thes amplified analog voltage is then quantized by an A/D converter. Single electrons (or holes) are not quantified.

Dear Folks,

Data is not information. "Maximizing data" does not mean you are maximizing the quality of your photos.

Being concerned with "maximizing data" is one of those wonderful bits of physical theory that has very little to do with real photographic image quality.

pax / Ctein

Let's not get too bound up with rules (Rules are for people who can't see. If you can see, why would you need rules?) Sometimes you want to blow the highlights because you don't want or need the detail.

If you want a full tone print you will need to have some white and some black in it anyway (otherwise it isn't full tone).

Dear CN,

I don't think that's entirely fair. I think most people who even heard of ETTR treat it as a rule of thumb. Oh sure, there are a few dogmatists out there. We've got a few comments here from people who think we're talking about hard and fast rules (any time you see someone say something like, "Well, yes, but there will be cases when..." you know they don't understand what "rule of thumb" means).

But, by and large, I don't see that as a bigger problem here than in any other part of photography. I'm arguing that it's a very poor rule of thumb, that's all. Not taking on the dogma issue (although that might be a good future column topic).

Blocked shadows are rarely objectionable in practice. Annoying from the practioner's side, who always knows what they started with in the scene but had to settle for in the photograph, but they rarely prevent attractive art or are an evident flaw to the viewer.

For the reasons Mike and I discussed in the "B&W sensor" column, blown highlights tend to be much more objectionable. Not always-- rules of thumb, remember-- but often-to-usual. It's not merely their presence, it's that they're abrupt holes in the middle of the composition. They may be the single thing that skilled practioners hate the most about digital photography.

Anyway, I'm still just producing a rule of thumb, not a dogma. As many people have said, what you really want to be able to do is get the exposure right. (For whatever that means for you.) Unfortunately, that's not simple, in total -- it's a complex and difficult subject. Rules of thumb are valuable practical tools.

pax / Ctein

Nice to see that even well known photographers like yourself can be wrong.

This seems to be a departure from your usual style; it seems to be 1.1/2 of two separate topics.

As you point out, actually achieving a correct ETTR exposure is tricky given the capabilities of current cameras. But you usually give advice on how to achieve difficult things rather than saying "let's forget about them". As I understand ETTR - exposing TO the right, not PAST the right - it is identical to exposing for the highlights. So the first topic (about 1/2 of which you have written) covers these things.

The second topic is about the use of tone curves to deal with any shot whose exposure (necessarily to protect highlights or for any other reason) looks dark. In this one you have, as usual, given a clear exposition.

It just seems a pity to me that you felt the need for the first part to get to the second.

One of the motivating forces for some who practice ETTR is a dread of dark shadows. Absence of shadow detail to them is a cardinal sin. You frequently encounter this kind of thinking in photo critique forums, and I have no doubt that it is also what motivates much of the hideous HDR photography that is so popular these days. (I'm not saying that all HDR is hideous.)

I write this having just browsed through a book of photos by Mario Giacomelli. Now, there was a man who was not afraid of black shadows, but rather saw them as strong aesthetic elements in his photography. Others might take a lesson from him and learn to embrace the darkness.

Dear David,

You flatter me; I do wish I were well-known enough to be considered "well known." But thanks, it's nice to hear.

I've been wrong more than once in the past; you probably missed it because it's not all that common.[ahem]

I try to thank people when they correct an error, because I like to learn things. I'd rather not be wrong.

Happily, this is not one of those occasions.

pax / Ctein


I guess I should qualify my final sentence as:

"ETTR has progressed from being a useful rule of thumb to dogma" by it's proponents, as experienced by me.

In my experience, the dogmatists, like the fetishists, are always the most vocal (rabid?) on a particular subject. I have had some personal encounters with ETTR dogmatists, so I guess my comment betrayed my own particular bias.

I agree that digital highlights are less likely to be aesthetically pleasing, but as you noted in your article, this is due to the characteristics of the digital medium. The few photographers that I can recall off the top of my head who have used areas of white in their images (e.g. Daido Moriyama, Rinko Kawauchi, Trent Parke) all use(d?) film.

More generally, I also wonder if it there is a discomfort in Western culture with the concept of having holes (as you put it) – i.e. emptiness – in a composition. (Again, betraying my own personal background and interests.)

An article by you on dogma would be interesting, especially as to how it may relate to the fetishists.



Looks like Friedlander exposed to the right :)

Dear Gerry,

I think you misunderstood the impetus for this article. Go back and read the discussion that Mike and I had about B&W sensors to get the backstory and the context.

This is not a gratuitous dig at ETTR on the way to some other topic/technique.


Dear CN,

Excellent points. Dogmatists are an inevitable part of our universe. I tend to ignore them (except when skewering them en masse) because they don't represent the merits of an idea or technique. It's unfair to us to reject potentially useful methods because they're championed by fanatics.

I don't have any kind of opinion on your very intriguing idea that our collective dislike of abrupt holes is part of a larger aesthetic. It's worthy of serious thought.

As Mike and I talked about, and I explained in this column, there's no objection to true whites (or blacks) in a print anywhere in our discussion. It's the slam-into-the-brick-wall effect which offends, and WHY it offends is a fine question.

pax / Ctein

pax / Ctein

Hi Ctein,

Thank you for a very nice article on how to do ETTR.

Despite your titling (and you're probably aware of this), what you're carefully explaining in this article *is* ETTR, it's just that "ETTR" is a horrible, misleading, confusing name. ETTR has always been about exposing as far to the right as you can *without clipping highlights* - in high-contrast situations, this is "exposing for highlights", but in low-contrast, you can get a bit of extra quality by overexposing.

ETTR is emphatically not about setting your camera to +1 exposure. In fact, even getting close to the highlights without clipping can twist the colors unpleasantly. But saying that ETTR is bull, is... bull. It just heavily misunderstood and needs a name that people can understand. Like, say, "expose for the highlights". Or "expose to just before blow-out".

And to various commenters: It has nothing to do with "rules" like the golden ratio. It's a technical recommendation just like "use a tripod" or "keep your glass clean". There are technical limits to when it applies - I wouldn't use it in street photography, as I would not be able to adjust the exposure fast enough to ensure lack of clipping. But it does not tell you how to compose your shot in any way, just how to get the best tonal values possible when the situation allows it.


Regarding the aesthetic quality of digital highlights, I have recently begun noticing a "crunchiness" to highlights in editorial images (whether in newspapers or higher-end magazines)*. This may or may not be related to improper sharpening and/or digital processing. I wonder also if the delivery of files in JPEG format may be compounding the problem. In effect, perhaps the inherent weaknesses of the digital process is being amplified by a combination of editorial expediency and/or lousy workflow practice.

Whether this also applies to exhibition/fine-art work is another matter, but I have not come across any glaring examples of such, and given the nature of such work, I don't expect to.

*Not to imply that newspapers are low-end, but the nature of newsprint stock doesn't exactly lend itself to high-quality image reproduction.

[Feel free not to post what's below, as it's very off-topic.]

Going even more off-topic, I mentioned the larger aesthetic considerations purely as a spur-of-the-moment thought in the context of my interest in graphic design and Buddhist art, especially with regard to constructivist and modernist-era graphic design and Buddhist ink painting.

Photography hasn't seemed to me to wholeheartedly embrace the concept of white/empty space. (As much as I love Irving Penn and Richard Avedon's personal work, I don't generally consider that the backgrounds in still-life and fashion photography count.)


Unfortunately, this is one of those occasions Ctein. I have compared the shadows in B&W conversions on prints of 16x20 to 32x40 in size from a Canon 7D and Pentax K20D. Exposing to the right rather than allowing the metter to do its thing resulted in lower shadow noise. If you were correct, there would be no difference. As there was a difference, you can add it to those rare occasions when you were incorrect.

No disrespect intended....but I trust my eyes and what I see over someone who tries to convince me that the difference that I find obvious isn't there.

Sorry, and all the best.

You can't compare film to digital sensors. It's just not a fair comparison. Film and digital see the light in different ways. With film, you expose for the shadows. With Digital, you expose for the highlights. There is more detail in the shadows (for digital) when you expose to the right.

Sure, as sensors get better there is a possibility that this may no longer be the case, but many of us are using older equipment and must expose to the right in order to make sure we have detail in our shadows.

It's really frustrating to hear this from you.

I always thought that ETTR was exactly what you're advocating here. You expose the image as far to the right as possible without clipping the highlights. By doing so, you get the most information possible in the highlights and the shadows. It might be hard to judge in the field, but that doesn't make the technique invalid.

On difficult shots such as this one, it pays to bracket.

Didn't read all the comments so not sure if anyone already mentioned this. Even if so, it bears repeating.

There are some flaws in this analysis of ETTR. There are two conditions that need to be kept in mind when using ETTR.

First, the exposure is pushed as far to the right as possible without blowing out any hightlights. Some will say without blowing non-specular highlights but that's wrong. It's push to the right without any clipping. Period.

The second, and more important aspect of ETTR is that the scene/subject contrast or dynamic range has to fit within the range of the sensor. So if you're shooting something with a 12 stop range using a camera with an 8 stop sensor, you're hooped as far as ETTR is concerned.

Ctein seems to be considering the first but not the second. The scene in question is high contrast and outside the range of the sensor. For those situations you have to make a decision. Are you going to have blown highlights, blocked shadows or are you going to use some other method (exposure blending/HDR) to get what you want?

To be clear, I'm not one of those ETTR zealots. I rail against them as loudly as anyone. But even with today's low noise, improved DRange sensors, ETTR can still have a place provided it's used properly.

All the falderal about the in camera histo being based on a JPEG is just so much noise. With a modicum of testing, it can be pretty easily figured out how the in camera histo is showing the brightness range and how much head room there is to push to the right.

if you are shooting at 1600 and over expose by one stop to the right, to have less noise,you also halve the shutter speed. your effectivly exposing at 800, in which case you would have less noise anyway, so why not just use 800. with a balanced exposure. if anyone can tell me something I am missing please reply. tony harrison.

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007