Exposing to the Right - Revisited

[Glenn NK]

A while back, we had some interesting discussion on ETTR - opinions varied considerably on the usefulness of the technique.

The following article can be DL'd from the following site. Of particular interest is the last paragraph on page 2, and the following comments on page 3.

https://www.box.com/s/qm2x26e6nmpeg7jwoktu

Looking forward to comments

Glenn

[Val Mansfield]

Thanks Glenn, very interesting reading.

Val

[shreds]

Does highlight the heavy reliance many have of the in camera histogram, which, as it points out, is a jpeg based version. Of course RAW has a wider latitude anyway, but I am glad that I have never been a camera histogram reliant photographer. (Then again I always shoot RAW+jpeg anyway).

Many are, however and I put this down, in part, to the way it is portrayed by the camera mags as essential to check all the time. Understanding basic technique BEFORE capture will result in many more 'keepers' regardless.

[Donald]

... I am glad that I have never been a camera histogram reliant photographer.

I think Ian gets to the nub of the issue here, with 'reliant' being the key word.

I do use the on-board histogram as an information tool. But it is only one part of the information I am gathering and assessing in preparing for the shot. I would agree that 'reliance' would turn you into a bit of an automaton, not doing any of your own thinking. Like any craftsperson, you know that you've got a range of tools to use in given circumstances and you've got to know how to use them to best effect.

I think the on-board histogram is an excellent tool, so long as I use it in an informed way to help me to make decisions.

[Colin Southern]

Hi Glenn,

It's a torch that Bruce started and Jeff (Schewe) carried on after Bruce's untimely passing. Technically it's quite correct, but my personal opinion is that it's a little misleading for 3 reasons.

1. In any given scene - when using ETTR - it would be normal for at least one of the channels to be close to clipping. In fact it would be correct to say that it would be normal for one channel to clip before the others. Couple this with the fact that although sensors are linear for the majority of their operating region, they start to exhibit non-linearity as saturation is approached. So ETTR runs the risk of pushing you right up to the "danger zone" - and if one goes INTO this danger zone (keeping in mind that ETTR is all about discarding the normal safety margin) then you're likely to capture information for one of the channels in a non-linear way ... that in turn will give you a weird colour cast in those highlights that's nigh on impossible to correct in post-production.

2. Deliberately "over-exposing" a scene by up to 2 1/3 stops means that the data then has to be shuffled down that amount in post-production. I'm not sure I quite understand why, but in my experiments back in the days when I was an ETTR person I found that I couldn't quite get the midtones where I wanted them. Close - for sure - but there was usually "something not quite right about them". It might be something to do with the gamma conversion - dunno - never quite got to the bottom of it.

3. The advice (a) doesn't take into account the dynamic range of the scene -v- the dynamic range of the camera (although Jeff does expand on this point in his latest book) and (b) it doesn't take into account more modern sensors. Case(s) in point ... if I'm shooting into the light and want to protect foreground shadow detail then you can bet your sweet bootie that I'll be using ETTR - but on the other hand - if I'm copying prints (that have a maximum dynamic range of only about 4 stops) then using ETTR with a camera that has a 12 stop dynamic range is simply pointless because with even a regular exposure (that typically has about a 2 stop RAW safety margin) I'll STILL be 6 stops from the noise floor (12 - 2 - 4 = 6). If we take a more typical example of bride in white dress - groom in black suit (so 4 stop range) - plus a couple of stops of shadow detail - with a normal exposure - then we're down to 12 - 2 - 4 - 2 = 4 stops above the noise floor -- probably a good "practical limit".

So what does all this mean? It really means that any potential benefits from ETTR are really proportional to the DR of the scene you're capturing. If it's a very high DR scene (eg shooting into the light but still wanting to protect foreground shadow detail) then ETTR is essential - but if it's of an average outdoor scene or something else of a limited dynamic range then it usually doesn't offer much benefit. And in the first example I'd normally either use a GND filter or bracket my exposures and produce an HDR composite to "hedge my bets" rather than rely on ETTR.

I admire Jeff's knowledge and skills immensely, but I do feel he's a bit like a "dog with a bone" on this one. It's a great theory, but any advantages aren't so clear cut in many (low to normal DR) situations. Keeping in mind too that today's camera typically captures 12 stops of info - but we display on monitors that can only display 6 or print on paper that can only display 4 ... so as a rule, the DR of the capture isn't the limiting factor ... even with modest DR compression.

EDIT: Keep in mind too that that article is now 9 years old ... it keeps appearing on every edition of Real World Camera RAW because each new edition is based on the old edition -- and I guess it's still "good padding"

[Colin Southern]

I think the on-board histogram is an excellent tool, so long as I use it in an informed way to help me to make decisions.

I tend to use it as an indicator of my approximate degree of under-exposure, and in contrast, I use the highlight alert ("blinkies") as an indicator of any degree of over-exposure.

For sure, an in-camera JPEG is conservative compared to a RAW capture - but - that conservativeness can also go by another phrase: "safety margin" (especially away from non-linear regions).

As I mentioned above - if we NEED to use the full DR of the sensor because the scene before us is a high DR scene then we have no choice but to use ETTR (unless we're using other techniques like GND filters or HDR composites) - but if it's just a normal DR scene then it's usually just a waste of time.

[pnodrog]

The capture is linear but from my reading of other articles the Analogue to digital conversion is not. So depending on manufacturer and the AD conversion used the RAW files vary in their linearity. The technology is evolving and is probably one of the reasons for the delay needed by Adobe to release the RAW conversion software on a new product. If RAW files were linear I think the conversion updates would be simpler and released more quickly.

In the past I tended to under expose slightly to ensure highlight detail but have found that I was being a unnecessarily cautious. ETTR may be a valid approach at present but I suspect it will become progressively less beneficial as sensor (including ADC) technology progresses.

[Colin Southern]

The capture is linear but from my reading of other articles the Analogue to digital conversion is not.

Good point.

In the past I tended to under expose slightly to ensure highlight detail but have found that I was being a unnecessarily cautious. ETTR may be a valid approach at present but I suspect it will become progressively less beneficial as sensor (including ADC) technology progresses.

These days I typically get my camera to EC = -1 and FEC = -2/3 for outside work ... I find it gives me richer (read "less washed out skies"), and doesn't result in an over-flashed look).

[pnodrog]

These days I typically get my camera to EC = -1 and FEC = -2/3 for outside work ... I find it gives me richer (read "less washed out skies"), and doesn't result in an over-flashed look).

I think you should revert to 50 asa Velvia and rate it at 80 - those were the good old days. It certainly improved saturation and avoided washed out skies and is still how I prefer images to look.

[Colin Southern]

I think you should revert to 50 asa Velvia and rate it at 80 - those were the good old days. It certainly improved saturation and avoided washed out skies and is still how I prefer images to look.

It certainly used to saturate like no other. Still prefer digital though

[herbert]

3. The advice (a) doesn't take into account the dynamic range of the scene -v- the dynamic range of the camera (although Jeff does expand on this point in his latest book) and (b) it doesn't take into account more modern sensors. Case(s) in point ... if I'm shooting into the light and want to protect foreground shadow detail then you can bet your sweet bootie that I'll be using ETTR - but on the other hand - if I'm copying prints (that have a maximum dynamic range of only about 4 stops) then using ETTR with a camera that has a 12 stop dynamic range is simply pointless because with even a regular exposure (that typically has about a 2 stop RAW safety margin) I'll STILL be 6 stops from the noise floor (12 - 2 - 4 = 6). If we take a more typical example of bride in white dress - groom in black suit (so 4 stop range) - plus a couple of stops of shadow detail - with a normal exposure - then we're down to 12 - 2 - 4 - 2 = 4 stops above the noise floor -- probably a good "practical limit".

Hi Colin,

When discussing the dynamic range of output media (e.g. around 4 for print, 6 for screen) it is important to consider the tonal gradation that is possible to show within that dynamic range.

Screens can output at 256 light levels (8 bit) or 1024 (10 bit) if you pay a lot of money. This is the upper limit for levels unless in between light levels are produced by flicking between brightness for each pixel. However the upper brightness and lower brightness set the dynamic range. Due to the inability to have a true black and the limits of display output this will be more like 6 stops.

Prints can output dozens of blobs of ink per pixel of different grey inks. This will allow a much larger number of tones to be displayed that the 256 levels of a monitor. So although the dynamic range is limited by Dmax (maximum black) and paper white, the number of tones that can be represented is very high.

In either case I feel it is valuable to understand that the 4 stops of print or 6 stops of screen is not a reflection of how good the output will look. If you have a 10 stop dynamic range scene captured in camera then you are not limited to only showing 4 stops of it in print. The ability to do tonal compression from a higher dynamic range capture allows many more stops of dynamic range to be represented successfully in a print or on screen.

I agree that poor tonal compression will lose the contrast in the scene where it is expected. For example you would want to print a white dress for the bride and black suit for the groom. If compressed too far then both these tones will tend towards grey and the resulting muddy midtone look. But with careful processing you can print much more than 4 stops of range because you have the advantage that the viewer will expect a certain appearance and the mind will compensate, e.g. it will accept a slightly less bright white bridal dress and less black groom suit if they fit within the entire image.

Alex

[revi]

(...)

In either case I feel it is valuable to understand that the 4 stops of print or 6 stops of screen is not a reflection of how good the output will look. If you have a 10 stop dynamic range scene captured in camera then you are not limited to only showing 4 stops of it in print. The ability to do tonal compression from a higher dynamic range capture allows many more stops of dynamic range to be represented successfully in a print or on screen.
(...)
Alex

That's true, but not relevant in the context Colin uses the 4 stop dynamic range here:
He is talking about making a reproduction of an print, and there the dynamic range to capture will be 4 stops (i.e. the ratio between the amounts of light reflected from white and black)

@Colin: I'm not sure I understand your calculation about the 6 stop safety margin though:
you say: '12-4-2 puts me 6 stops about the noise floor' while NOT using ETTR.
But that would imply NO margin at the highlight side, which is what ETTR is all about... Confused...

[herbert]

That's true, but not relevant in the context Colin uses the 4 stop dynamic range here:
He is talking about making a reproduction of an print, and there the dynamic range to capture will be 4 stops (i.e. the ratio between the amounts of light reflected from white and black)

Thanks for the correction. It makes more sense now.

In this case ETTR would not be a good idea due to the colour shifts that may occur in the highlights.

Alex

[Graystar]

I have a few points to express...

I'm a computer programmer and have looked into the process of turning RAW data into RGB image data, and can safely say that any concern over "RAW levels" is pretty much unwarranted. The reason is that bit values from the sensor don't directly become RGB values in the image. During the demosaicing process, 2/3rds of the data is simply invented from surrounding photosite values. Even the value of a photosite's own color is influenced by the surrounding photosites. This processing occurs in 16 or 32 bit (depending on the converter.) The demosaicing process is complex and has been improved over the years, which is just one reason why image quality has improved. And the bottom line is that every pixel value is the result of a complex series of calculations that greatly reduce the effect of the logarithmic nature of the data. So there's really no need to be concerned with levels.

As for ETTR in general, sure the concept is sound, but I find that I rarely photograph scenes that don't have something in them that's white or red...both of which severely limit the ability to ETTR. Consider the following image...



The exposure was set using my gray card (which I "calibrated" to my camera.) A second image with a third EV more exposure had clipping in the FLASH button and highlights in the white buttons. So despite the histogram and dark nature of the image, there is no way to ETTR the image. So one really has to rely on the blinkies function to know when things are overexposed. Well, I found that all too troublesome to perform, as the benefit was rarely there. Currently, in constant-light conditions, I just set exposure using my gray card. It's fast and accurate, as bright whites are never clipped. Yes it's true that if there are no bright whites or bright reds in the image, one could say that some highlight DR is being wasted. To that I say, but if there were bright whites and reds in the image, my exposure would be just fine, right? Well...to me, if the exposure level is just fine with whites and reds in the image, then it's still just as fine with whites and reds missing. So I just don't worry about it anymore.

I do agree that underexposing is always bad. I matched the spot-metered exposure given by my gray card to the RAW levels of my camera. The exposure I get when spot-metering my card will maximize the signal for reflected white. Specular reflections will clip...but to me that's okay because I think they should. Some people don't feel that way and prefer to underexpose to reduce the clipping. It just comes down to personal choice.

[Inkanyezi]

As for ETTR in general, sure the concept is sound, but I find that I rarely photograph scenes that don't have something in them that's white or red...both of which severely limit the ability to ETTR. Consider the following image...

/image/

The exposure was set using my gray card (which I "calibrated" to my camera.) A second image with a third EV more exposure had clipping in the FLASH button and highlights in the white buttons. So despite the histogram and dark nature of the image, there is no way to ETTR the image. So one really has to rely on the blinkies function to know when things are overexposed.

But then, that is exposing to the right. Mostly the histogram would not guide you to the correct value for keeping the brightest parts of the image near clipping or in the case of specular reflections, actually clip. The histogram is a statistical image of the tonality of the image, it does not convey highlights that are very small and appear only as ripples above the baseline or are even too small to appear at all. What way we arrive at correct exposure may vary. When there are large highlight areas that are easily identified, spot-metering the highlight with a tried-out plus compensation (on many cameras +2, but on some up to +3) may give the correct ETTR exposure, but for a black object as the strobe, that won't work so some other method is needed. The image, as I see it is exposed to the right, because it does have clipping or near clipping highlights, and that is what ETTR is about. It has absolutely nothing to do with the black areas piling up against the left wall of the histogram.

Mostly the problem areas can be identified with highlight control, and some cameras can do that prior to exposure. Live highlight on some mirror-free systems, and many compact cameras, most of them Canon, when CHDK is used, can show clipping before the image is taken, so that proper compensation may be applied. Red however is a problem there, as such firmware blinks only when all channels clip. Also the histogram indicates that clipping exists, but that's not much help as long as one often would allow clipping in certain areas.

[Graystar]

But then, that is exposing to the right.

Yes, that's right. Sorry if I wasn't clear...I meant that to ETTR such images you need to rely on blinkies and not the histogram, which is limited. I did not mean that I used the blinkies function to set the exposure.

The image, as I see it is exposed to the right, because it does have clipping or near clipping highlights, and that is what ETTR is about.

That's true. We're all after the same thing...best exposure possible. The difference is the method of getting there. Rather than intentionally adjusting exposure to achieve a maximum signal, I simply spot-metered my gray card and used the exposure that it gave me. The difference would be better demonstrated with an image of a black cat sleeping on a pile of coal, in which there likely are no bright highlights. In that case, my exposure would leave unused DR on the right, whereas ETTR would not. Yes, by taking a few shots and tuning the exposure, the ETTR image may have less noise...but I'll likely only get one shot before waking the cat!

It has absolutely nothing to do with the black areas piling up against the left wall of the histogram.

Well, lots of ETTR proponents say to use the histogram to set exposure. I just like to use this example as a demonstration that trying to set exposure via the histogram doesn't always work out.

[MilT0s]

I have a few points to express...

I'm a computer programmer and have looked into the process of turning RAW data into RGB image data

Can you post any links about that if handy?

[Graystar]

Can you post any links about that if handy?

Not really...

Here's an interesting PDF describing several demosaicing methods...
http://www.unc.edu/~rjean/demosaicing/demosaicing.pdf

Raw Therapee's AMAZE algorithm and the latest Adobe Camera Raw algorithms are even more complex.

I haven't found a description of the entire transformation process. You have to download the source code for DCRAW and review it. You can get it here...
http://www.cybercom.net/~dcoffin/dcraw/

You have to understand the C programming language...even then, a good debugger with a robust "watch" function is very helpful in following the process.

[Inkanyezi]

It has absolutely nothing to do with the black areas piling up against the left wall of the histogram.

Well, lots of ETTR proponents say to use the histogram to set exposure. I just like to use this example as a demonstration that trying to set exposure via the histogram doesn't always work out.

It's a pity that they do, and I know that many misinterpret the histogram in that way. It can sometimes be used to determine correct exposure, but particularly in the case of a scene with large dark areas and very small highlight areas, it is bound to fail. Blinking highlights is a more reliable way to find it out, and it is a great help to have highlights blinking before the shot is made. Unlike the histogram, the blinkies won't only show you that there is clipping, but the exact spots where clipping occurs. In that way the photographer may choose whether to allow those areas to clip or not.

I mostly accept that light sources and specular reflections blink, and sometimes I use the blinkies as a guide to where the sky would be washed out, knowing that setting exposure well under the blinkies, at least 2/3 steps, will render the sky OK. But then, I have blinkies prior to shooting. It's part of my light measuring. The histogram will show a red diamond when clipping occurs, but it won't tell how large or where the clipping area is.

[Glenn NK]

Blinking highlights is a more reliable way to find it out, and it is a great help to have highlights blinking before the shot is made. Unlike the histogram, the blinkies won't only show you that there is clipping, but the exact spots where clipping occurs. In that way the photographer may choose whether to allow those areas to clip or not.

We know that the histogram is based on an in-camera jpeg. What are the blinkies based on?

Glenn