What is "Middle Gray"?

[xpatUSA]

Many would say "halfway between Black and White" and would leave it that. Others would say "18%" reflectance a la Gray Card.

After much research lately, I find that it is whatever Chuck Norris says it is ... sorta.

For example, Adams perpetrated 18% while Kodak preferred 16% ... both being "Norrises" in their time.

In the (in)famous Zone system, 18% is indeed the logarithmic middle of Zone V. For Kodak 'steps', the logarithmic middle of their gray step is 16%

Comments welcome!

 

[Ken Rennie]

I used to use 18% but now I don't use grey cards for exposure, only white balance. Nowadays I just push the exposure to the right just stopping before highlights are overexposed. This, of course, means that my final processed image may or may not reflect the brightness of the scene captured. Ken

 

[xpatUSA]

In the (in)famous Zone system, 18% is indeed the logarithmic middle of Zone V.

So the boundary where Zone V becomes Zone VI is not 24% as one might expect but is actually 25.5% and similarly the boundary with Zone IV is 12.7%, not 12%.

 

[xpatUSA]

I used to use 18% but now I don't use grey cards for exposure, only white balance. Nowadays I just push the exposure to the right just stopping before highlights are overexposed. This, of course, means that my final processed image may or may not reflect the brightness of the scene captured. Ken

Indeed, one could almost say that ETTR is the king of digital exposure, whereas the Adams-meister says to "place" the region of interest where he wants it and to hell with stuff in the scene that is too bright or too dark which he will dodge or burn later.

The classic drawback to ETTR is the sunlit motor-cycle shot with all those specular highlights, as I'm sure you know. In other words, Exposing-To-The-Right for some highlights can get one a pretty dark image ...

 

[JimFox]

That's an interesting thought, that's something I haven't thought much of since digital came out. ETTR, has been how I have exposed now for the past 20 years.

 

[AlanLichty]

I am yet another in the ETTR camp. I meter my captures based on metering off of the brightest spot in the frame and allow no more than one half to one stop over the metering midpoint. At that level I can recover everything in the frame with our present day software. I haven't really seen any reason to get lost in the techniques that were all important in the era of film. I shot film for over 40 years and have no desire to go back to its limitations.

 

[xpatUSA]

I am yet another in the ETTR camp. I meter my captures based on metering off of the brightest spot in the frame and allow no more than one half to one stop over the metering midpoint. At that level I can recover everything in the frame with our present day software. I haven't really seen any reason to get lost in the techniques that were all important in the era of film. I shot film for over 40 years and have no desire to go back to its limitations.

Out of interest, may I ask what type of metering, e.g. Spot, etc? In-camera or separate?

 

[AlanLichty]

Out of interest, may I ask what type of metering, e.g. Spot, etc? In-camera or separate?

Spot - in camera. I typically shoot in manual mode so I seek the bright area of the desired scene and set my exposure values on that followed by framing and focusing the shot.

 

[Ken Rennie]

Manual mode for me with blinkies set to 109%+ this gets me a generally accurate idea of the clipping point. The exceptions are red poppies and bluebells where the colours very quickly run out of gamut, especially if you are going to convert to sRGB for social media. I also have an external Sekonic meter

 

[xpatUSA]

If we like cloud contrast, I believe ETTR could reduce cloud contrast because many cameras' raw to RGB show an S-shaped function:

image courtesy of DP Review

Derived from photos of a Stouffer 21-step transmission wedge normalized to set the mid-tone (+0.0) output as 118/255. Clouds often occur up around 240 due to ETTR and it can be seen that there is a shoulder there with reducing contrast as the step-wedge value increases. Looks to me ETTR aficionados might like the Ricoh in their bag ...

Sometimes I find that a reverse 'S' curve improves both highlights and shadows, albeit with a reduction in mid-tone definition. Sorta kinda like a transformation back to a linear rendition (no gamma).

 

[AlanLichty]

You can eliminate the S curve by creating/using linear profiles as a starting point for your post processing. I frequently find them useful when dealing with very wide dynamic range as well as shots with intense reds and blues as Ken mentioned above.

 

[Ken Rennie]

If we like cloud contrast, I believe ETTR could reduce cloud contrast because many cameras' raw to RGB show an S-shaped function:

image courtesy of DP Review

Derived from photos of a Stouffer 21-step transmission wedge normalized to set the mid-tone (+0.0) output as 118/255. Clouds often occur up around 240 due to ETTR and it can be seen that there is a shoulder there with reducing contrast as the step-wedge value increases. Looks to me ETTR aficionados might like the Ricoh in their bag ...

Sometimes I find that a reverse 'S' curve improves both highlights and shadows, albeit with a reduction in mid-tone definition. Sorta kinda like a transformation back to a linear rendition (no gamma).

Thankfully things have moved on from 2009. All modern camera sensors are linear. However Adobe Camera Raw and Lightroom from Process version 3 onwards, we are at 6 is non linear at the bright end and tails the brightness off. Anyone who uses both Adobe and Capture 1 will hsve come across blown highlights in C1 that ACR thinks are still good. Setting the clipping point can't be done in ACR. Adobe's highlights and whites sliders are both non linear and dynamic ie their effects vary depending on the range of brightnesses in the scene. The dark end sliders are the same. It is posdible to fool ACR into behaving linearly (almost) but this reply could get very long (and boring). I spend a lot of time photographing clouds and bracket shots. If none of the channels are clipped then I can make my clouds look identical from shots several stops apart, apart from the brighter ones have less noise. However determining clipping can't be done using ACR and trying to darken clouds that have had lots of highlight recovery applied will result in a horrible looking mess. This is not due to non linearity of sensors but Adobe's implementation of RAW processing. Ken

 

[xpatUSA]

You can eliminate the S curve by creating/using linear profiles as a starting point for your post processing. I frequently find them useful when dealing with very wide dynamic range as well as shots with intense reds and blues as Ken mentioned above.

Yes, I've done that with the GIMP, too. Quite useful!

In fact, at one point this year I had accidentally configured it such that it only exported Linear images ... not quite so useful, grump.

Ah, yes, the classic red flower petals with no greens, ergo 100% HSV saturated, often where someone (not here of course) edited to taste in ProPhoto or Adobe and then saved as sRGB.

 

[AlanLichty]

Yes, I've done that with the GIMP, too. Quite useful!

In fact, at one point this year I had accidentally configured it such that it only exported Linear images ... not quite so useful, grump.

There are images where linear profiles are a must but in my experience there are more that don't. Don't ask me how I know this

 

[xpatUSA]

Thankfully things have moved on from 2009. All modern camera sensors are linear.

Yes. However, DPR is showing that processing from raw to RGB is generally not linear. I had one model which, in extended 50 ISO was ,like the Ricoh, dead straight up to 255. On the other hand, some OM cameras gain "DR" just by considerably flattening the highlight part of that curve.

DPR's shooting of that Stouffer step wedge really does tell the tale.

 

[Ken Rennie]

Yes. However, DPR is showing that processing from raw to RGB is generally not linear. I had one model which, in extended 50 ISO was ,like the Ricoh, dead straight up to 255. On the other hand, some OM cameras gain "DR" just by considerably flattening the highlight part of that curve.

DPR's shooting of that Stouffer step wedge really does tell the tale.

Yes it tells us that most processing defaults introduce a contrast increasing s curve. However it is not that difficult to reverse this especially as the last stop contains 50% of the data so when we do apply the reverse S we are unlikely to get into posterisation. I do not apply a reverse S but in ACR reduce the exposure, reduce the highlights and increase the whites. This increases the (almost) linear range of the whites and allows images with many subtle levels of white, handy for clouds, snow, waterfalls and bride's dresses. I could use a linear profile but would then spend some time increasing contrast ( not globally) as this would negate the linear profile. Alternayively I could use an older camera with a linear camera to RGB profile and put up with the reduced dynamic range and increased noise. Ken