Overexposing

unpopular · Nov 14, 2011

^^ what exactly do they mean by 12% or 18%? 18% or 12% of what, and what is 100% or 0%?

I'm guessing this is something left over from densitometers and film, right? Or percentage on a logarithmic scale, maybe? If you dial in 18% of black in your favorite imaging app, you'll get a very dark shade. So what's up with this "middle grey" at "18%"?

This is teh one thing that always confused me.

MLeeK · Nov 14, 2011

12 to 18% of LIGHT/Dark. 100% is pure black and 0% is white. Why is that so difficult to understand. 12-18% on the scale from light to dark... Not color, LIGHT.

MReid · Nov 14, 2011

Adjust the brightness on your screen, set up your blinkies to tell you when highlights are blown. Learn to shoot in Aperture priority and use the exposure compension dial.....are you messing with us ???

clanthar · Nov 14, 2011

unpopular said:
^^ what exactly do they mean by 12% or 18%? 18% or 12% of what, and what is 100% or 0%?

I'm guessing this is something left over from densitometers and film, right? Or percentage on a logarithmic scale, maybe? If you dial in 18% of black in your favorite imaging app, you'll get a very dark shade. So what's up with this "middle grey" at "18%"?

This is teh one thing that always confused me.

18% reflectance. A grey card reflects back 18% of the light that strikes it.

Joe

unpopular · Nov 14, 2011

Ok.

But why then does 18% reflectance relate to 50% transmission on film, or level 128 in 8-bit? Why doesn't 18% reflectance = 18% density? Why does it render as "middle grey"? Is it just how our eyes see things, or is this one of those logarithmic scale things that always confuses me so much?

Mike_E · Jun 17, 2014

DIY: Making Your Own Gray Cards - DIY Photography

Instead if 82% substitute 88% brightness. An 8.5 X 11 inch matte photo sheet works best.

coffeefilter · Jun 17, 2014

MLeeK said:
Why is that so difficult to understand. 12-18% on the scale from light to dark... Not color, LIGHT.

Chill out. No need to get up in arms over a question about what 18% gray means.

WayneF · Jun 17, 2014

unpopular said:
Ok.

But why then does 18% reflectance relate to 50% transmission on film, or level 128 in 8-bit? Why doesn't 18% reflectance = 18% density? Why does it render as "middle grey"? Is it just how our eyes see things, or is this one of those logarithmic scale things that always confuses me so much?

This one is a can of worms, but of course you are right, 18% is obviously 18%. It is NOT 50%.

However, the quirk of the human eye/brain is that our eyes response is logarithmic, not linear, and our brain does think we perceive 18% as being about middle gray. That is the only thing "middle" about 18%.

In the past ages, printing press printers calibrated their ink flow so that an 18% half tone test pattern printed to look to be about middle gray to the eye.

Ansel Adams had the notion that his middle Zone V ought to be 18% too. Eyes might see it that way (Ansel did this in the 1930s, with film only, and surely he never saw a digital histogram). So he called 18% to be 50%, and we were sort of stuck with it. But reflective light meters today are calibrated to 12.5%. Ansel's was not a big error, we always tweak things anyway.

Anyway, we humans got confused (can't keep one concept of middle separate from another concept of middle), and when digital came along, and histograms had an apparent middle point. So we continued to imagine 18% (analog) ought to be about mid point of a (digital) histogram. 18% ought to be 18% of course, and of course it is, in linear data (like Raw data).

But we don't see raw data, we see RGB data. All of our RGB images are gamma encoded (data in the histogram too), and then when gamma encoded, 18% comes out at 117, which is about 46%. Which is close enough to middle, so we imagined our dumb 50% midpoint notion was correct, when of course other factors we didn't understand were the only cause that 18% appeared there. But the actual 50% point is gamma encoded to be up around 183, or about 73% (in a gamma histogram).

As very easy and obvious proof to see this:

If you tweak the exposure of something bright or white, so that the right end of histogram data aligns very near 255...
and then underexpose exactly one stop (which we know is 50%), it will not move anywhere near 128 at histogram middle.
It will be closer to 3/4 scale. It won't be exactly 73% because the camera is shifting data to do white balance and contrast and stuff, hard to predict, but one stop down will be near 3/4 scale (not 50%), which is of course where 50% ought to be in the gamma histogram.

No one ever wants to know actual details however, so most ignore this, and continue thinking the midpoint of our (gamma) histogram is 50%. Which is quite wrong, they cannot explain why that could be true, but it really doesn't matter what we think 50% is... whatever happens happens anyway.