[AstroDave]

Dave, who told you that if you expose for 18 percent gray you're just one stop from blowing?

That's not quite what I said, but it's just arithmetic:

The usual convention is that "correct" exposure with an 18% gray card gives you R, G, B values at essentially 128 pixel value for the card - half way to full scale. Twice as much exposure, for a linear detector (one where twice as much light gives you twice the pixel count - what we use in astronomy) puts you at pixel value 255/256 - FULL SCALE.

Partially to save up from ourselves, and improve dynamic range, I guess, our cameras don't do that. The scale is not one-to-one. My experience shows that a 1-stop increase or decrease in exposure near mid-scale causes about a 40-50 pixel value change (for my K-1 and K-3), so the camera scale is not 1-to-1 for our DSLRs.

At any rate, you are right - you can blow out the highlights more easily with a white card, but a little bit of care easily gets around this.

[stevebrot]

Dave, who told you that if you expose for 18 percent gray you're just one stop from blowing?

More like 5-6 stops


Steve

[clackers]

That's not quite what I said, but it's just arithmetic:

The usual convention is that "correct" exposure with an 18% gray card gives you R, G, B values at essentially 128 pixel value for the card - half way to full scale. Twice as much exposure, for a linear detector (one where twice as much light gives you twice the pixel count - what we use in astronomy) puts you at pixel value 255/256 - FULL SCALE..

Nuh, doesn't work that way, Dave.

The dynamic range of a K-1 is not two eV, but more than fourteen.

And a further danger with white is that such surfaces will tend to glare, throwing off your spot metering unless you're using a polarizing filter or something. Best to stick with matt grey. You can substitute the palm of your hand for a quick and dirty reading.

[stevebrot]

The usual convention is that "correct" exposure with an 18% gray card gives you R, G, B values at essentially 128 pixel value for the card - half way to full scale. Twice as much exposure, for a linear detector (one where twice as much light gives you twice the pixel count - what we use in astronomy) puts you at pixel value 255/256 - FULL SCALE.

Nuh, doesn't work that way, Dave.

The dynamic range of a K-1 is not two eV, but more than fourteen.

Dave is sort of right. The proper exposure placement of a gray card will result in a 24-bit rendering of 127, 127, 127...half the numeric value of full scale. Half of full scale, however, is not the same as half the exposure. It is simply half the scale and the scale is limited to 9 or 10 meaningful stops.* Counting up from from 25 the sequence of 1-stop increments goes like this:

25 (0011 1001)
50 (0011 0010)
76 (0100 1100)
101 (0110 0101)
127 (0111 1111)
152 (1001 1000)
178 (1011 0010)
203 (1100 1001)
229 (1110 0101)

There is nothing magical about these numbers. They are simply where the curve falls.


Steve

* 255 is not really valid since that value is clipped. I supposed 1 (0000 0001) could have been included as the lowest increment of a ten stop scale, but I chose not too.

---

Last edited by stevebrot; 01-21-2018 at 08:20 PM.

[clackers]

Dave is sort of right. The proper exposure placement of a gray card will result in a 24-bit rendering of 127, 127, 127...half the numeric value of full scale. Half of full scale, however, is not the same as half the exposure. It is simply half the scale and the scale is limited to 9 or 10 meaningful stops.*

Sure - his flaw was assuming the 0-255 range was linear.

[BruceBanner]

Meanwhile for Bruce;