Shooting in log?

[Stu Maschwitz]

The 11-stop figure comes from measuring the RAW conversion to RGB LOG, but not necessarily in Rec709 space... but are you sure that there are only two stops between 45 IRE and 110 IRE? That sounds like each stop equals a jump of 30 IRE...

I seem to have conflicting sources here, so any insight would be welcome—but according to one published formula, the Rec709 transfer function is:

EDIT (thanks Joofa): (1.099 * pow(x, 0.45)) - 0.099

...for values of x > 0.018.

That converts decimal scene values to decimal image values. So let's start with a scene value of 0.18.

(1.099 * pow(0.18, 0.45)) - 0.099 = 0.4090

Close-ish to 0.45 or 45%.

Next stop up is twice as much light as 0.18, or 0.36. After that is 0.72, etc. Here's how that plots out:

18% gray = 0.18 = 40.9%
+1 stop = 0.36 = 59.5%
+2 stops = 0.72 = 84.9%
+3 stops = 1.44 = 119.6% = clipped

If you expose 18% gray at 0.45 you get these values:

18% gray = 0.18 = 45%
+1 stop = 0.36 = 65.1%
+2 stops = 0.72 = 92.55%
+3 stops = 1.44 = 130.05% = clipped

-Stu

 

[Dj Joofa]

I seem to have conflicting sources here, so any insight would be welcome—but according to one published formula, the Rec709 transfer function is:

pow(x*1.099, 0.45)

...for values of x > 0.018.

The correct formula is:

1.099 * pow (x, 0.45) - 0.099 for x > 0.018, and
4.5 * x for x < 0.018.

 

[Stu Maschwitz]

The correct formula is:

1.099 * pow (x, 0.45) - 0.099 for x > 0.018, and
4.5 * x for x < 0.018.

Oops, you're correct Joofa! I'll fix my maths, thanks.

EDIT: My post above has now been corrected, both the formula and the values (without substantive change to the findings or my overall point). Thanks again Joofa!

-Stu

 

[Dj Joofa]

Stu, You are most welcome.