Originally Posted by sotti
As far as I know, there isn't any commercially available software (this week) that has the BT.1886 function in it, at least that I know of.
That's no excuse.
It wasn't all that long ago when I was writing everything down on paper to input into HCFR manually ... and before that I didn't even have HCFR. Now, where did I leave that slide-rule.
But seriously, I can see where BT1886 might throw a monkey wrench into your workflows since you need to know Y at 0% and 100% first.
To be honest, I'm not quite sure what those JND graphs are trying to do ... I can easily notice the difference between the sub 5% stimulus bars on clipping patterns even with power law functions.
To me, power law "crushes" near blacks for two reasons:
1) Our standard method for setting "black level" ensures that we don't clip anything above 0% stimulus at the expense of compressing or "crushing" the dynamic range near black.
2) Following the power law curve exactly as far "down" the curve as we can adjust, will ensure that at some point (~4.6% Stimulus on my set @ 2.2 gamma) near blacks will simply merge with black (aka clipping.) IOW, if we follow standard practice for setting black level, but neglect to apply at least a linear offset black level compensation to our power law EOTF, we haven't really solved anything.
In actual practice, we don't really clip anything, we just wind up compressing the dynamic range near black. BT1886 appears to be designed to drop a nuclear bomb on the issue once and for all.
PS: On further review, I think your JND graphs are demonstrating exactly what I attempted to describe above, I'm just not sure how they got there.