Hello all
Since the creation of this challenge, and I remind you that this was one of the goals, I have brought several remedies to more or less annoying bugs. Some were difficult to resolve. The “Local Adjustments” code represents approximately, between the algorithms and the GUI, more than 50,000 lines of code.
Thanks to everyone who pointed them out to me.
Your image production with truly different processes is a source of significant improvements. Without these challenges, there would have been no significant progress. Thank you all.
Among the latest improvements, there is in particular, when several Spot (using either Cam16 or Log encoding), a better correspondence between the desired settings and their taking into account.
From my point of view, even if it appears in a hidden corner, the most important recent improvement in terms of algorithm is what ends up with the “Highlight attenuation” header. In particular the “Scope based” part, which, depending on the settings used, is either a “simple” highlight attenuator (rolloff), or a powerful “Tone mapping” function (over 23Ev).
Technical section for interested parties only- (aficionados):
For those who may be interested, the original algorithm (MIT - Thatcher Freeman) that I’ve modified for RT (also used in ART for the “Rolloff” part - thanks to Alberto ), uses a final formula, which connects a variable gamma to a linear part and is similar in spirit to a TRC, but with a huge advantage: the highlights part is asymptotic.
The formula :
Result = (dr * (x / (x + b)) + c) * kmid;// dr = dynamic Range, b = take into account slope, WhiteEv, BlackEv, Yb scene, Yb viewing, etc. , c = black point, kmid = relationship between Yb scene and Yb viewing.
If you use console mode with Verbose=true, you can see the impact of the settings - a line with:
b, gamma, slope, DynRange (dr), kmid, black (c), Yb-scale. (lost in all settings Ciecam)
Thanks again to all, the challenge isn’t over yet
Jacques