Sorry for the HTML! The problem is on how to change the value of the module's slider before doing the computation. The effect still have to be applied to the pixels, replacing the old ones. The "preserve lightness" option would only rescale the chosen values in a way that preserve the overall lightness. I played a little with the blending operator and it seems it's not the same (am I doing something wrong?). It helps to scale down the lightness when the module's output is set to any kind of color information. But it doesn't help on a grayscale conversion. Furthermore it becomes difficult to know exactly what value to apply to the blending opacity in order to keep the original lightness. For a better explanation of this feature, see the topics "Preserve Luminosity" and mainly the "In Monochrome mode" from https://docs.gimp.org/en/plug-in-colors-channel-mixer.html. Please, let me know if there is another way to achieve this effect.
Cheers -- Caio S. Souza Laboratório de Biologia Teórica e Computacional Universidade de Brasília www.lbtc.unb.br 2015-11-23 6:12 GMT-02:00 Roman Lebedev <lebedev...@gmail.com>: > On Mon, Nov 23, 2015 at 1:44 AM, Caio S. Souza <caiobio...@gmail.com> wrote: >> Hello, >> To preserve luminosity when using the Channel Mixer module one must be >> careful in choosing values for the red, green and blue sliders such that >> their sum equals 1. If we need to change the sliders values a lot (e.g. when >> experimenting with their importance on a black and white conversion), it >> becomes very slow and painfull to drag all others sliders appropriately to >> keep the overall luminosity. Mainly when you use differents values for >> different parts of a single image. It would be much faster and easier to >> have an option to keep the image luminosity, like in Gimp's channel mixer >> (https://docs.gimp.org/en/plug-in-colors-channel-mixer.html). This is >> relatively easy to implement: we just need to divide the computed value by >> the sum of the three chosen weights. >> I took a look at this module's source code and identified the pieces that >> must be changed (including the opencl kernel). If you agree this feature is >> a good idea, I may help implementing it. What do you think? >> > > 0. html mail is bad :) > 1. some blending operator is the solution here, i believe. > have you tried, e.g. color/Lab color/HSV color ? > >> Cheers >> -- >> Caio S. Souza > Roman. > >> Laboratório de Biologia Teórica e Computacional >> Universidade de Brasília >> www.lbtc.unb.br >> >> ___________________________________________________________________________ >> darktable developer mailing list to unsubscribe send a mail to >> darktable-dev+unsubscr...@lists.darktable.org ___________________________________________________________________________ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
