I did some tests, unfortunatly this is where we will hit the limitation of not using a nodal interface. With nodals blocks, you could assemble image filters in a Lego way, apply the output of a channel mixer to the input of the blur, then composite the output of the blur in some arbitrary RGB/Yuv channel over the initial image. All that from combining basic blocks that individually make sense.
The color stuff doesn’t belong in a lens blur module, from an UI perspective. It comes from another framework with another mindset. If we are to allow that, I will basically need to duplicate the channel mixer inside the blur module and that will become your usual bloated module, especially since we don’t have the luxury of working in display-referred here, so you need additional middle-grey and white fulcrums everywhere if you want contrast and brightness adjustments.
I understand now. Thank you for your patience and time to explain it!
Even without these options, I can’t wait for the new blur to show up in master. Diffusion module I have already tested and has become a permanent part of my workflow.
FYI - I’ve been encountering occasional crashes on DT 3.7.0+279~g5ffce60f0 when running the new diffuse module on Win 10, although the backtraces attribute the access violation to libcolorbalancergb.dll. I’ll submit a report to Github if/when I have enough information, but in the meantime I wanted to get word out in case anyone else has had similar issues.
Taking the discussion away from the topic (I believe it’s becoming a specialty of mine), I must say that this the finest grade I’ve seen from you. Sublime! Would love to see a run through of that edit in video format, but I’m quite sure other things are keeping you busy.
I narrowed down the issue to CMake… FFTW3 is not declared at all in there, while RT has it in CMake since forever, so that’s why dt doesn’t find it. Mind you… I’m not a soft engineer and never wanted to be, linking and building are stuff I don’t really understand.