(The sun wasn’t yellow)
Playing around with my image formation, which is now available, in a very janky state, for anyone to try, if you want (https://github.com/ilia3101/LuminanceDRT)
This file is licensed Creative Commons, By-Attribution, Share-Alike.
Another of your atmospheric sunsets, thanks for sharing it.
A chance to play with blends (my latest field of experimentation). The result is not intended to bear any relationship to reality! When I examined my luminosity layer masks I discovered three more people lurking on the right.
I have some difficulty to admit that at sunset, the sun, and it’s reflection in the water are white. If the sun was not yellow, it certainly was not white. So, it doesn’t seem to me a realistic rendition of the scene.
The sun was an intensely bright, beautiful, orange.
Zero yellow. Zero white.
But try representing that sensation on a display with smooth tonality.
@age I like yours.
Very nice work on your display transform, Ilia! It seems incredibly smooth indeed.
I’ve started prototyping a filmic v7 for darktable (nothing published yet though…) Here’s a result from that one.
Yes, depending on weather, color of sun at sunset can vary from yellow to orange and red.
I was surprised as in your rendition the sun is quasi white and so is the reflection on sea.
I see this beautiful orange only on some renditions like the one of @Jean-Marc_Digne .
That’s not your first rendition😗
hi @flannelhead , good to see you’re looking at things like this.
flannelhead and all display transform people, in respect of the gamut mapping in V6, someone suggested quite recently that the gamut mapping could/should be a process separate from filmic. I think that makes sense, after all, what if you’re changing things later in the pipe than filmic? Should it be a separate module? Should it be part of a revised Output Profile module?
Indeed, that’s what I’d like to propose if I ever get around to publishing a PR with filmic v7. It would make sense especially as some are using modules like local contrast after filmic. And users of base curve would benefit from this, too.
I agree. In my display transform experiments, the gamut mapping occurs separately after the display transform is applied (I’m using @bottosson 's OkLab gamut mapper - it works well and is computationally simple). It depends on the display transform you are using though: some of them do gamut mapping as part of the transform (I think @ilia3101 's is one of these, the ACES ZCAM DRT candidate is another).
I went for orange, or as close as I could get to it. I used the HLR guided laplacians and filmic reconstruction to, I think, good effect.
darktable-4.1.0+1204
DSC00162_06.ARW.xmp (40.7 KB)
Thanks for the play. Nice sunset shot.
