Not sure how many of these edit were done with the default DT web safe srgb profile… It seems geared to offer a perceptual rendering intent and although you can select rendering intents I am almost certain it doesn’t have the luts to render them… Switching to an icc from icc consortium you can change the rendering intent which in the case of flowers will also change the result…
Here are 4 exports of my edit… First one is perceptual with DT srgb profile and the second is relative…they seem the same and the next two are with the icc from the icc consortium…First one perceptual which is a good match for the 2 from DT and the last is rendered as relative and I actually like how it renders in this case… IMG_7089.CR2.xmp (9.1 KB)
Todd, they all look about the same to me. You embedded the relevant profile in each (proper practice), and most browsers are going to convert the image to sRGB (default) for rendition.
They have another one on the same page called appearance… this one is supposed to help preserve color when there are color space conversions at least I think this is what it is for… its gives a slightly different result not sure I fully understand the difference between those two and this newer iccmax yet… Might have to read a bit to fully understand it…
IMG_7089.CR2.xmp (20.7 KB)
I tweaked color calibration until the flowers became red. I had to drop saturation afterwards. filmic v6, with preserve chorminance in luminance Y mode.
Halina3000,
Further to my previous comment, I’ve also noticed that camera lenses can differ considerably in how well they convey colours at the red end of the spectrum. I suspect that some lenses filter out some light frequencies to some extent. So it might be worth trying a different lens or two, and comparing the results, if you can’t resolve the issue to your satisfaction in post-processing.
I’m curious about that, having assumed my entire lens collection is “achromatic”. If you have a couple of example shots I’d be interested in examining them.
Bright luminous saturated reds are also one of the problem areas for filmic v6/v7 gamut mapping.
People switching to sigmoid or filmicv5-no-preserve for those kind of situations.
Personally i set it up with filmic, but then render out to a rif in rec2020. This disables filmics gamut mapping basically , and give better results in those cases. It’s a preference thing though. And also one of the issues i have with Darktable , is that there is no way to control the gamut mapping. So i akways get a magenta look in DT, while the export to rec2020 has no magenta cast in those reds.
The football team of my son plays in bright pure ree shiny shirts (50% red, 50% white pattern) .
In rendering pictures of that in DxO with a profile from Color Fidelity , i get good reds without issue. Rawtherapee with the same profile and a simple logencoding for tone mapping gives very pleasing results , and the same looking reds.
Even with the same profile converted to ICC with dcamprof, i have a hard time to get Darktable render the reds in a pleasing way. Even sigmoid and filmic5.
Filmic6/7 but exported to rec2020 are only ones being close (with the regular standard matrix in DT btw).
Bright, saturated reds just seem like a DT problem at the moment to me. I haven’t tried a base curve edit to be honest , who knows.
Red flowers do have some green in real life but over-saturation pushes the captured green channel down to zero. So the flowers in the posted image are only bi-chromatic e.g. red and blue w/no green. Purple casts can often appear - a sure sign.
Contrary to popular belief, shaded colors are more saturated than brightly-lit.
One way to over-saturate an output is to edit in PhotoPro or Adobe color space to taste and then post straight to sRGB for the web: gamut- clipping can occur.
Still interested in the lens subtopic. I still think modern lenses are largely achromatic, but I do realize there isn’t perfectly uniform power distribution through any one of them. Simple to measure, just shine a full-spectrum light through one and take its SPD measurement, then compare it to the reference measurement taken directly off the light. Oh, another application that needs that i1Studio I never got around to buying…