R0011392_01.dng.xmp (11.4 KB)
With 2 tone eq instances (the compress – expand trick), plus local contrast on the highlights.
With 2 tone eq instances (the compress – expand trick), plus local contrast on the highlights.
Surely a naive question, but why do you need two instances? Isn’t this module linear, i.e. can’t you sum the two curves into a single curve to get the same result?
Not a naïve question. In this case the brightening exceeded 2 EV and I probably could have forced the hard limits, but I have found with many modules that multiple instances with smaller adjustments rather than a single sledgehammer instance produces less artifacts such as halos.
I personally like to use the shadow and highlights module but I will never brighten the shadows by more than 50% in a single instance. Applying a second instance results in no halo artefacts where one heavy instance produces halos.
With tone equalizer some users like one instance to focus on highlights and one instance to focus on shadows. Keep in mind than TE works on a mask and is prone to artefacts including halos if used heavy handed. Hence why I defer to multiple instances.
BTW, I recently started setting the tone equalizer adjustment and when happy exploring what happens if I change the “preserve details” masking option. I wish I had done that with this image as changing the first instance mask type produces a nicer result, while the same change on the second instance would result in halo artefacts. I no longer accept the default mask with no questions asked.
I also like RGB sum as the estimator for the mask and also sometimes the old guided filter mode can preserve more detail for some images
I hadn’t experimented with that until I just read your post and it makes another huge difference. Thanks for the tip.
Here is a video with a different use of two instances of tone equalizer from Jeddi-master Boris H:
Thanks a lot for the tips and answers, indeed the fact that it preserves the details via regions makes it non-linear.
Great video thanks. It also mentions color lookup table that works in LAB space, is it still recommended now, as I saw recommendations to avoid modules working in LAB space since, for instance, it cannot deal with large dynamic range?
As long as you don’t make huge modifications, the Lab-based modules that have not been deprecated and hidden, only available for old edits, are safe to use.
What exactly do you mean by that?
“RGB sum” is one of the lightness estimators in the masking tab of tone equaliser. As there is no one perfect way to convert an RGB value to a lightness value, you can select the one that best fits the image and your intentions. (and no, Lab doesn’t work here: you can be far outside the valid domain for L)
In any given color space, perceptual lightness is well-defined.
I would certainly hope so (or conversions would be even more complicated than they are now).
It’s also irrelevant in the context here: