I noticed this harsh highlight when the photo is processed with Darktable which wasn’t exist in the original JPEG. As soon as I open the raw file in darkroom, or simply when I exported the JPEG from the light table, the harsh white appears.
Edit: Looks like that spot is a completely blown highlight - while it is possible to make the transition less harsh you won’t be able to get rid of it entirely. If you post it as a ‘play raw’ and license it appropriately, the community can try editing it and give you some pointers on how to do this.
Not sure it’s the best approach but I’ve started experimenting with a masked instance of the lowpass module to remove the harshness of the transition on an overexposed patch:
Thank you. Now I know how to check for the clipping warning.
Since the OOC JPEG looks fine, I thought there’s nothing “wrong” with the raw…
My DT crashed when opening your XMP. I noticed your xmp version was v4 while mine is v3, don’t know if it matters. Anyway I’ll try to edit using FilmicRGB and other modules suggested by @Nilvus.
You can get a great looking image even with blown out over exposed areas. People including some of the best do it all the time.
Your issue is with the highlight roll off. The camera jpeg is brighter and has less contrast between the area near the bright spot and the blow out zone itself.
In your case you are well above base ISO already so the easiest solution for the future would be to just lower it a bit while keeping the other variables the same. This shouldn’t result in much more noise but makes the result easier to work with.
Thank you @Jonas_Wagner for your pointer to Aurelien’s article, very interesting and I will keep this in mind.
Regarding the ISO, I actually used auto ISO most of the time due to quick changes in the scene. As you can see I’m photographing bird, I want to be able to easily switch from perched bird in dark place or flying bird on a bright sky. Matrix metering with auto ISO and a little bit exposure compensation is very handy. But sometime I’m just not quick enough to adjust the exposure compensation…
Anyway in my example photo which is not the greatest photo afterall, I wasn’t really interested in the lost detail in the clipped region. It was just an empty background. If I can make the transition a bit smoother as in the JPEG, it would be good enough.
The main problem I had in getting the clipped patch of sky to look natural is that the clipped highlight handling assumes that clipped areas are white (R=G=B), since the clipped area here is sky and isn’t white I had to cheat a bit.
Set white balance so that the sky near the clipped patch comes out as grey (this makes the whole image look rather yellow)
Use ‘clip’ highlight mode (‘blend’ may be better, but it introduces a magenta cast around the edge of the clipped region)
Add a Colour Adjustment node and scale R, G, B so that the white balance is correct (I assumed the birds neck was neutral)
Add my current default processing workflow (which is a variant on filmic: a colour adjustment node to implement a power function, then a tone mapping to roll off the highlights smoothly, then a curve to add a toe for shadow contrast)
Here is what I could achieve in a couple of steps… I have used Photoflow, but probably this can be reproduced in DT as well.
I have used the “blend” highlights reconstruction mode, and added a simple tone mapping curve that preserves the shadows/midtones and flattens the highlights, mapping input=2 (white level = +1EV) to output=1:
