Auto-Matched Tone Curve is a cool feature, but I have noticed that sometimes it fails in a rather spectacular way, yielding images that look really bad.
It happened to me (again) yesterday: the embedded jpeg was OK but the RT image had too much contrast and bad colors.
This time I was not able to get a good image with any adjustment so I looked at my old copy of ACR and it worked well too.
To make a long story short I downloaded the default Adobe tone curve from Rawpedia and I assigned it to the image: it worked and now I have a good image to work with.
This leads me to a couple of thoughts.
1 - Not everybody will feel comfortable with having to download a tone curve file and save it to the correct folder. It would be nice to have it installed by RawTherapee in a default curves folder.
2 - Since Auto-Matched Tone Curve is a default feature it is not easy to understand that it is the culprit for rare bad results.
Now in the Exposure tab there is a large button labeled āAuto-Matched Tone Curveā: IMHO it would be nice to put there a couple of buttons, one with the Auto-Matched Tone Curve feature, the other to load the Adobe standard curve. It would make it simple to use it and it would be easy to think about it when auto-matched does not work well.
@agriggio is the author of the auto-matched curve algo you see in RT now. He develops his own fork of RT called ART.
I reported aut-matching issues in RawTherapee and ART.
As you can see in the ART report, Alberto thinks that auto-matching is a difficult task and there will always be cases where it fails (and it will never be perfect).
Usually the āStandard Film Curveā profiles come quite close so in case I have severe issues I switch to that and with slight tweaking of the curve/brightness/contrast I get good results.
In the end if Iām happy with the JPEG I use that. When I edit the RAW file I usually tweak it a little bit more so starting from jpeg look or something that isnāt that far off doesnāt matter too much.
Adding another button IMO clutters the user interface too much for the little benefit it offers. But other users/devs might see this differently.
I see your point here and I donāt have a really good solution for that.
Having some sort of back-up curve that is applied instead of a āfailedā auto-matched curve is highly subjective (why would Adobeās curve be the way to go?). The whole point is that we donāt want to force any vendor specific flavouring to your raw files if it is not necessary.
If you really like the curve from your cameraās JPEGs, you can always try to copy and paste the curve from another JPEG where the matching works better. See here on how to copy/paste: Exposure - RawPedia
The Adobe curve is just an example of a curve that can be expected to yield decent results. It could also be a different one as long as it solves the problem.
Anyway it would be just a starting point for further editing.
Thank you for your suggestion, but I have already lost some time on this issue and I have solved it.
I was thinking of new users that maybe are not particularly technically minded. An auto matched mode failure could be difficult to diagnose so maybe it would be nice to have a simple way to override it. Just having such a way might help in understanding where the problem lies.
Without providing an example (a raw file where it fails or at least a screenshot of the curve), noone will be able to improve the auto-matched tone curve.
Anyway, I realize that tone matching cannot be perfect so there will always be cases where it fails. I was wondering about a better way to handle the failure for inexperienced users.
I donāt know if a tone curve can be copyrighted, but anyway it ought be enough to ship a similar curve without any reference to Adobe. The idea is having a simple falloff when auto matching fails.