I seem to be hitting issues here and there lately with RT - this is another one.
I was working on the raw uploaded in my earlier post tonight - http://filebin.net/afu7qh7bk1 - and adjusting with the CIE controls. I decided to do a custom curve for brightness, and noticed the point at the top was having little or no effect. I was adjusting it along the top of the graph control; but no response. I made increasingly severe changes to the top point, ending up as per the screenshot in the upload “…strange CIE curve…”. This should surely have produced an awful image, but didn’t! Any thoughts please?
Andrew
I tried to reproduce your CIECAM settings on a raw file, starting form the neutral processing profile. On my machin, with the latest Master build I made on linux (4.2.788) it works as expected, the rightmost control point does have the required strong action.
I’m gonna try compiling the Gtk3 branch, as it is a build from this branch you’re using.
Thanks for having a look at this Sebastien. I went back into this photo to get some more details to help your investigations, and found several things were switched on which I wasn’t aware of, e.g. there were some Lab settings. I’ve switched all these off but the tone curve is still misbehaving. I’ve uploaded some more files for this photo.
Jpeg ending “-A” - this is what I saw when I went back into RT, strange histogram on the right.
After switching off unnecessary adjustments:-
-B - noticed that with the last point at this height, it was making a difference to the image, see the trees on the horizon on the right. But of course much of the image should be dark with this curve, yes?
-C - this second point at the top is at a boundary, it seems to me. To the left of the position shown, parts of the image go white, rather as you’d expect. But this stops from the position shown. And in any case, much of the image should be white with this curve.
The .pp3 confirms the settings I was using.
The files are here: http://filebin.net/grbufg47jp
regards, Andrew
@RawConvert, regarding the B image, what happens, I think, is that your curve is turning the highlights darker (whites become grey). But I do think your curves are too “strange” or “extreme” and bring results that are hard to predict.
I also tried with the Gtk3 build (4.2.915) and it looks correct to me, the curve behaves normally as long as I use less extreme settings.
I too think CIECAM02 curves behave strange (sometimes the right half of the curve seems to have no effect, also the curve doesn’t seem to match the histogram behind it), but I first must read this:
http://rawpedia.rawtherapee.com/CIECAM02/fr
1 Like
Sebastien, I agree example B is extreme, but C-style is quite reasonable in trying to use the full range of tones - but finding the point at the top behaves oddly and is inconsistent with the image, as Morgan says also.
Morgan, I can’t read French! - but there is the English version of course. I just tried ticking “output histograms” and this gives a different histogram, also some further odd behaviour as shown below. If you slide the bottom point to the right bit by bit, once you get to the histogram a chunk breaks away and forms its own spike:-
All this is not a big deal of course, the process pretty much works, and I take my hat of to those who understand the principles involved, the maths, and those who built it. I can’t really say in what way it looks different from say Lab processing, but from the few photos I’ve used it on so far, it has given very pleasing results. Andrew.
I mean hat off …