I’m trying to ‘correct’ a couple of rolls of slides/diapositive film which were exposed in the 1960 but which appear to have been poorly developed (they all have a marked purple colour cast). They have been scanned to tiffs using Vuescan and a Nikon Coolscan V, but possibly not with the optimum settings. In darkroom mode in dt I have toggled the clipping indication on and set it to show ‘saturation only’. In the slide I am editing as I type this, about 65% of the image is the profile of a large building and is glowing bright red. The remainder of the image (mostly sky) does not show any 'over saturation…
What is confusing me is that the majority of the building is in shadow, some of it quite deep, but not clipped (setting the clipping indication to ‘luminance only’ does not show any blue or red pixels). I did not expect it to be clipped in saturation . I cannot seem to find a module to control saturation when using scene-referred modules, so I have resorted to using ‘contrast brightness saturation’ - which is display-referred. But, very strangely, if I decrease the saturation (move slider to left in that module) then more of the building and then the sky starts to glow red. This seems to be completely the reverse of what I expect. Furthermore, there appears to be no way to eliminate the clipped saturation.
What exactly are the bright red pixels showing? Is there the implication that my original scan is the cause of the problem ?
There are several modules that can affect color saturation. If you use developer version of darktable, I recommend color balance rgb module. Also old color balance module is good for this. Furthermore in the color calibration module in the tab colorfulness you also have the possibility to desaturate the single color channels. I will also use color zones module for this. With this you can desaturate single colors.
You can perhaps check whether you have selected an inappropriate profile for display, histogram or softproofing (darkroom view, lower right, gamut icon, right click). Just a thought…
Yes, of course it does; thank you.
The sky, road and grass are now properly exposed, with far better control of saturation and luminance than I achieved. The problems arise when I try to correct the obvious under-exposure in the profile of the buildings. I can select those parts of the image quite well with a parametric mask, within an incidence of the exposure module, but when I increase the exposure there are great colour shifts in the appearance of the main building, especially in the darkest parts of the tower.
To be more specific: the building is faced with sand coloured limestone - a very light coloured stone. Admittedly, 52 years ago this building was still heavily coated in carbon deposits, but it should be much lighter than it appears - by at least 1.5 to 2 EV, I guess. Increasing exposure by as little as 0.1 EV starts to cause a colour shift back to the original purple colour cast, while doing little to increase the luminance. How do I manage that problem?
Is it valid to mix display and scene referred modules in working o one image? Does this matter for an image taken with an analog camera?
of course you can mix the modules, you even can change the order of the modules if it’s useful for your purpose - this doesn’t kill kittens, you just can’t blame darktable in case there’re side effects messing up your edit. But even then everything is reversible.
Try an error is a valid approach to become familiar with darktable
Move the exposure module after the rgb curve , another dirty trick is to move the color calibration module at the end of the pipe( but before the output color space conversion) and use the gamut compression for reduce the very saturated colors
For shadows try the relight preset in the tone eq…if it’s too much don’t try to tweak it just use opacity to pull it back…I find this often does a very nice controlled lift of the dhadows
The image has blue shadows and yellow highlights, I wasn’t able to directly neutralize shadows so I’ve modified my strategy and used two eyedroppers in the sky and in the gray street, this was the started point
Yes, it certainly is. I’m going to have to study the history stack more to understand what has been done.
What fascinates me is that the image still looks like it needs some work, even after some pretty good corrective work has been applied. Does the fact that it is still not a completely accurate color rendition suggest that the original scan tif needs to be improved?
