Auto Input Levels in the GIMP is the quickest and easiest way I know for removing colour casts and generally improving colour contrast. As I mentioned above you can do it in RawTherapee by shifting the end points of the RGB curves but it is fiddly and time consuming and I generally find it easier to export to GIMP.
1 Like
My aim is not to contest the notion of ‘Levels’, I know perfectly well that this exists, but to emphasize the problems of vocabulary, habits, understanding… Some will say that I am repeating myself ( I am old), but my training in human sciences (in addition to that of science) has demonstrated to me several times the importance of these sources of incomprehension.
A series of words or concepts are confusing and without going into detail we can cite : black and white point, white balance, gamma, tone curves, levels, CAM (color appearance), profiles (input, output, working profiles, etc.), etc.
When we export the data to process it elsewhere - which I do not dispute - we change (except by choosing a linear gamma in output) 2 things: the primaries which are generally more reduced (sRGB or Adobe instead of Prophoto) and the Tone response curve which instead of being linear is replaced by most often an sRGB gamma, that is to say the connection of a straight line - slope=12.92 and a gamma=2.4. It is obvious that the responses will be different.
Another important point to note. When using the histogram, the default one is obtained with the output profile (screen or TIF output, etc.) which is with reduced primaries and an sRGB gamma. From my point of view, when processing the data, it is better to visualize in linear mode and in the working profile.
Another point is to make sure that the changes induced by ‘Levels’ take into account the real data, and are not bounded. If ‘Highlight reconstruction’ is enabled and Color Propagation is chosen, the real data can be in the range of 1000 to 200000, while the normal range is [0…65535], or something close to it.
In other words, it is necessary to differentiate the processing (and to use the Darktable language) - Scene referred and Display referred, or in CIECAM language - Scene condition and Viewing conditions.
When we talk about output, it’s a wish because I don’t know how to do it and I’m not equipped with an HDR monitor, it would also be necessary to take into account the value of Absolute Luminance of the output device (HDR PQ, etc.), a sort of improved TRC
Jacques