Ok I see, I didn’t understand what you meant, sorry…
but I’ll keep hiding histograms in RT as possible, I know it’s my fault 
My own personal opinion, darktable histograms seem no better to me, or better still even worse (if possible):
…but I haven’t a clue of what a waveform histogram is, I think I must do some homework.
I could explain but dt already has a lovely article.
1 Like
@afre thank you for the interesting link.
Well, had they been satisfied with histograms, there wouldn’t have been the need of any waveform I think. So I take some comfort from this, after all I’m not alone in criticizing histograms.
The fact is I like waveforms even less than I like histograms, and that’s really a problem. Should I switch back to film photography then? 
RawTherapee - Tonal distribution in a luminance histogram (linear with gamma correction):
To achieve real blacks and whites, both extremities next to clipping appear to be relevant. The most part of the x-axis refers to the interval -2/+2 EV, with the darkest tones here represented at -3 EV.
We can say that the -2/+2 EV interval (a small interval indeed compared to the real world luminance) is the descriptive part of the photograph, while the extremes are its glory.
But be careful to clipping.
Hi @geldo, I just read your remarks in this thread, but to be honest, what is the point you are trying to make?
I authored the change in RT to have a more flexible histogram with linear, log-linear and flexible log scaling (click and move your mouse left to right to see what I mean). These can be set to taste, depending on your intentions.
What is it that you look for in a histogram?
Hi @Thanatomanic , first of all congratulations for your work on the histogram, and for your final question, that has not an easy answer really. I’m just trying to learn RawTherapee, trying to get the most out it it, and we can safely assume I am in the initial stage of the learning process, still having myself a lot of doubts about the real usage of a raw converter. Moreover it takes me some time to write in English, for all this I can’t give you a thorough answer all at once (provided yours wasn’t just a rethoric question).
So to begin with your question, to me a histogram should be visually descriptive. I would say I would exclude a logarithmic y-axis which I think gives a wrong visual idea of pixel distribuition, and of course you can avoid a flattening of the remaining levels by truncating vertical spikes (the truncated spike is a good visual indicator too, you don’t really need to know the exact pixel count).
To beat you: But you can ignore it by using only linear scaling, is not a good point to me. 
Then there are other things, of course.