IMHO, something that is still missing in Darktable and is hindering a more widespread use is a concise, clear guide to a principled approach to basic raw development. I have watched quite a few video from Bruce Williams and Aurelien Pierre, but I still feel I don’t quite have a sure footing about how to proceed in general. I understand that in DT you can do similar things in many different ways, but that is not too helpful unless you understand in depth all the details of the different procedures (and their combination).
Aurelien did publish a quick guide that addresses this aspect a while ago, but it is written for a previous version of filmic. Would it be possible to update it for the new version?
Perhaps I should clarify better my point by illustrating where in the workflow I get a bit stumped:
import an image into DT using the new scene-referred workflow (ok, here!)
use the Exposure module to adjust the overall exposure…. but according to which criteria? I can think about at least two alternatives here:
a) move the exposure slider so that the average image exposure looks right. I understand that the default values are conceived for average exposures using matrix metering in the camera, but if one uses manual exposure (eg, for ETTR), the defaults won’t work well. So, for example, if I have exposed the picture taking care not to blow highlights, by raising the exposure slider in this module to a reasonable middle grey value, the highlights are going to clip now.
b) move the exposure slider taking care not to clip highlights (toggling on overexposure indication): I think, though, that Aurelien advised against this, saying that highlight management will be done later on in the Filmic module.
And what about Black Level Correction? Should I adjust that so to avoid clipping or should I pick a value that just “looks good” for the image?
The problem with this is that, whichever one of the above strategies I use — even when the image has no clipped highlights or blacks in raw — I find it quite difficult to adjust the parameters in Filmic to get an image with nice contrast and no clipped highlights or blacks. (I have to say that I go back and forth moving sliders by trial and error, and hence my plea for a principled approach).
Now, if the purpose of Filmic is simply to do a nice, natural tone mapping, why couldn’t Exposure simply set the black and white point automatically according do the image data (min and max intensity values) and then use Filmic to add contrast without clipping extremes? Indeed, shouldn’t a curve like Filmic’s, which is anchored to the max and min point, never produce clipped values barring floating point precision limits (i.e., it should go asymptotically towards the clipping values)? What I see instead, that as I adjust contrast (or other sliders in Filmic) in order to regain some contrast, it is very easy to clip both highlights and blacks.
And while we’re at it, a final question… (sorry for the lengthy post :-). If there are indeed highlights that are really clipped in the image (= clipped in raw), couldn’t the Reconstruction part of Filmic automatically set a mask for those pixels? That would be another step towards simplification.
I don’t mean at all to criticise Aurelien and the all the others DT developers who are doing an outstanding job, for which I am deeply grateful. I just feel that some clarification in the sense described above would greatly help in increasing the adoption of DT. I myself am using Lightroom (last standalone version) as my main software, and would really like to switch to DT as I can see its better potential, but as of today I am still unable to match LR results in a reasonable amount of time. BTW, if somebody was willing to write such a principled guide to DT workflow, I, for one, would be happy to buy it.