First of all - brief introduction. I’m quite new to DT.
My background is on years upon years of Lightroom CC, which I binned when I moved all my software to FOSS a year ago.
After moving to Linux, I used RT for a while, but it never felt quite ‘pro’. Yeah, silly argument I know. But knowing that DT was out there and there being a good consensus around it being more powerful, I felt compelled to get myself to learn it.
I used a Fuji X-E2 for the longest time, which seemed to always turn out well with minor editing. I recently got back into Canon with an EOS RP (had been a 5D2 user before the Fuji), but I am struggling quite badly to get the look I’m after (and which has sort of become my de-facto).
One of the important parts of it is that I push the histogram to the right and leave a void on the far left. Whenever I try to achieve this with DT it’ll always leave a trace of something.
The way I understand it is that the black level correction is merely biasing the expsure, but there is still information from the raw file that shows up after pushing.
I’m trying to achieve the same result one would get from completely clipping out part of the histogram. I’m looking for a sharp removal of the lower section, without leaving anything.
This is how my current histogram looks like after trying my best:
The issue is not so much with the resulting exposure - it’s with the colour cast that’s left behind.
If you think I’m doing something fundamentally wrong, please do let me know as I feel I’m not quite understanding the flow I should be following. My go-to reference has been this bit of text so far: darktable user manual - process
if just one channel clips then this will result in a colour cast in the highlights since the non clipped channels are kept by default.
You can change that quite early in the pixelpipe by setting a custom behaviour in highlight reconstruction module. But better do this at the end of the pixelpipe in the tonemapper when you’re sure you don’t need the information in the unclipped channels any more …
The issue might also be your white balance, but it’s (as usual) hard to be definite without a sample image.
And if you indeed want hightlights clipped, you may want to try filmic, instead of sigmoid: with sigmoid, the output shouldn’t reach 1.0, with filmic you can get clipping in the highlights by lowering the white reference (and under “optionts”, select “safe” for the “contrast in highlights”, perhaps).
I think I didn’t explain myself properly - it’s the shadows what I want to clip. It’s merely an artistic interpretation - I find too much detail on the shadows distracting, so I tend to systemmatically clip the darkest tones out and lift the black point to a very dark gray instead.
I hope this also explains why I haven’t attached any raw to the post - it’s not particular to any image, it’s more of a workflow-related aspect.
That would completely discard the lower end of the histogram and make it look like the screenshots on the OP - completely regardless of any other edits. This would remain constant throughout any other editing steps I take. Now I understand there’s differences in how DT and Lr/RT approach editing, but this is fundamentally what I want to do.
As you can see, black isn’t black entirely and there’s not much to see in the very dark areas of the shadows - entirely an artistic decision and not a technical one (I don’t pay too much attention to technical aspects of pictures - I edit entirely by feel).
The issue I have is that oftentimes a colour cast remains:
It is not really practical to me to go image by image removing that cast - clipping the tone curve should, in my mind, clip the tone curve. Neither RT or Lr exhibit this behaviour. The sample photo above was edited with Lr, but I can get pretty much identical results from RT with an editing preset I can apply to other pictures from that same journey without unwanted side-effects.
Thanks!
Edit: I just saw your post, @kofa . I’ll have a look and update accordingly, thank you! Mind, this message was written before yours so it’s not meant as a reply to you!
Hey @SuperGrip welcome to the forum. You have both tone curve and rgb curve available in darktable, why not use them in the same manner as you do in other editing software?
After trying what @kofa and @Donatzsky said, plus having a deeper look at the tone curve module, I believe I’ve found the issue. Selecting ‘none’ under ‘Preserve colours’ seems to do a far better job at it and not induce any casts:
The histogram still looks a bit funky compared to what I get from RT, but again - the look is the look, and that’s all fine. I am pretty sure I can work a bit more on the base curve to really dial in the look, but I’m fairly confident to be on the right track.
This is what I’ve settled on - please confirm I’m not breaking the laws of physics here:
If you mean the base curve module, then don’t. It’s a tone mapper (along with Filmic and Sigmoid) and only one of those should ever be active at the same time.
They want to clip the shadows completely before raising the black point, which target black doesn’t seem to do.
Without knowing what you use the base curve for, I can’t really answer that, except to say that curves aren’t used much, if at all, in darktable, since there are typically much better tools available.
I recommend you read this article:
And watch this tutorial to better understand the concepts behind the modern scene-referred workflow:
You can also manage your color cast with the 4vways tab…If you use the picker and select your shadows it will attempt to cancel out the cast…usually you can leave the hue and increase or decrease the correction by change the chroma…You can do that for mids and highlights as well… an overall cast on the image is often well corrected by doing this for the highlights…
This worked near enough perfectly without the need to touch the base curve, as per @Donatzsky 's recommendation. I’ll study better the post (just skimmed over it after dinner!) and try to figure out what might fit my case best.
Thanks everyone for so many options and taking the time to explain them!