Somewhere along the line I have missed Aurélien’s suggestion regarding this. In fact, it has always been my my understanding that color balance RGB’s contrast would somehow ‘break’ filmic’s settings. I would much appreciate any pointers regarding this.
As with many things, investing time to master something is often rewarding. DT is more complex than LR for sure (not that much, but sure it is), but, for me anyway, it provides much better results on many fronts and additional controls on what is really happening behind the curtain. The “photographic community” is not homogeneous : many will choose to invest additional time to learn a tool like DT in order to get the benefits that come with it. Others wont, and this is perfectly OK.
1 Like
Well stated in the manual…you are correct that AP I believe had said not to use it for "global " contrast adjustments… use it more selectively… but if one can use the tool knowing how it behaves in a way that suits them then I guess it is fine…
From the manual…
" The contrast algorithm gives natural results that mimic the central part of the contrast curve of analog film. However, it will also increase the image’s dynamic range, which may void filmic settings in the pipe. For global contrast adjustments, you should normally use the tone equalizer module – the color balance RGB contrast slider is best used with masks, e.g. for selective corrections over the foreground or background."
Explained here at 45 min mark through to around 48 min…
5 Likes
I don’t get it. How do I apply these presets?
This is great, I can’t wait to try it out. Thank you for sharing!
Have you imported them… do that first if not then they are choices in the dropdown of the module that they are designed for…
Thank you so much for your input Todd, and for directing me to such a pertinent part of AP’s video.
AP says that color balance RGB’s contrast “is not recommended as a general correction”, which is how I remember it.
Aurélien does however describe the problem as coming from the tool shifting the white point, whereas the manual says it stems from increasing the dynamic range. Whilst I recognise that increasing the dynamic range will potentially shift the white point, for me, the distinction between the two descriptions has proven enlightening.
I typically photograph people illuminated by a window, or other single light source, and within darktable, I will usually aim to establish a fairly bright facial highlight, then aspire to tweak the image by adding contrast and grading the light as it falls off across the face.
What I hadn’t appreciated, was the degree to which my ‘tweaks’ were affecting the white point, which now explains lots of going around in circles!
I don’t think there is a difference.
If your color balance rgb pivots around some brightness level, it means values brighter than that will be made even brighter, values darker that that will be made even darker. That is increasing the dynamic range – so a pixel which was just below filmic’s white point (or just above the black point) will be shifted outside.
1 Like
Ah, I thought using Color balance RGB’s contrast (and adjusting filmic when needed) was the recommended approach.
I wanted to share what I had and what worked well for me.
Perhaps we should swap it out for Filmic RGB’s contrast?
Also, someone on Mastodon said it might be better to swap out diffuse and sharpen for local contrast, even though that’s a legacy module. The problem with diffuse or sharpen here in quick settings is that it’s useless and is intensive (especially if you don’t have good opencl support). Local contrast provides the “clarity” slider Lightroom users are used to, and it’s not too bad if done at smaller amounts (or masked, but the quick UI doesn’t support that either). I sometimes also use it even with diffuse or sharpen, as it provides local contrast a bit differently and more cheaply (on resources). Although, I usually use a mask to control it.
I guess a “version 2” could have these changes? What do you all think?
1 Like
Nor do I now, but I hadn’t previously grasped this.
It was never my intention to criticise your efforts. It was simply that I had never heard Aurélien say what you attributed to him.
As for filmic RGB’s contrast slider, I get the impression that Aurélien is not a particular fan of this either, even though he wrote the module. I once saw him describe it as “damage control”, which I wrote down at the time, but I cannot point you towards a reference.
I use local contrast liberally with masking and blend modes… I like the result. But as I often forgo filmic and sigmoid modules as I don’t need them I don’t have any worry about messing with filmic is usually not a problem. For many images to me there is no gain as you can often lose a lot of detail and then you just work to get it back. I find for the main edit I can use exposure, diffuse and sharpen, tone eq, color balance, local contrast and color calibration and often never need to use filmic… if things are blown out a bit I will then use it but even then I often can manage with a few tone eq instances…
1 Like
Define “better” 
I haven’t used filmic v7 yet, but I would encourage everyone to try out sigmoid and filmic to see which they prefer. I don’t think there’s an answer as to which is “better”. Filmic may give more vivid highlight colours, maybe even more accurate, but sigmoid creates a look that you might find much more pleasing (more filmic, ironically). And for newbies, it’s much more of a set-it-and-forget-it kind of module.
But thanks for writing this up. It’s always great when someone puts in the effort and time to help others out who may be having a hard time understanding the software.
3 Likes
Well done!
I also arrived at a “exposure only” workflow although I highly prefer sigmoid which in my opinion gives better results and with a straight forward interface, too. My default is at 1.2.
My look includes local contrast, contrast equalizer, tone equalizer and highlight reconstruction with inpaint opposed. All kinds of masking in the tools are turned off.
The cameras are set to underexpose around one stop to get that non-digital look going that the analog camera sensors should be providing and we have been denied for so long.
2 Likes
Almost exact same settings I have, just that on my camera the 1/3-2/3 underexposure is the sweet spot.
Regarding filmic rgb vs sigmoid: I don’t find either intrinsically “better”, in the sense that I can do everything I want with either. I just compensate in different places — darktable is so flexible.
I prefer sigmoid because it is less work to get the results I want.
2 Likes
I did not have the time yet, but with the mostly stable processing flow I now have (again), I will calibrate my main cameras using their inbuilt spotmeter (again) for those times when I want to be precise.
My compact Panasonic GF7 has a zebra-display that can be set in 5% increments. Set to 100% I get very close to perfect, might have to dial it a bit down to 95%.
I also love how Panasonic stuffs all these features into almost all of its cameras. Many manufacturers, eg Canon, start zebra at “enthusiast” models.
Lately I have been reconsidering my approach to exposure. After studying a lot of b&w film photos, I realized that in a lot of cases I actually prefer to have large amounts of background burn out to near-white (eg photo taken inside a not so bright room, including a window, that overlooks some irrelevant building or whatever).
In these cases sigmoid just shows some hint of details in the white, enough to make it non-uniform but not to draw the eye away from the main theme. The inpainting algorithms of Darktable 4.2 are more than sufficient for this purpose even when some channels are clipped.
Consequently, I no longer do ETTR unless I really want to bring back the detail from the highlights for whatever reason.
5 Likes
Same here, and also a recent realization. But I do struggle with getting the highlight rolloff as smooth as a modern color negative film.
One small step further … managed to do at least my main camera, so the following settings are very specific.
On the Nikon D500 using spotmetering the threshold seems to be between +2.67 and +3 EV overexposure when measuring the brightest areas that should still contain content.
This is for using RAW of course, highlight reconstruction “inpaint opposed” and sigmoid of 1.3 or less. This allows for pretty much arbitrary changes of the exposure setting in darktable with a smooth rolloff into the lights without blowing anything out while retaining color information up to the top end.
The highlight reconstruction does not have to kick in yet, but it is there for the details that one can not measure with the spotmeter of the camera.
Motives where single channels are extremly over represented - red flowers, anyone? - will need a much more conservative setting to preserve the color information.