@s7habo, I’ve finally finished the latest video and your series on Sigmoid and contrasts. What a fantastic set of videos! Thanks so much for doing these and revealing your editing secrets. I always learn something new from your videos, even when I think I’ve mastered some part of Darktable. I have already incorporated your trick of doing an linear invert in Tone EQ for images with high dynamic range, and then using another instance to selectively add back contrast exactly where you want it. It’s also great to see the Color Lookup Table get some love, because I think this is a really underrated module.
And thanks for making something out of a couple of my rejected photos. I thought they might be useful for practising with contrast, and that certainly seemed to be the case. I was particularly impressed with how much detail you got out of the snowy mountain picture. My only gripe was that you didn’t crop the railway shot to make it more symmetrical, thus highlighting my bad composition, haha
Hi! Perhaps a little naif question… Is it possible to achieve nice contrast in both highlights and shadows? For instance in this play raw https://discuss.pixls.us/t/st-monans-salt-pans
Thank you for the gradient. I have downloaded it in order to reproduce some of the editing in @s7habo videos. But it turns out not to be linear (screenshot below) or am I doing something wrong?
You are right.
Setting the input color profile to sRGB made the gradient histogram linear. Thank you.
Input profile is something I rarely (never) change or think about.
In which situations is it necessary to manually set the profile?
Mostly if you are working with synthetic or test images you might need to and of course if you use a color checker and calibrate your camera to make a custom icc profile then this is where you would specify it…
@s7habo Another great and useful post. Many thanks. I think you obliquely referred to AgX. I wait for it in anticipation of another very useful and informative post
@s7habo Thanks for another great tutorial, Boris. Colour Lookup Table is one of my favourite modules, so I’m really happy you gave it some love on your channel.
A while ago, I put in a feature request to convert the module to unbounded RGB to try and make it more robust in the scene-referred workflow. But apparently it was too complicated. I actually don’t mind too much that the sliders are LAB, but I still think the module could benefit from some updates. So I was wondering if you had any thoughts for possible tweaks.
One idea I had was to simply make the sliders less sensitive to adjustments. Currently, tiny adjustments can have a huge effect, so I’m wondering if tweaking the math for the sliders could make them a bit less sensitive.
Another nice feature to have would be some kind of visualization to see what is selected when you select/create a patch. I’ve no idea if its feasible, but I think that would be helpful.
I also find the changes you have listed here very desirable. However, the question is who could make them. Unfortunately, I have no development skills whatsoever to tackle this myself.
The only way is the one you have already taken—to make a request for each of the points and hope that someone will be able and willing to implement them.
Yes, there is the slight issue of who could do them I don’t have the skills either, but I have found many developers to be quite responsive on Github. I put in 3 feature requests before the last stable release (including side-by-side snapshots), and all three of them were accepted by various developers, including two students from Paris who wanted to learn through contributing.
But I never submit a feature request expecting it to be implemented. Maybe one day I will have the time to teach myself these things, but until then, I hope that my requests are small contributions to improving the software for everyone.
I’ll submit the ideas and see if there are any takers. No harm in that!
I think as well if you look and many color charts and the math around calibration using them that it is often using LAB for reference values and the connection space in profiles. I think the history of the module is linked to that as well to DT chart for making jpg matching or color calibrated styles.
This might to a degree lock it in and what you really might have to ask is could what I am missing from the color eq wrt how I use the CLUT module be somehow added or derived from that…
What I like about CLUT so much is the interface. I love that you have all those patches, which you can use as-is, but also just add your own with just a couple of clicks. And then on the same “page”, you can adjust the hue, saturation and lightness of that selected colour. It’s very neat, and I find it very user-friendly in this regard.
But with regard to its functionality, Darktable has so many tools that you can usually find multiple ways to do the same thing. So, yes, I can mostly replicate what CLUT does elsewhere, such as with Color EQ and Color Balance RGB.
With the latter, you would need multiple instances and parametric masks to isolate the various colours, so this is obviously much more cumbersome. But it’s fine for 1 or 2 colours.
With Color EQ, I think it mostly replicates CLUT, and arguably it’s better suited to the modern Darktable workflow. I think why I still like to use CLUT mostly comes down to its ease of use. While I understand the design choices of Color EQ, I’m not a big fan of the multiple tabs, and each tab works quite independently. For example, the “hue curve” feature to restrict/expand the hue selection does not apply to the saturation or brightness tabs. So I simply find I’m doing a lot more clicks with that module.
But this is in no way a complaint. It’s great that we have so many choices and options. I’m just a big fan of the interface of CLUT!
I like the chosen theme very much. The hare part I consider a bit of magic as I would not have had much idea what to do with this picture myself.
Generally I’m fond of your video’s because quite often you elaborate on how you analyze the picture and then improve on that. For me your video’s strike a very nice balance between analysis and how to use darktable.
Good to see you introducing the input sharpening. I started using it and like it very much.
Capture sharpening is a great tool, and I am enthusiastic about its addition. However, most cameras and lenses from the last 10 years are super sharp already for most applications.
Our eyes are drawn to sharp, contrasty, bright, and colorful objects, so outside the subject, sharpness is a distraction. In practice, if capture sharpening is used, one will need to tone down the sharpness (either with contrast eq or D&S) outside the subject with a mask later in the pipeline.
This is also the general principle behind the latest excellent video in this series: darkness emphasizes light. Just like grey emphasizes saturated colors, blurry areas emphasize sharp areas, etc.
I wait for it in anticipation of another very useful and informative post
I don’t have the skills either, but I have found many developers to be quite responsive on Github. I put in 3 feature requests before the last stable release (including side-by-side snapshots), and all three of them were accepted by various developers, including two students from Paris who wanted to learn through contributing.