Those who are familiar with Davinci Resolve know that the processing approach in that program is based on a node structure.
The idea is to train the tools to work with lightness by angle of influence on the image only on a certain channel, if we are talking about encoding an image in the RGB paradigm.
What’s the idea. We only affect a certain channel of the image, but in “color” overlay mode. So the node responsible for adjusting lightness has no effect on color. In such a node, we leave only one channel in operation, and the point is that nothing happens to the node in this overlay mode. The contrast will not change until we start influencing it.
That is, we indicate to the tools that there are shadows and lights in the image by shifting the hue. So we are not affecting the channel that we left for the lightness node, but another one - so we can affect not the red channel, as an example, but the virtual purple one)) or any other one.
The color node is not neutered by turning off the rgb triad - everything is normal there. However, there we can compensate for the change of visual saturation and it does not prevent us from getting the result we want.
Sorry, I do not understand what you are aiming at. The video is also not really helpful for me to understand the purpose. If this works in DaVinci Resolve, why are you want to repeat it in darktable?
Maybe you could post a raw image and the result you got from DaVinci Resolve. Please also explain what kind of effects you tried to achieve in that image. Otherwise it is to theoretical for me and also, I do not use DaVinci Resolve and can therefore not repeat the procedure. As a proprietary software DaVinci Resolve is also not the focus of this forum.
If you mean applying parametric masks to such goodness, a mask is a mask. A mask simply overlaps a certain range of processing. And in my case, it affects the entire lightness curve of a particular channel.
I also really have no idea about what you are talking about sorry by channel do you mean the r g and b channels and are you talking about alpha channels as well?? Sorry just no real idea exactly…
Hopefully someone else can dissect this
EDIT…it seems you are looking at working in an environment like this right…
You have no nodes and and tools like this in DT …just a linear string of modules. You can add multiple instances and move the order from default but that is it…each module will take the output from the previous module as its input…
This is the framework you have to work with but nothing like what you show in this video example
no, it’s not possible in general since darktable isn’t node based so there’s no 1:1 match - you need to find a solution the darktable way using:
blend modes: can be also done per r/g/b channel, luminance or chrominance
if you just want to affect color, then select chrominance etc., if you just want to affect luminance, the select luminance as blend mode, if you just want to affect a certain channel, then select that.
parametric masks: restricts the effect depending on e/g/b channels, Luminance, Chrominance or hue
if you just want to influence purple pixels, then mask for purple hue, if you want to affect low chrominance areas, the set the chrominance sliders, and it’s similar for luminance, r/g/b channels. You can combine the different parameters
multi instance editing: you can use several instances with independent masks or blend modes. In difference to davinci these can’t be parallelized so the output of the former instance is the input of the latter.
There’s just an option to reuse a mask defined former in the process chain so you can affect the same or the other pixels - in combination with blend modes but not in an combination with further parameters
(e.g. 20220514-IMG_4731.CR3.xmp.txt (8.1 KB) boosting greens first then desaturate using the inverted mask)
that’s the only way to get something like a parallelized processing
Thank you.
I’ll try to think about it. In Davinci, I managed to put together this combination. Maybe I will solve this issue in my own way here, or even go another way. The important thing is the result itself, and not the task to do such a thing in itself.