color eq slider names about darktable HOT 27 CLOSED

An alternate idea is to have unequal spacing between points. That would allow to merge 'azure' and 'cyan' (which is currently very narrow) into one, and split 'blue' from 'lavender' into two. It would still be eight points, just spaced differently.

Nope for sure. Has been discussed in the original PR and AP's arguments are absolutely valid on this.
And, a) nodes are all 45°-spaced and b) guided filter depends on this.

from darktable.

Hello,

Very nice analysis, you beat me to it ;-).
I made a few test patterns to better visualize the colors and the action of the various settings.
I found that the problem lies more in the distribution of colors. I find the angle of the colors turquoise and blue too narrow and the angle of the color lavander too wide.

You'll find the various test images here https://www.dropbox.com/scl/fi/awwnbhpg4ux2jnglncbyf/Mire-HSL.zip?rlkey=orsshzbhmi1756ntngckfqueu&dl=0

@jenshannoschwalm @TurboGit You are doing a magnificent job of devolleppment, thank you.

Greetings from Brussels.
Christian

from darktable.

Revision of above comment:

1. purple -> magenta? yes
2. turqoise -> cyan? yes
3. blue ->azure ? yes

from darktable.

Honestly I don't care what we call the colours, and I'd much prefer not to display those sliders at all. I now know I can hide them but the fact we've chosen to have that (hide) option suggests to me that we suspect they might be pointless. Same goes for the sliders in the tone equalizer. Do they add anything? No

Another advantage of removing the names is that we could do away with the fixed nodes. I understand we need to keep the nodes equidistant but if we could shift them all left/right at the same time to place a colour of interest directly under a node that would be great.

from darktable.

I have strong opinion to keep the sliders. The current interface should be fine for everyone with or without.

from darktable.

I think there is no reason to discuss further. We'll keep them, enough people want them and it doesn't hurt ...

from darktable.

In general i agree on "some names are not good". Just took those from original author.

1. purple -> magenta? yes
2. turqoise -> cyan? yes
3. all blue changes - not sure

from darktable.

@jenshannoschwalm Yes those are the most obvious two.

The name teal is certainly debatable, as many different sites seem to have a different opinions on what teal is. My reasoning was that rgb 255 128 0 is orange. Teal and orange is a famous complementary harmony, and the opposite of orange is rgb 0 128 255, thus should be teal. However, according to wikipedia the name teal comes from a bird with that colour on its head, which does look greenish, closer to turqoise, and therefore teal wouldn't be a suitable name for the range of this slider.

The name most commonly found for that rgb trio is azure, so that one seems better.

If you don't change lavender it might not be necessary to change blue to teal/azure, however it doesnt seem correct to name a slider after a primary colour it doesnt contain.

Green slider range also contains turqoise and we name it after the primary, so it would be consistent to do the same for blue/lavender and call it blue, but I can understand the hesitancy, since purple looks more distinct from blue than turqoise does from green.

EDIT: If you were to stick with 'lavender', 'violet' may actually be a better name. Lavender RGB is 230, 230, 250 or 181, 126, 220 (floral). Contains a lot of green because of its lightness. Violet is 127, 0, 255. Purple is similar at 106, 13, 173.

from darktable.

Names could align with those often used to illustrated the names used with mixing as noted in this document about the channel mixer...
The Channel Mixer.pdf

from darktable.

3. all blue changes - not sure

Why does the blue slider not act on a range of hues around the blue primary as red and green do around their respective locations ... its very clear in the vector scope that it is quite different... I think at first I thought it was maybe a color space things but I changed a few and that didn't seem to make a huge difference

from darktable.

@todd-prior that's a document I put together. As above, I'm thinking now azure (instead of teal) is a more accurate and established name for that rgb colour. You're right in that it doesn't effect the blue primary.

from darktable.

from darktable.

An alternate idea is to have unequal spacing between points. That would allow to merge 'azure' and 'cyan' (which is currently very narrow) into one, and split 'blue' from 'lavender' into two. It would still be eight points, just spaced differently.

from darktable.

@TurboGit @s7habo @elstoc, any opinion on this?

from darktable.

purple -> magenta? yes
turqoise -> cyan? yes

Yes for both. I would keep blue.

I had also thought about Lime -> Yellow, I have no strong opinion.

from darktable.

I don't have a strong opinion on this either.

You can see which colors are influenced in the background. I think this visualization is crucial. Besides, this is not a module with which you should precisely cover something like pantone color chart but is rather intended for rudimentary changes.

So for me, everything you agree on fits.

from darktable.

Nope for sure. Has been discussed in the original PR and AP's arguments are absolutely valid on this. And, a) nodes are all 45°-spaced and b) guided filter depends on this.

Would you have the number or link of the orginal PR... I'll try to find it but in case I don't...no rush and still I feel like unless there is some color space mapping that puts "blue" where blue is that it act barely on blue and so given that red and green seem fairly aligned how can blue come to land where it does .... maybe the PR explains it and if so don't bother wasting your time on me....

from darktable.

@elstoc i agree with you on removing the sliders.

from darktable.

from darktable.

Personally, I like the current interface, and the fact that you can choose to show or hide the sliders. I find it more user-friendly to work with the sliders, which allows me to make finer adjustments (same goes for the tone equalizer). I admit it's all very subjective.

As for the color names, I'd keep the blue name even though the node isn't positioned on the primary color (240°).

from darktable.

I don't have a strong opinion about keeping or removing sliders, but even if they are removed, the name of the hue point still shows up in the pop up label when you hover over, and will presumably be in the manual too, so is still relevant.

Although I think cyan and magenta are the two most obvious names to change, my main motivation to look into this was to see if some of the confusion people had around why the blue slider didn't affect blue could be resolved by more accurate naming. If the names of blue and lavender remain unchanged, that issue still remains. I see two other solutions:

1. Removing the sliders (and also colour names from pop ups and manual). Since the graph shows colours, many people wouldn't have a problem with that, though it wouldn't be very helpful for the colour blind.
2. Put colour charts like those in the original post, showing which range each slider effects, in the manual. Not as immediately discoverable, but at least its documented.

I understand we need to keep the nodes equidistant but if we could shift them all left/right at the same time to place a colour of interest directly under a node that would be great.

If that were to happen I guess they would be named numerically, 1 to 8.

from darktable.

from darktable.

Hello,
Yesterday, I used the module for a few photos with good, easy-to-get results :-).
As for the names, here's my proposal, taking into consideration the various proposals quoted above and the French translation.

1 red (Fr = Rouge): unchanged
2 Orange : unchanged
3 Lime (Fr = Vert citron) : Lemon (Fr = Citron)
4 Green (Fr = Vert) : unchanged
5 Turqoise (Fr = Turquoise): unchanged or Cyan
6 Blue (Fr = Bleu): unchanged
7 Lavander (Fr = Lavande) : unchanged
8 Purple (fr = Violet) : unchanged or Magenta

In conclusion, in the end I'd only change lime to lemon, what makes me hesitate for cyan and magenta is that they're well-defined colors (0,255,255) and (255,0,255). I'd give a preference to more subjective color names.
Best regards,
Christian

from darktable.

@elstoc i agree with you on removing the sliders.

removing the sliders needs to make the nodes in the graph accessible for shortcuts. So the current show/hide solution is quite more useful

from darktable.

We could just hide them from presentation....

from darktable.

I think the sliders can be handy really. If I know I am going to say target orange then I will just go to that slider.... drag it until i am close to what I want ...then I often double click and leave my mouse in place and then click. Doing this I can go back and forth and guage the effect.. In a similar way if I add one or two more small tweaks I can zero either or both of them by double clicking the slider.... If I have only the graph I have to drag the node back to zero being more careful and with more ergonomic effort. I think if if you take the sliders away it would be nice to have a way to reset one node. I think if you double click now on the graph or a node it resets the whole curve does it not.... If I am behind development in my understanding then ignore me

from darktable.

Closing this for now. UI has some significant progress so "naming" should be discussed in another FR if new arguments are coming ...

from darktable.