I’ve played around with all kinds of combinations, and I could never get red at the top of the image to 255, and I could never get rid of a small amount of green, I did get it very low by zeroing green, but then the grass was red.
How do you improve on this?
You can’t do this without a mask, because the sky is not only blue, but has the proportions of green. Also the grass is yellow, which means a combination of red and green. So if you remove green, the yellow part of the grass will be red.
Okay. I thought the challenge was to do it without a mask. My bad. Thanks for your reply.
In fact, it was intended that way, and you actually did it. That the sky isn’t 100% red doesn’t matter. More important was understanding the logic of how the channel mixer works. You understood that, which fulfilled the purpose of the game. 
I forgot to say that if you understand the logic, you can put it into practice and do corrections quite well. For example, if this photo was taken early in the evening at sunset and I had the impression that there was still too much blue in the sky, I could use the channel mixer to make the sky a little more “reddish”.
I may be a little late here, but I had written up something similar a while back on a similar question for RawTherapee. My GIMP example was here (maybe it can help in some way?):
I had read that, Pat. It helped me understand. 
In fact, I looked in the manual on editing I have made up, and under Channel mixer the first link listed was your article.
Mind to share how you determine that?
You can see that from the histogram.
I thought I had gray/lightness blend figured out, but no. I reasoned that in gray channel, when you increase a color, that lightens it because you are saturating it, i.e., removing gray from the color. Adding lightness blend to the equation: increasing red makes areas with red lighter. But using the jpg of the original road, field, and sky scene and only increasing red to 1.163–no other changes–darkened the sky. How does that happen? If there is red in the sky, shouldn’t it make it lighter?
What tool are you talking about.?
???
Can you please make make a snapshot that could help us understand what you mean.
Are you talking about Mono mixer in GIMP?
I was talking about channel mixer and your post above of March 28. Sorry for the confusion.
My bad. I should have opened the reference to see what post you were referring to. I thought you’d respond to my last post.
In gray mode with the color mixer in darktble, you combine the color channels from zero. This means that if you increase the red channel, you will actually only see the red channel. As long as you don’t increase the values for other channels, they will remain off at first. Since the sky is blue, and you have not increased the blue channel, the sky is dark.
Yes, but you see a darker sky, because blue channel is not yet on and blue is the predominant color of the sky.
BTW, this is the big difference to the mono mixer in GIMP, which has the same function as the gray mode in the color mixer in darktable.
In the case of the sky, all RGB values drop, but in the case of the road, all go up. Can you explain what makes the difference? Is it because red is below 128 in the sky and above it in the road?
Maybe you are confused by the fact that nothing happens before you move the slider. That means you see the image as a combination of all three color channels. The moment you move the first color channel, the gray mode is activated. Try it without Lightniss blend mode to see what I mean.
Here is the explanation of the Blend Mode “lightness” in darktable : "This blend mode mixes lightness from the input and output images. Color data (chroma and hue) are taken unaltered from the input image. " https://www.darktable.org/usermanual/en/blending_operators.html How exactly the values will be affected by this blend mode, I don’t know. Maybe you should ask the darktable developers.
Yes, I have been confused by the behavior of the module if you reset it. When you reset, the colored image shows up in its unmodified state, and I thought that is where I was starting from. But if you move one of the sliders off zero ever so slightly, the image goes black. That, as you say, is the real starting point.
As you move the red slider up, the sky, the whole image, in fact, becomes lighter. But because the maximum is 2, the sky stays darker than the original. Using the green slider, moving it up a little leaves the image darker, but if you move the slider up more, the image crosses over from being darker to being lighter than the original–before the slider reaches 2.
Sorry to be so dense. Thanks for your patience.
No problem. The important thing is that we got this straightened out.
Yeah, it’s really confusing. Maybe gray mode should be activated immediately upon clicking on “gray”.
Actually the channel mixer already has this mode where you can use channels to control only the brightness of the image, (“lightness” mode), which is the same as “gray” mode with lightness blend mode.
The problem is, it starts exactly the same from zero. I think it would be much more intuitive if the sliders for the channels were already moved by default so that the brightness of the image is identical to the original brightness before the mode was switched on. That way, you could see changes immediately when you move the sliders afterwards.
A hint: To influence the brightness of the picture with the channel mixer, I prefer to use “gray” mode with HSV lightness blend mode, because with HSV lightness the colors won’t flip over if you go too far with the values
This should be documented in the user manual to avoid confusion.
Another question:
In order to have the same blue in the sky across several images the area selected by the color picker should be changed from rgb 57 99 151 to rgb 74 108 154. This is very easy to do using the channel mixer and raising the red slider in the red channel, the green slider in the green channel and the blue slider in the blue channel. Since the whole sky is blue and selected by means of a mask there is no problem with the white balance (neutral grey).
What would be a systematic approach if you had to maintain the correct neutral grey?
