Hi
After watching some videos by Boris on using the colour mixing in color calibration, I am wondering if the use of color mixing could be simplified in the following way:
Let us imagine we want to add brightness to a region of the photo with a certain color (A), and not affect other region with a not so different color (B). We could select an area of color (A) with the mouse, and then move the mouse weel to add brightness by multiplying each channel value in proportion to the color content in (A).
Then, if we see that (B) is too much affected, we could select an area of (B) and reduce the multipliers of each channel in proportion to the color content in (B).
If (A) and (B) are far appart we might have finished in a single step, but if they are close together, we might have to repeat it several times.
For sure, we could use the same to affect the individual channels, instead of the brightness. We can make that some particular color (A) becomes more red by selecting the region and using the mouse wheel to add/reduce redness to that region.
I hope I explained the proposal clearly, and that it has some sense!
Hi Miguel and welcome to the forum!
I’m not sure I understood what you meant. Could you demonstrate that with one example?
Actually, it is already possible to mask the corresponding region with the help of the parametric masks in order to treat it separately.
Hi Boris, great to get in contact with you, I am a big admirer (is that an english word? I mean something like ‘fan’) of you!
I can’t imagine how I can produce an example on a photo, but my inspiration came from a video you produced where you showed how to get the color composition (or content) of a region to change its tonality, and at the same time don’t change the tonality of other region(s). And I just watched another video by you on giving more contrast to the highlights, where you proposed to reduce the brightness of the blue channel. I thought that you could reduce the brightness of the color with (R;G;B) content of the sky instead, but it would require to measure that content and change the channels in proportion to that content.
Let us imagine that we want to give more red tint to a region with a content (R,G,B)=(200, 152,127), while not changing the redness of a region with content (180, 170,100). Then, following your proposal, or what I understood, I would add to the red channel a*(200,152,127) and reduce the red channel by b*(180,170,100). Then we are adding two vectors which are not very perpendicular, but with the proper combination of ‘a’ and ‘b’, which could be controlled by the mouse wheel, you can obtain a compromise which satisfies you: add some red to the first region while not changing it (or reducing) on the second region.
Of course, you might add other regions in the game. The final result can be the same you can obtain by doing ‘manually’ the calculations you proposed, but I think that in this way you will not be obliged to analyze numerically the content of each region involved.
I hope I clarified the proposal, otherwise, please, tell me.
| Remitente notificado con
Mailtrack | 22/01/22 11:36:12 | |
- | - | - | - |
Hau idatzi du Boris Hajdukovic via discuss.pixls.us (noreply@discuss.pixls.us) erabiltzaileak (2022 urt. 22, lr. (10:56)):
Hi again
As I guess you are very strong in mathematics, here I explain my proposal in a more precise way:
Color mixer matrix at start:[1,0,0;0,1,0;0,0,1]
we add red ‘a’ times colour content [r1;g1,b1]
we reduce ‘b’ times colour content [r2;g2;b2]
the new matrix would be:[1+ar1-br2,0,0;ag1-bg2,0,0;ab1-bb2,0,0]
I hope this makes things clear
Best regards and thanks for you time
Mailtrack | 22/01/22 11:54:43 | |
- | - | - | - |
Hau idatzi du Boris Hajdukovic via discuss.pixls.us (noreply@discuss.pixls.us) erabiltzaileak (2022 urt. 22, lr. (10:56)):
Sorry Boris, my matrix was wrong, it is the first row of the transform matrix which has to be changed:
[1+ar1-br2,ag1-bg2,ab1-bb2;0,1,0;0,0,1]
Mailtrack | 22/01/22 12:05:34 | |
- | - | - | - |
Hau idatzi du Boris Hajdukovic via discuss.pixls.us (noreply@discuss.pixls.us) erabiltzaileak (2022 urt. 22, lr. (10:56)):
Thank you for your detailed explanation but I am actually the opposite of what you think. I have no idea about these technical aspects of your proposal.
It would be better if you address them to the developers. Maybe @anon41087856 can say something about it.
It would be even better if you put your proposal on github:
@Kaskarina : Is there a particular reason you have that tracking item in all your posts?
I use that tracking since some years ago. I find it interesting, and sometimes important, to know if the mail was opened. Is that a problem for you?
Regards
Miguel
Hau idatzi du rvietor via discuss.pixls.us (noreply@discuss.pixls.us) erabiltzaileak (2022 urt. 22, lr. (12:41)):
Thanks, Boris, I will proceed to github.
By the way you explained the color mixer I thought you have an strong mathematical background. Well, I think you must be good at mathematics, anyway!!
I am a retired engineer, very familiar with mathematics, and since three years ago, struggling to develop my artistic skills. It is being difficult, I find mathematics much easier 
Best regards
Miguel Zatarain
Mailtrack | 22/01/22 14:33:16 | |
- | - | - | - |
Hau idatzi du Boris Hajdukovic via discuss.pixls.us (noreply@discuss.pixls.us) erabiltzaileak (2022 urt. 22, lr. (12:23)):
Here? Yes. I really don’t feel the need to be tracked in addition to what the forum does
You can use the clut modules to select any color…replace one of the patches and then you can use L a b and saturation controls on that color…you might be able to map some of that to a key and scrolling
Ok, I will try to remember disabling tracking when posting here
You may want to watch the two videos on Channel Mixer by Nicolas. I think they address some of your questions, particularly the second video.
Thanks, Christian, the videos explain well the basics of channel mixer. What i am proposing is a way to avoid the long, difficult process of calculating the coefficients in the matrix, as Nicholas is doing in the video.
I already sent a request to the developers, in case they find the proposal interesting. For some reason i thought pixel.us was the way to proceed but Boris suggested me to do it on github.
Thanks again
Christian Pfister via discuss.pixls.us <noreply@discuss.pixls.us> igorleak hau idatzi zuen (2022 urt. 22a, lr. 22:47):