Copy link to clipboard
Copied
This came about while I was trying to understand the difference between
the Darken and Darker Color blending modes.
The documented explanation for the Darker Color blending seems straightforward:
Darker Color
Compares the total of all channel values for the blend and base color and displays the lower value color.
But this does not always seem to be the case in RGB mode.
If I fill a base layer with sRGB (225,161,49) and a blend layer with sRGB (174,174,174)
then I expect Darker Color blending will compare the totals of the RGB values (435 vs. 522),
determine that the base layer is "darker" (435<522) and thus display the base color,
which is indeed the result.
I would also expect similar results for any permutation of RGB values which total 522.
For example Darker Color does interpret the base color as darker than sRGB (194,154,174)
and sRGB (154,194,174).
However, when I fill a blend layer with sRGB (174,154,194), Darker Color blending
displays this color instead the base color which has the lower total RGB value.
Conversely Darker Color blending displays the base color instead of sRGB (200,200,0)
which has a lower total RGB value (400 vs. 435).
Clearly my understanding of this blend mode is inconsistent with my observations.
Could someone offer a more accurate explanation?
Or is this a bug?
(I am using CC 2018 Release 19.0.)
The actual comparison is done using luminance, which is a weighting of each channel based on human color perception:
Y(Luminance) = SUM(R*0.2126 + G*0.7152 + B*0.0722)
(these may not be exact for Photoshop - they're just common weights)
So for a base of (225, 161, 49), you get a summed value of 435, but a weighted value of 166.52
A blend of (174, 154, 194) gives a summed value of 522, but a weighted value of 161.14
In this example, the blend wins the weighting, but loses the summation.
Copy link to clipboard
Copied
This works per channel, you're not supposed to add the numbers up.
If your base layer is red 255, and the blend layer red 174, then the result is red 174.
And then in your example you have base green 161, and blend green 174, so the result is green 161.
And so on...
Copy link to clipboard
Copied
Thank-you for your reply, however your explanation is for DARKEN blending.
DARKER COLOR blending is supposed to differ in the way "darker" is determined and
is supposed to display one color or the other, not mix values to display a third color.
Here are the explanations from the user guide page (dated February 15, 2017):
Darken
Looks at the color information in each channel and selects the base or blend color—whichever is darker—as the result color. Pixels lighter than the blend color are replaced, and pixels darker than the blend color do not change.
Darker Color
Compares the total of all channel values for the blend and base color and displays the lower value color. Darker Color does not produce a third color, which can result from the Darken blend, because it chooses the lowest channel values from both the base and the blend color to create the result color.
Copy link to clipboard
Copied
Interesting. I see what you're saying. Not sure how they calculate the results. Just asked a friend who wrote a book on it. Hopefully, he will chime in.
Copy link to clipboard
Copied
Got a responds from my friend. He's going to look at it later then he gets home, but basically he said that while the darker color is suppose to be a comparison of all the channels, it also includes a normalization process that accounts for human perception. You can kind of see how Adobe applies this in Camera Raw when you do a B&W conversion. The default slider values aren't all at 0:
Copy link to clipboard
Copied
Thank-you for your replies, Chuck.
Your first reply reassures me that I am not alone in seeing this.
Your second reply jives with Scott Valentine's later reply,
so hopefully we're all heading towards a better understanding
of this blend mode.
Copy link to clipboard
Copied
I'm the guy Chuck is referring to
For a fun exercise, you can create a spreadsheet to do the calculations. I sent one to Chuck to play with that identifies whether the base or blend color shows up in the result - I'll try and create one in Google Docs so you can see how it works.
Copy link to clipboard
Copied
Thank-you for your replies, Scott.
Your explanation is consistent with a rule-of-thumb that Dan Margulis
describes in his book "Modern Photoshop Color Workflow": the RGB
channels contribute to luminosity in roughly a 3:6:1 ratio.
This would also be consistent with how Darker Color blending seems
to work in Lab Color mode, where I presume "darker" is determined
by comparing lightness values.
Hopefully someone from the development team can confirm this and
update the Darker Color description (including for Lab mode).
Copy link to clipboard
Copied
I'll send a note to the PS engineering team and see if they can add some clarification. I replied with a link to the spreadsheet, which is being moderated before it shows up here.
Copy link to clipboard
Copied
Thanks, Scott, the spreadsheet is helpful.
I'll leave the thread as unanswered in case someone from adobe can post a better description.
Copy link to clipboard
Copied
I'm not sure how much better of a description you want - check the spreadsheet's answer against Photoshop's actual results and let me know if I need to tweak the weighting factors. I haven't tested the exact numbers through a significant range of combinations, but I can tell you that Adobe views their equations as proprietary, so they won't tell you with any more precision I'm afraid.
What other information about the blending mode do you need to feel comfortable with using it?
Copy link to clipboard
Copied
Thanks Scott and everyone else who replied.
Copy link to clipboard
Copied
The actual comparison is done using luminance, which is a weighting of each channel based on human color perception:
Y(Luminance) = SUM(R*0.2126 + G*0.7152 + B*0.0722)
(these may not be exact for Photoshop - they're just common weights)
So for a base of (225, 161, 49), you get a summed value of 435, but a weighted value of 166.52
A blend of (174, 154, 194) gives a summed value of 522, but a weighted value of 161.14
In this example, the blend wins the weighting, but loses the summation.
Copy link to clipboard
Copied
Here's the spreadsheet: Darker Color Blend Mode - Google Sheets
You can edit the white number cells with RGB values, and the red cells below them will update. The light blue cells are the weighting factors. As you change values, the empty cells to the right of the matrix will show which of the layers is going to show up on the canvas.
Copy link to clipboard
Copied
So I learned something new here.
I normally don't use blend modes I don't understand, because then it just becomes a random effect I can't control or visualize. Darker Color is, I confess, one of them.
Copy link to clipboard
Copied
You can check out Scott's book here. It has the calculations in it, and he's very much an authority on it. That's why so many of the industry best volunteered to be in his book. I doubt the engineering team is going to tell you anything more.https://www.amazon.com/Hidden-Power-Blend-Modes-Photoshop/dp/0321823761
https://www.amazon.com/Hidden-Power-Blend-Modes-Photoshop/dp/0321823761
Copy link to clipboard
Copied
All of the color blend modes use CIELab as their basis for the changes. There will be no easy to see relationships to RGB or CMYK values.