Human perception in color mixing

[Tomáš Pažourek]

Look at this image:

Loading Image...

50% of its pixels are blue (0, 0, 255), 50% of its pixels are yellow (255, 255, 0).

When I look at my computer display from a distance where I cannot distinguish the individual pixels anymore, the whole image appears in some shade of green.

However when I use computer to mix these colors (e.g. apply gaussian blur on the image), the resulting color is dark grey (128, 128, 128).

This seems very wrong to me, the color mixing that we use in our computers is completely different to how our eyes mix the colors.

My question is:

Is there a color model, which would give me results more similar to how my eye mixes the colors?

[Thomas Richter]

Post by Tomáš Pažourek
https://dl.dropboxusercontent.com/u/1829396/yellow-blue.png
50% of its pixels are blue (0, 0, 255), 50% of its pixels are yellow (255, 255, 0).
When I look at my computer display from a distance where I cannot distinguish the individual pixels anymore, the whole image appears in some shade of green.
However when I use computer to mix these colors (e.g. apply gaussian blur on the image), the resulting color is dark grey (128, 128, 128).
This seems very wrong to me, the color mixing that we use in our computers is completely different to how our eyes mix the colors.
Is there a color model, which would give me results more similar to how my eye mixes the colors?

This is monitor dependent. It appears grey here on my screen. Problem is
that color mixing by position is a linear process in the physical
intensities, but color mixing by pixel value in sRGB space is not. The
reason is that the sRGB color space is not linear, neither is the
monitor, or the overall signal path from the computer memory to the
screen - unless the screen is properly calibrated. That is, pixel values
are not proportional to physical intensities. And physical intensities
are not proportional to human perception either.

If you see it as "green", then apparently because your monitor has a
stronger green luminance than red and blue, which may be plausible. As
said, does not happen on my laptop (LCD) monitor.


Greetings,
Thomas

[Tomáš Pažourek]

Post by Thomas Richter

Post by Tomáš Pažourek
https://dl.dropboxusercontent.com/u/1829396/yellow-blue.png
50% of its pixels are blue (0, 0, 255), 50% of its pixels are yellow (255, 255, 0).
When I look at my computer display from a distance where I cannot distinguish the individual pixels anymore, the whole image appears in some shade of green.
However when I use computer to mix these colors (e.g. apply gaussian blur on the image), the resulting color is dark grey (128, 128, 128).
This seems very wrong to me, the color mixing that we use in our computers is completely different to how our eyes mix the colors.
Is there a color model, which would give me results more similar to how my eye mixes the colors?

This is monitor dependent. It appears grey here on my screen. Problem is
that color mixing by position is a linear process in the physical
intensities, but color mixing by pixel value in sRGB space is not. The
reason is that the sRGB color space is not linear, neither is the
monitor, or the overall signal path from the computer memory to the
screen - unless the screen is properly calibrated. That is, pixel values
are not proportional to physical intensities. And physical intensities
are not proportional to human perception either.
If you see it as "green", then apparently because your monitor has a
stronger green luminance than red and blue, which may be plausible. As
said, does not happen on my laptop (LCD) monitor.
Greetings,
Thomas

Thank you, Thomas. You are right. On most common monitors I tried afterwards, the color is grey. When I do the mix in gamma-adjusted sRGB, the result is around (188,188,188) or (186,186,186) depending on the transformation used (full sRGB vs. simplified sRGB), which seems very similar to what I see.

But still there is the human perception point of view, which you mentioned. I know about some of the research that focuses on more uniform representation of colors in the sense of human perception (LAB and LUV color spaces). Is there some research which focuses more on how to mix and blend colors to be natural to human eye?

(Still the gamma-adjusted sRGB mixing works good enough for my perception, at least in this case.)

Greetings,

Tomáš