Re: still blue

["Thomas Schmitt" <scdbackup@gmx.net>]

Hi,

Nicolas George wrote:
> > - Our brain lies to us about colors all the time.

This begins already with the combination of light receptors and the first
nerve cells in the chain which creates the color evaluations.

As mammals we should normally be blind against red-yellow-green contrasts.
Dogs and cats perceive colors only as contrast of blue and yellow.
But as primates we benefit from a mutation which has split the yellow
receptors into read and green ones and a copy of the blue-yellow contrast
evaluator nerve cells. (This hacked us into the communication between
plants and birds or insects).
So our color perception is the combined evaluation of a blue-yellow
contrast and a red-green contrast.

The old yellow impression is now the impression of less blue than the
average of red and green plus red and green being quite equally strong:
  xterm -bg '#FFFF80' &

Brightness and color strength are not the same, because brightness is
mainly perceived by the black-white receptors of which we have much more
than color receptors. (Thus nightly black-and-white view.)
On my screen i (individually) see pure green brighter than pure red which
is brighter than pure blue:
  xterm -bg '#ff0000' &
  xterm -bg '#00ff00' &
  xterm -bg '#0000ff' &
Nevertheless a slight overweight of blue can create a blue-ish impression:
  xterm -bg '#e0e0ff' &
Note that the sum of red and green is 1.75 times brighter than blue. So the
calibration of colors matches my individual eye color chemistry quite well,
at the expense of a skew in brightness.

In my youth i had discussions with anthroposophists who disputed the physical
nature of light colors as spectrum of wavelengths. Their main argument for
Goethe's color ideas was the fact that at the blue end of the spectrum
they can see red comming back as violet. Thus the idea of a color circle.
Actually this reddish impression comes from the fact that the red receptors
are a bit more sensitive to blue than the green receptors. So the sensed
color is much-more-blue-than-yellow + slightly-more-red-than-green =
deep blueish violet.


Emanuel Berg wrote:
> what would you do to
> write a function that will return a RRGGBB that is always blue
> and as many blues as possible possible?

I'd try to follow above principles and expect a certain grey zone where
individual light sources and eyes deviate in display and perception from
the numerical RGB values.
I.e. every RGB vector where B is significantly larger than (R+G)/2 should
appear as blueish, or else the RGB light source or the eyes can be
considered non-mainstream.


Have a nice day :)

Thomas