[Physltest] [Phys-L] Re: Color (was LED mini-flashlight price break)

[John Denker <jsd@AV8N.COM>]

I was very impressed with Michael Edmiston's recent note
about spectra, filters, and color-names.  It was nicely
grounded in experiment and clarified some important issues.

I was particularly struck by this passage:

> * * * My point is... * * *
>
> I am not only talking about a conflict between a physics definition of
> color and a public definition of color.  Right now I am talking about an
> inconsistency totally within public perception.  The spectral regions
> reflected/transmitted by objects people call blue do not match the
> portion of the spectrum these same people identify as the blue region of
> the spectrum.  That doesn't always convince them to call it cyan, but
> they at least realize there is just as much green light coming from that
> object as there is blue light.  There is no doubt about this, and they
> all agree.  They saw it with their own eyes, but they don't necessarily
> agree to change what they call it.

That's a point I had never thought about before.

After thinking about it for a while, I'm not convinced the
inconsistency is as pronounced as it might seem.  There is
a flaw in the logic, a slight mixing of apples and oranges.

Take a look at the CIE chromaticity diagram.  I hope you have
a nicely-printed version you can use;  failing that, you can
look at an approximate version such as
   http://www.av8n.com/imaging/color-management.htm#fig-cie2
which is necessarily very imperfect because of the limitations
of your computer's display.

In any case, the following remarks are true without
approximation:
  -- The pure spectral colors wrap around the boundary of the
   diagram, from roughly 7:00 clockwise to 3:00.
  -- Cyan can be found on the diagram ... in the interior, not
   on the boundary.

When you project a spectrum on the wall, you can get people
to agree that moving halfway between spectral blue and spectral
green, by moving _along the spectrum_, along the boundary of
the CIE diagram, lands you on a point that should not be
called blue.  This is the "apples" question.

However (!) you have not thereby pinned down their opinion
about cyan.  There is an entirely separate question about
moving halfway between spectral blue and spectral green
_via the interior_ of the diagram.  This is the "oranges"
question.

I'm not playing devil's advocate here.  I believe people
on the street are within their rights to consider 460nm
spectral blue to be one shade of "blue", and cyan to be
another shade of "blue".  I wish they wouldn't ... but
to this point we don't have any firm grounds for saying
they are being illogical or self-inconsistent.

A real self-inconsistency is revealed in the following
example:  Today I saw a music CD entitled "Primary
Colors".  The cover art showed three colors, evidently
drawn with crayons.  They were red (a deep dark shade of
red, not even remotely resembling magenta), yellow, and
blue (a deep dark shade of blue, not even remotely
resembling cyan).  Now regardless of the names, I can
look at the actual colors and be confident that they
will not work at all well as primary colors.

The point here is that if we adopt (temporarily!) the
street terminology, we can say that "blue" in some
sense is a primary color for subtractive mixing.  But
alas not any shade of "blue" will work ... only the
cyan shade.  The CD artwork chose the wrong shade.
They were using an imprecise notion of "blue", and
it led them down the garden path.

Meanwhile, the scientific and technical usage is different
from the street usage.  We say the additive primaries are
RGB and the subtractive primaries are CMY.  Our notion of
B does *not* include cyan, not even close.  Similarly, on
my computer, filling a rectangle with "blue" means something
very specific, and is very different from filling with "cyan".

=======================

Returning to the roots of this thread:  It really is
great fun to do the following demonstration.

Get three lights (R, G, and B).  The Walgreens $1.00 LED
flashlights work just fine for this.  Get a helper to
shine two of them at the same point on a white screen.
You shine the third at the same point.  This produces
a fairly good version of white.

Now interpose something that casts a shadow.  A spoon
works well.  There will be three main shadows:  one
cyan, one magenta, and one yellow ... i.e. the secondary
colors of the RGB system.  People generally ooh and
aah over the colored shadows.  They are pretty, and
somewhat unexpected.

In places where the main shadows overlap, you can get
additional colors, namely the RGB primaries and
possibly black.

At this point some serious teaching can begin.  You
point out that the cyan color is significantly
different from the primary blue color.  It's blue
plus green.  Cyan is just as closely allied to
primary green as it is to primary blue.  Similarly
people on the street may be tempted to call magenta
a shade of red, but physically it is no closer to
red than it is to blue.

You can also remark that inkjet printers use cyan,
magenta, and yellow inks.

At this point there may be a rebellion, along the
following lines:  "But everybody knows that red,
yellow, and blue are the primary colors.  That's
what I've been teaching my class for thirty years,
so it *has* to be true.  This talk about RGB
primaries is nonsense.  It's a trick.  It's an
optical illusion or something."

Well, I say, there's a multi-jillion-dollar industry
that relies on additive RGB primaries for making TVs
and computer displays.  And there's another multi-
jillion-dollar industry that relies on subtractive
CMY primaries for printing.  This is no illusion;
this is where the rubber meets the road, all day
every day, worldwide.  Would you rather believe what
was said in your third-grade art class, or believe
the demonstration in front of your eyes right now?

=============

If you can't lay hands on theatrical gels or lab-grade
Melles-Griot filters, here's an ersatz that you may
find useful:  Some (not all) color printers and some
color copiers can print onto 8.5x11 transparent acetate
sheets.  Print yourself a big rectangle of cyan, a big
rectangle of magenta, and a big rectangle of yellow.
You can then use them as filters.  You can stack them
in series to perform subtractive color mixing.  How well
this works will depend on details of your print system,
so I make no guarantees ... but I've used this trick
on several occasions with good results.
_______________________________________________
Phys-L mailing list
Phys-L@electron.physics.buffalo.edu
https://www.physics.buffalo.edu/mailman/listinfo/phys-l