Index Home About Blog
Newsgroups: rec.pyrotechnics
From: nelson@cs.rochester.edu (Randal Nelson)
Subject: Re: Blue + Yellow= Green?
Date: Thu, 12 Mar 1998 14:51:04 GMT

> aaron vincent wrote: (original question)
>
> I was wondering since Barium compounds are fairly toxic, would it
> be possible to obtain a decent Green with CuO and and a small amount
> of a sodium compound?

> Donald Boscoe  <dlb39@cornell.edu> wrote:
>
> It may be possible to adjust the levels of the two coloring agents to a
> desirable green color, but the fine tuning would likely be far beyond
> the reaches of the average pyro lab.

> Randal Nelson wrote:
>
> Actually, it would would tend to produce white, not green, even if
> it were fine-tuned.  Blue mixed with yellow only produces green in
> subtractive processes i.e. mixing paints or inks.
> Combining emissions is is an additive process like the pixels on
> your color monitor. In that case you get yellow by turning on the
> red and green pixels. Add in the blue ones, and you have red + green + blue
> giving you white.
>

> Donald Boscoe  <dlb39@cornell.edu> wrote:
>
>	Where do you get red light out of CuO and Na salts?  If you had the
> right mixture of blue (from the Cu) and yellow (from the Na) I think it
> would make green light.  This is illustrated by Mr. ROY G. BIV, in whose
> name green is between blue and yellow.
>

OK, I oversimplified for the purposes of illustration with the color
monitor analogy. You are perfectly right that there is no appreciable
red in the sodium emissions.
The actual perceptual process involves how the so-called
"red", "green", and "blue" cones in the retina are stimulated.
The spectral tuning curves of these cones are quite broad, and the the
"red" and "green" curves overlap over about 75% of their area, and
both actually peak in the yellowish region of the spectrum.
(The "red", "green" and "blue" peaks are at 570, 540, and 445 nm
respectively - a very uneven distribution. So much for what we are
taught in high school).
The upshot is that a line-like yellow source (as in the sodium spectrum)
stimulates both the "red" and "green" cones more or less equally,
giving the same perceptual effect that a red line and a green line
would have together.
Add in enough blue, and the perception is whitish, as the
"blue" tuning does not overlap either the "red" or the "green"
significantly. (The situation is not symmetrical at all, and in fact
the "blue" cones are peculiar, compared to the red" and "green" cones
in a number of ways)
The reason that blue and yellow produce green in a subtractive process,
is that both blue and green pigments usually have substantial
reflection in the green portion of the spectrum. When you mix them,
that is all that is left unabsorbed, so the eye sees green.
There are actually some blue and yellow substances that do not
have much overlap, and the result of mixing these is not green, but
a muddy brownish color.

Perception is a complicated business...
I've been trying to get computers to "see" for years, and
it still doesn't work all that well.

RN


--
  Randal Nelson			 716-275-8488	University of Rochester
		      nelson@cs.rochester.edu	Computer Science Department
..!{allegra,decvax,rutgers}!rochester!nelson	Rochester, New York,  14627



Newsgroups: rec.pyrotechnics
From: nelson@cs.rochester.edu (Randal Nelson)
Subject: Re: Blue + Yellow= Green?
Date: Fri, 13 Mar 1998 19:21:15 GMT

In article <35085540.4A1@cornell.edu>,
Donald Boscoe  <dlb39@cornell.edu> wrote:

>Now a few questions ... What is
>Shimizu trying to say on page 46 of F.A.S.T. when he notes that the peak
>sensitivity of the bars is at 5070 angstroms and the peak sensitivity of
>the cones (presumably an average of all three types) is at 5550
>angstroms?  It seems that this fact is somewhat irrelevant to the next
>statement that "the black and white stimulus [in fireworks] is about ten
>times larger than that of colours".  Once he has made that statement
>clear, as an individual fact, his ideas about colored smoke follow
>logically.  Your comments inspired me to re-read this section of the
>book and I was a bit confused.
>	And a second question: Is the yellow sodium flame so intense and so
>able to mask other colors because of the inherent nature of the sodium
>atom, or because the light stimulates twice the number of cones in the
>human eye?
>

Since I don't have Shimizu's book, I have to guess a little bit.

The first statement is probably referring to the relative sensitivity of
the rods (not bars) and cones. The peak sensitivities are about right.
Basically, there is only one type of rod cell, as opposed to
three types of cone cells.
The main issue is that the rods are a couple orders of magnitude more
sensitive than the cones, and are essentially responsible for vision in
dim light, hence the lack of strong perceptual colors at night
and the idea that rod vision is "black and white".
Basically, there is a light level below which you can see quite well,
but colors are not very apparent.
In normal daylight, the rods are pretty well saturated, and are not thought
to contribute much to the perceptual signal.
Fireworks are mostly viewed in the dark, but the stars are quite bright
from an absolute standpoint, so both rod and cone systems are operating.
My guess is that Shimizu is saying that because the rods are intrinsically
more sensitive than the cones, their signal is (10 x) greater than the cone
signal when both are operating as is the case with fireworks against
a night sky. This makes sense, but I would not be surprised if it turned
out to incorrect if someone were to devise a psychophysical test to
measure the ratio (I'm not at all sure how one would devise such a test)
The retinal system is highly non-linear; various signals inhibit
other signals, and the sensitivities of the cells themselves are
non-stationary (dark adaptation etc.)
Color mixing in bright light is one area where a linear model seems
to work fairly well to describe perception, but other parts of the
envelope are pretty complicated to deal with.

The swamping effect of sodium yellow is mainly due to
the intrinsic intensity of the 5889 and 5895 angstrom (D) lines
in thermal emmisions.
The fact that they occur close to the maximum cone sensitivity plays
a roll in that if the yellow signal is strong, added blue or green
just pushes the perception towards white but it is secondary in this case.


RN
--
  Randal Nelson			 716-275-8488	University of Rochester
		      nelson@cs.rochester.edu	Computer Science Department
..!{allegra,decvax,rutgers}!rochester!nelson	Rochester, New York,  14627



Index Home About Blog