Any final authority on color controversy?
Robert Truscio
an art test that asked for the primary colors for mixing paints. The teacher
marked him incorrect, stating that red yellow and blue was the correct
answer. Meanwhile, the hallway of another school was decorated with color
mixing wheels using color paints of cyan magenta and yellow as the
primaries. Why have they done this?
I don't care so much which answer is correct, but why does this dichotomy
exist? There is an abundance of books and sources that point to or argue for
one or the other group. I know this pertains to art, which may be considered
subjective, but can anyone show definitively that the child answered
correctly or incorrectly? Should a child be taught differently depending on
the teacher or school? Or is it merely, take your pick? Wouldn't it be best
to learn correctly from the start? (Please note that this is exclusive of
the designated primary colors of light for which there appears to be no
controversy.)
Is there any final authority that holds a definite answer?
R. Truscio
Steve Hehr
I'm sure others more expert than I am will reply; but here's my two
cents:
I don't think it's a question of which is correct -- I would say both
answers are correct, but that the student's answer is more exactly
correct. The primaries ARE cyan, magenta and yellow; and cyan is a
particular shade of blue, and magenta a red, though quite a bluish-red
-- according to their dictionary definitions, anyway. Cyan and
magenta are sub-classes of blue and red, respectively. So the
teacher, I feel, is correct in saying that yellow, red and blue are
the subtractive primaries, but definitely wrong in marking the
student's more precise answer incorrect.
I suspect that the teacher in question used the simpler classes and
terminology because that's what she/he found in the course materials.
Quite possibly the teacher was unaware that cyan and magenta ARE blue
and red... I don't know, one would think an art teacher would be more
knowledgeable, but stranger things have happened.
Ironically, then, the test may have revealed more about the state of
the teacher's knowledge than the student's. I think the more
important issue is not which set of terms is used -- the simpler or
the more precise -- but that a teacher should understand just what an
answer is saying before marking it wrong. Greater knowledge on the
part of the student should be rewarded, not discouraged. A really
good teacher, I feel, would have used such an answer as an example to
the class, showing why the student's answer, though different from the
standard response, is not only acceptable but, in a sense, even
superior.
--
Steve Hehr
To send me email, replace the "OUT" in my address with its opposite.
Ben Newsam
header saying 'EarthLink Inc. -- http://www.EarthLink.net'
>
>I don't think it's a question of which is correct -- I would say both
>answers are correct, but that the student's answer is more exactly
>correct. The primaries ARE cyan, magenta and yellow; and cyan is a
>particular shade of blue, and magenta a red, though quite a bluish-red
>-- according to their dictionary definitions, anyway. Cyan and magenta
>are sub-classes of blue and red, respectively.
Huh? Hardly.
> So the teacher, I feel, is correct in saying that yellow, red and
>blue are the subtractive primaries, but definitely wrong in marking the
>student's more precise answer incorrect.
The question was about mixing paints, so red yellow and blue are the
ones.
>I suspect that the teacher in question used the simpler classes and
>terminology because that's what she/he found in the course materials.
>Quite possibly the teacher was unaware that cyan and magenta ARE blue
>and red...
No they're not. Cyan is no more blue than it is green, and magenta is no
more red than it is blue. Same with yellow... Is that red or green in
your weird system?
--
Ben
Steve Hehr
As I said above: "according to their dictionary definitions..."
Per Webster's 7th New Collegiate Dictionary:
Cyan: dark blue: blue
Magenta: a deep purplish red
So it is Webster's "weird system" as well.
Gernot Hoffmann
Robert,
my opinion (instead of authority):
1. C M Y are the primaries for offset printing, for lasers and
inkjets.
Cyan, Magenta, Yellow are merely names. The inks or toners are defined
by reflectance spectra. It´s well known that people´s opinions differ
considerably about the color of Cyan. Offset Cyan doesn´t look like
monitor Cyan. Offset Magenta doesn´t look like monitor Magenta.
2. R Y B cannot be used as primaries because Y and B are complementary
colors (literally interpreted for technical color systems). It would
be impossible to mix green.
3. Fine art painting is not based on the mixing of just three
primaries.
My wife has more than 50 cans or glasses with pigments. Each pigment
has its specific spectrum and these colors exceed probably the gamut
of offset inks. Some are very vibrant.
Best regards --Gernot Hoffmann
kahrs
> A grade school student answered, "Cyan magenta and yellow," to a question on
> an art test that asked for the primary colors for mixing paints. The teacher
This was a question on mixing paints. When mixing paints,
colors mix in a subtractive way. The primary colors in
subtractive painting are CMY, so the student was right.
> marked him incorrect, stating that red yellow and blue was the correct
> answer.
The teacher was obviously thinking about additive mixing.
> I don't care so much which answer is correct, but why does this dichotomy
> exist?
The dichotomy exists because mixing paints/pigments is
completely different from mixing light-emitting sources.
Ink only reflects light (with spectral components
"damping" within mixtures) while light sources really
produce radiation (LEDs for example, with spectral components
adding up).
> There is an abundance of books and sources that point to or argue for
> one or the other group.
There is no contradiction in it. Read page 386 and Plate 11 of
Rodgers, Procedural Elements for Computer Graphics
or chapter 13.3 and plates II.3 and II.6 of
Foley/van Dam, Computer Graphics: Principles and Practice
Danny Rich
grade school student learned the terms cyan and magenta since his teacher
did not teach him those. One of the unfortunate things about the current
educational system is that exams do not "examine" what a student knows but
what the teacher knows. But that is a separate discussion for another
newsgroup.
To the dichotomy on color terms I contribute the following. The terms cyan
and magenta, as documented in English language dictionaries, so well
documented by Steve Hehr, are relatively new terms. Yet the use of color in
art is much older. In his treatise on the colors of lights, Newton list the
colors as Red, Orange, Yellow, Green, Blue, Indigo, Violet. Indigo is the
dye used to produce the nice, deep color of denim which we today call "blue"
but in earlier times the color "blue" was considerably greener than that -
what we would likely call cyan - equally placed between green and indigo
blue. Most all true reds (not orange reds) have a blue component.
Psychophysically, it is said that the unique or pure red is extra spectral.
So the texts on paint mixing for artists use the terms Red, Yellow, Blue to
mean the same colors that today color engineers call Magenta, Yellow, Cyan.
That is why, if you mix an artists yellow and an artists blue you do, in
fact, obtain a green and not a gray - though a pure green pigment is much
nicer.
Since art is highly influenced by history it is unlikely that this situation
will ever change. My good colleagues, Joy Turner-Luke who teaches at Studio
231 and Mark Gottsegen who teaches painting in Art Department of the
University of North Carolina - Greensboro still use the terms Red, Yellow,
Blue even though they are technically compentent about the spectral
signatures of the pigments from their work in ASTM D01.57.
Danny Rich
"Robert Truscio" <B-P.T...@worldnet.att.net> wrote in message
news:VSeTc.444004$Gx4.3...@bgtnsc04-news.ops.worldnet.att.net...
Mauro Boscarol
The primary color of a printing press are the inks of that press.
Every printing press has different inks.
The primary colors of a monitor are the phosphors of that monitor.
Every monitor has different phosphors.
The primary colors of every mixture (additive or subtractive) are the
colors used in that mixture.
Mauro Boscarol
Ben Newsam
header saying 'Optimum Online'
>So the texts on paint mixing for artists use the terms Red, Yellow,
>Blue to mean the same colors that today color engineers call Magenta,
>Yellow, Cyan.
Pure nonsense. I challenge you to produce evidence for this wild
assertion.
--
Ben
Don Klipstein
>Steve Hehr <stev...@OUTfionline.net> writes, despite their Organization
>header saying 'EarthLink Inc. -- http://www.EarthLink.net'
>>
>>I don't think it's a question of which is correct -- I would say both
>>answers are correct, but that the student's answer is more exactly
>>correct. The primaries ARE cyan, magenta and yellow; and cyan is a
>>particular shade of blue, and magenta a red, though quite a bluish-red
>>-- according to their dictionary definitions, anyway. Cyan and magenta
>>are sub-classes of blue and red, respectively.
>
>Huh? Hardly.
>
>> So the teacher, I feel, is correct in saying that yellow, red and
>>blue are the subtractive primaries, but definitely wrong in marking the
>>student's more precise answer incorrect.
>
>The question was about mixing paints, so red yellow and blue are the
>ones.
I hear people telling me this, but I have tried mixing yellow and blue
paints in kindergarten and in first grade and the result was not that
green a green.
>>I suspect that the teacher in question used the simpler classes and
>>terminology because that's what she/he found in the course materials.
>>Quite possibly the teacher was unaware that cyan and magenta ARE blue
>>and red...
>
>No they're not. Cyan is no more blue than it is green, and magenta is no
>more red than it is blue.
How about in practice? Ever look at how printing inks look in a test
print?
>Same with yellow... Is that red or green in
>your weird system?
>--
>Ben
- Don Klipstein (d...@misty.com)
Gernot Hoffmann
...
>
> So the texts on paint mixing for artists use the terms Red, Yellow, Blue to
> mean the same colors that today color engineers call Magenta, Yellow, Cyan.
> That is why, if you mix an artists yellow and an artists blue you do, in
> fact, obtain a green and not a gray - though a pure green pigment is much
> nicer.
>
> Since art is highly influenced by history it is unlikely that this situation
> will ever change. My good colleagues, Joy Turner-Luke who teaches at Studio
> 231 and Mark Gottsegen who teaches painting in Art Department of the
> University of North Carolina - Greensboro still use the terms Red, Yellow,
> Blue even though they are technically compentent about the spectral
> signatures of the pigments from their work in ASTM D01.57.
>
> Danny Rich
>
Danny,
some people may teach that artists mix their paints by three "primaries".
But it´s wrong - artists have always used plenty different pigments
(confirmed by Dr.Volker Hoffmann, professor for History of Art at the
University Bern, Switzerland, one of my brothers).
Mixing just three "primaries" doesn´t create pleasant colors.
By the way, it´s not even possible to mix all kinds of pigments because
some have affinity to water but others affinity to alcohol or terpentin
(don´t know the English word).
Best regards --Gernot Hoffmann
Jim Palmer
<<clip>>
>
>> marked him incorrect, stating that red yellow and blue was the correct
>> answer.
>
>
> The teacher was obviously thinking about additive mixing.
>
green out of those three.
Jim Palmer | Univ. of Arizona | Optical Sciences Center |
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 100,000 Newsgroups - 19 Different Servers! =-----
kahrs
>> The teacher was obviously thinking about additive mixing.
>>
> Since when are RYB useful in additive mixing? I challenge you to get a
> green out of those three.
You are right, I had misread RYB as RGB.
I have never heard of RYB and thought he meant RGB.
The sources I mentioned clearly speak about RGB.
So, it is even more surprising that the student's
opinion was not accepted as being right.
Danny Rich
Technology by Billmeyer and Saltzman. Perhaps my 30 years in color science
and technology is not strong enough evidence.
Danny
"Ben Newsam" <ne...@microser.demon.co.uk> wrote in message
news:l61yTwCn...@microser.demon.co.uk...
Greg
news:a815dbcf.04081...@posting.google.com...
> some people may teach that artists mix their paints by three "primaries".
> But it´s wrong - artists have always used plenty different pigments
> (confirmed by Dr.Volker Hoffmann, professor for History of Art at the
> University Bern, Switzerland, one of my brothers).
I thought it was clear enough that question should be interpreted to mean
that if an artist could ONLY choose three colours, which three would they
use?
Greg.
Steve
I went to college for Fine Art (Painting, Drawing and Sculpture), and also
took many classes in High School.
Although this was a while ago (1980s),
All the classes I ever had, taught that the "color wheel" and the 3
primaries are Red Yellow and Blue, the secondaries are Orange, Green and
Violet (Sometimes they say Purple). They also all state that complementary
colors are Red - Green, Yellow - Violet and Orange - Blue. Now this is
taught as "theory", artist pigments aren't Red, Yellow, Blue but Canary
Yellow, Cobalt Blue, Burnt Umber etc. The number of pigments artists use can
easily be in 100s. And white is used to tint (lighten) a color
Usually when they teach the color wheel they display as the R - Y - B color
model a red that is a fire engine red and a blue that is closer to cyan. In
practice they don't have just a blue but 20 blues to choose from etc so
they are not painting with just primaries nor are they necessarily trying to
teach that, but how to mix paint, and complemary colors etc. They also teach
Tints and shades which are produced by using white and black paint.
I don't know if mixing paint and than putting on media is different from
ink/toner laid down on top of each other. I don't how or why the CMY vs BRY
are taught. I know that others here try to say that the art teachers are
really meaning cyan and magenta, but call it red and blue. For blue it's
almost correct, but they really mean a color somewhere in-between blue and
cyan. But Red they defiantly mean Red, every slide I can remember seeing in
class was Red not magenta.
I would also like to point out that the inks used in printing are not
completely CMYK either, and in fact magenta can be very different on 2
different printers. Ink manufactures can't create a perfect C, M, Y. It's a
mixed blend of pigments to produce these inks
Of course with computers and a lot of art departments now teaching art on
the computer they may teach the color wheel differently today. However the
color wheel in an art class isn't really about printing with 3 primaries but
with a palette of colors 12 colors the 3 primaries, the 3 secoundaries and
the 6 tercearies (yellow-green, red-oragne etc) The Model is more like HSL
in a way.
I would say that the student was correct with his answer but was wrong in
that he either didn't listen to the teacher or is trying to "show up" the
teacher, which should have been brought up durning disscussion not on a
test.
Steve
"Robert Truscio" <B-P.T...@worldnet.att.net> wrote in message
news:VSeTc.444004$Gx4.3...@bgtnsc04-news.ops.worldnet.att.net...
Danny Rich
"Gernot Hoffmann" <hoff...@fho-emden.de> wrote in message
news:a815dbcf.04081...@posting.google.com...
>
> some people may teach that artists mix their paints by three "primaries".
> But it´s wrong - artists have always used plenty different pigments
> (confirmed by Dr.Volker Hoffmann, professor for History of Art at the
> University Bern, Switzerland, one of my brothers).
>
> Mixing just three "primaries" doesn´t create pleasant colors.
> By the way, it´s not even possible to mix all kinds of pigments because
> some have affinity to water but others affinity to alcohol or terpentin
> (don´t know the English word).
>
> Best regards --Gernot Hoffmann
In deed they do use many pigments as does the printing industry use more
than CMYK - my own company has created and maintain 1200 unique ink formulas
for one large consumer products company. Most of them cannot be reproduced
even by the dye-based inks in commercial ink jet printers. But, as
mentioned in several other posts, in North America, art schools still teach
that RBY are the primaries for mixing a range of color from a small set of
paints.
Respectfully,
Danny
Gernot Hoffmann
Narciso Silvestrini + Erns-Peter Fischer
Idee Farbe
Baumann & Stromer, Zürich 1994
This is a very interesting catalogue for an exhibition
(which I had visited 1995 in Lausanne).
Even the German version should be readable by anybody
(left side structured description, right side illustration
and references).
Maybe English, French and Italian version exist.
Google search was without success.
It turned out in the previous discussion that the question
"which primaries are used in art ?" can be interpreted in
the sense of "primaries = pigments = inks" or "primaries =
abstract colors".
The historical models seem to be ambivalent in this sense
as well.
I don´t intend to start the discussion again. Just wanted
to mention this booklet.
Best regards --Gernot Hoffmann
Jo Taylor
are red, blue, yellow and are subtractive - i.e. if the paint is added to a
white sheet of paper, the amount of light reflected becomes less as more
paint is applied - light is subtracted. Cyan, magenta, yellow (with black)
are the primary colours for printers' ink and are also subtractive for the
same reason. Only red, blue green, the primaries in light, are additive as
can be witnessed when a black stage is illuminated by each successive colour
until the three overlays produce white light.
There are three systems in common use, rby for opaque paint, cmyk for
transparent inks and rgb for light.
AT
"kahrs" <Juergen.Kah...@vr-web.de> wrote in message
news:2o8ut3F...@uni-berlin.de...
Jürgen Kahrs
> The teacher was correct and not thinking additive. Primary colours in paint
> are red, blue, yellow and are subtractive - i.e. if the paint is added to a
I have to admit that I misread rby as rbg.
> There are three systems in common use, rby for opaque paint, cmyk for
> transparent inks and rgb for light.
Sounds convincing. Then I wonder why the sources I quoted
(like many other scientific sources) don't mention the
two kinds of subtractive systems. I know that my reading
is biased toward computer science texts. Therefore I would
be interested in a link that explains this "trichotomy"
more closely.
We have seen much controversy in this thread.
If this question is so controversial (potentially
only because it was posed incompletely), why does
a teacher expect his students to go for just one
of the options you mentioned above ? This was the
original question in this thread.
Jo Taylor
wife's name.
Scientific sources tend to rely on computers for their information.
Computers, working in light, work on the RGB system. CMY(K) can be produced
by mixing two RGB primaries as CMY are the secondaries of RGB. It is
impossible to recreate the RBY system using a computer. Unless your
scientific sources received teaching in 'Art', as distinct from 'Science',
there is no reason why they should be aware of RBY.
The teacher in the example quoted was an art teacher so paint is assumed to
be the medium in question and thus RBY. Had the teacher been in a school of
printing, CMY would have been the correct answer, just as RGB would have
been correct if the teacher had been concerned with computers, television,
stage lighting, etc..
AT
"Jürgen Kahrs" <Juergen.Kah...@vr-web.de> wrote in message
news:2sdht7F...@uni-berlin.de...
Chris Cox
<jo.t...@wanadoo.fr> wrote:
> Actually, it is Alan Taylor, for some reason the messages come through in my
> wife's name.
>
> Scientific sources tend to rely on computers for their information.
> Computers, working in light, work on the RGB system. CMY(K) can be produced
> by mixing two RGB primaries as CMY are the secondaries of RGB. It is
> impossible to recreate the RBY system using a computer. Unless your
> scientific sources received teaching in 'Art', as distinct from 'Science',
> there is no reason why they should be aware of RBY.
>
> The teacher in the example quoted was an art teacher so paint is assumed to
> be the medium in question and thus RBY. Had the teacher been in a school of
> printing, CMY would have been the correct answer, just as RGB would have
> been correct if the teacher had been concerned with computers, television,
> stage lighting, etc..
Even in paint, the primaries are CMY, not Red Blue Yellow.
Other than bad teaching (as in this example), there is no difference in
color between art and science.
Chris
Jo Taylor
printers inks or light, it is generally understood one is referring to
non-reversable opaque oil or acrylic paint. This may be mixed on the palette
or on the support in such a way that red and yellow produce orange, blue and
yellow produce green, etc.. If one of the colours is applied to the support
and allowed to dry then a second colour cannot be mixed with it. The red
applied to a dry yellow does not produce orange but obliterates the yellow.
Water colour is different in that a dried layer can be reconstituted with
water and another colour but the basic facts remain the same - red, blue and
yellow are the primaries in paint.
Gouache can indeed be anything between transparent and opaque just as
acrylic, water colour and oil colour can but that is not the issue. 'A kind
of blue' is not pure blue and pure blue cannot be produced by mixing any
paints.
'Artists don´t use just three "primary" colors' is also not correct. I am an
artist and am working on a series exploring exactly what can be achieved
using only the three primary colours. I could buy ready mixed tubes of
Cobalt Turquoise, Magenta or any one of hundreds of other colours but these
would have no family relationship to the three primaries I am already using.
I prefer to mix my own versions from my three primaries plus judicious use
of black and white.
'They had learned in centuries to mix plenty pigments (organic, mineral,
chemical).' This is indeed true but only because achieveing a pure red, blue
and yellow containing compatible ingredients that could produce a purple,
green and orange was extremely difficult and time consuming prior to the mid
19thC. Today, one can rely on any of many reputable manufacturers to produce
tubes of compatible paint primaries and these are red, blue and yellow.
Regards, Alan Taylor
> Alan,
>
> according to my wife, Gouache paint can be anything between
> opaque and transparent. Depends on the layer thickness.
> A fast test shows: CobaltTurqoise + Magenta delivers a kind
> of blue.
> As already stated in my previous post: artists don´t use just
> three "primary" colors. They had learned in centuries to mix
> plenty pigments (organic, mineral, chemical).
>
> Best regards --Gernot Hoffmann
Jo Taylor
am an artist, not a scientist. As far as I am concerned, yellow is a primary
colour, defined (as many thousands of books will confirm) as a colour which
cannot be produced by the mixing of other colours. Such a statement is based
on the traditional and accepted medium of opaque paint - the fluid stuff in
tins, buttery stuff in tubes or solid stuff in little blocks. Within that
context, and I stress 'within that context', yes, paint means 'an opaque
layer' and, yes, 'yellow is a primary'.
It must be considered that 'paint' (I use the word in its widest sense) has
been in use for many thousands of years and the RBY system has been known
and taught for a similar period. RGB has only existed since a sufficiently
strong source of light has been capable of being projected through coloured
filters - the word 'electricity' comes to mind. CMYK was still in its
infancy in the 1930s. Both RGB and CMYK are very new ways of understanding
colour and are not part of the traditional process of making paintings.
And just in case you want to mention stained glass, I am a stained glass
artist and, despite coloured glass being available since the 11th century,
nobody, to my knowledge, has noticed that colour of the image created by the
sun shining through a red panel and a green panel in an single overlapped
location produces yellow. This could be because of the single source of
light and the windows being flat - no chance of overlays - hence the
reference to electricity when the possibility of projecting two or three
coloured light beams onto a single spot became reality.
Regards, Alan
> Alan,
>
> interesting discussion. Is it really safe to say: "paint means always
> an opaque layer" ?
> About the primaries: without knowing the reflectance spectra it´s quite
> useless to guess about the available gamut.
> For example yellow:
> If it is a more or less spectral yellow then no red can be mixed.
> If it is a metameric red+green yellow then red can be mixed.
> Therefore a statement like "yellow is a primary" is meaningless, IMO.
>
> Best regards --Gernot Hoffmann
Gerhard Fuernkranz
>[...] As far as I am concerned, yellow is a primary colour, defined (as many thousands of books will confirm) as a colour which cannot be produced by the mixing of other colours.
>
Watching the discussion for a while I've got the feeling, that the
actual problem seems to be rather the lack of an agreed, mathematical
definition, specifying which physical properties a color must fulfil in
order that it may be called "primary color".
Asking Google, I found http://www.wordiq.com/definition/Primary_color saying
"A primary color is a color that cannot be created by mixing other
colors in the gamut of a given color space. Primary colors may
themselves be mixed to produce most of the colors in a given color
space. [...]"
Basically this definition corresponds to your words. But IMO the
important issue in the above definition are the words "IN THE GAMUT OF A
GIVEN COLOR SPACE".
The consequence is, there does not only exist a single, fixed set of
primary colors, but for each combination of a particular color space
with a particular color mixing model (which is basically implied by the
physical characteristics of the used color reproduction process), there
exists a different set of primary colors.
Another implication of this definition is, that the set of primary
colors is not limited to three ones. You will be able to find color
space + mixing model combinations, where the set of primary colors is
even INFINITE (e.g. if I choose an additive mixing model, and choose the
complete CIE horseshoe as color space, then according to the above
definition, ALL visible monochromatic spectral colors are primary colors
for this color space, since they cannot be mixed from other colors in
the color space, using the given mixing model).
Or conversely, if the set of primary colors and a color mixing model are
given, then these primaries and the mixing model determine the
achievable color space gamut.
And if a painter mixes his colors from, say, twenty or fifty highly
saturated pigments, then it is IMO pretty likely, that MANY of the used
pigment colors become also primary colors for this particular painting
process, according to the above definition (and not just
Cyan+Magenta+Yellow, or Blue+Red+Yellow). Each additionally used color
pigment which increases the achievable gamut of the painting process,
becomes an additional primary color for this particular painting process.
On the other hand, you will surely also be able to define a color space,
where particular blue, red and yellow colors are the only primary
colors, or a different color space, where particular CMY colors are the
only primaries (But these particular color spaces will likely have a
smaller gamut than the color space you can achieve by using fifty
different pigments).
My conclusion:
If P is the set of the used colorants (pigments) for painting, then the
set of primary colors for this particular painting process is the set of
all colors x, where x is an element of P, and where x cannot be created
by mixing any combination of other colorants y <> x, which are also
elements of P. And depending on the used pigments, this set is not
necessarily CMY or BRY, but very likely the particular painting process
will have rather more than 3 primary colors, if many saturated pigments
are used, whose colors cannot be mixed from any combination of the other
used pigments.
Regards,
Gerhard
Robin Myers
> <snip>
>
> It must be considered that 'paint' (I use the word in its widest sense) has
> been in use for many thousands of years and the RBY system has been known
> and taught for a similar period. RGB has only existed since a sufficiently
> strong source of light has been capable of being projected through coloured
> filters - the word 'electricity' comes to mind. CMYK was still in its
> infancy in the 1930s. Both RGB and CMYK are very new ways of understanding
> colour and are not part of the traditional process of making paintings.
>
> And just in case you want to mention stained glass, I am a stained glass
> artist and, despite coloured glass being available since the 11th century,
> nobody, to my knowledge, has noticed that colour of the image created by the
> sun shining through a red panel and a green panel in an single overlapped
> location produces yellow. This could be because of the single source of
> light and the windows being flat - no chance of overlays - hence the
> reference to electricity when the possibility of projecting two or three
> coloured light beams onto a single spot became reality.
There were examples in the 19th century of people putting red, green and blue filters on
"magic lanterns" and projecting monochrome images through each filter to produce color
images on the screen (blending several images on the screen). They experimented with
various color filters, struggling to find the ones to give good color. The lanterns were
not electric powered, but powered by various burning fuels. Check some of the books on the
history of photography for more details.
Robin Myers
Jo Taylor
in use - RBY, RBG and CMYK - and each has its own set of primaries.
--
regards, alan
Gerhard Fuernkranz
>This is effectively what I said in the first place. There are three systems
>in use - RBY, RBG and CMYK - and each has its own set of primaries.
>
This is at least my understanding according to the definition I
have found. I would even say there are not only three systems,
but there may exist an infinite number of different color
reproduction processes, and each of them may use a different
set of primaries (which is not limited to three elements).
Or if applied to painting, I'd say that an artist may choose
different primary colors for different artworks, depending on
the desired gamut of the particular painting.
(In practice I guess this is more complicated, since the artist
probably simply uses 20 or 50 colorants, without caring too
much about primary colors, and I guess he does not always know,
which of these colorants he could have omitted (and mixed colors
in a different way), without need to renounce some areas of the
artwork's gamut)
Regards,
Gerhard
Jo Taylor
that complicated. Within the realms of being a painter, I could use 'a red'
, 'a blue' and 'a
yellow' and still mix a range of greens, purples and oranges plus a range of
browns. Any of the primaries could have a bias towards one or the other
primary but the result would be the same. A brick red, for example, has a
yellow rather than a blue bias. It also contains some of the complimentary
(green) making it more brown and could also include some black and white or
both (grey). The same could apply to the blue and the yellow but it would
not alter the fact that the chosen red and blue would create a unique
purple, the chosen red and yellow a unique orange or the chosen blue and
yellow a unique green. This is the essence of a harmonious range of colours
which can be produced using any red, blue and yellow plus black and white.
No artist needs more than these five for any painting. Having a wide palette
just leads to a non-harmonious composition and is the basic mistake most
people make when starting out.
--
regards, alan
Chris Cox
<jo.t...@wanadoo.fr> wrote:
> Sorry but I have to disagree with this. Red, blue and yellow are the
> primaries in paint and are defined as those colours that cannot be made by
> mixing others. Cyan, magenta and yellow are the primaries in printers inks.
You can disagree all you want, and you'll still be wrong.
Someone was lazy and told you that the primaries were Red, Yellow Blue
instead of explaining the idea of Cyan Maganta and Yellow. But that
over-simplified explanation doesn't make it the truth.
Red, Yellow, Blue paint can only make Magenta, Cyan, and Green with
greatly reduced saturation - which is a hint that Red and Blue are not
primaries in paint.
Cyan, Maganta, Yellow can make Red, Blue, and Green with saturation
similar to the original pigments - which is a good indication that they
are primaries.
If you want to use 3 primaries, then you can either add light (Red,
Green, Blue), or subtract light (Cyan, Magenta, Yellow). There is no
magical "other" set of primaries for those conditions (and the fact
that paint is opaque doesn't enter into the problem)
Chris
Jo Taylor
Cyan, Magenta and Yellow are the primaries in transparent printers' inks.
Red, Blue and Green are the primaries in light and Red, Blue and Yellow are
the primaries in paint.
No one was lazy. Red, Blue and Yellow paint can produce any colour desired
and, with the addition of white, can also produce Cyan and Magenta - tones
of blue and red. I did a test yesterday, would you like to see a copy?
> Cyan, Maganta, Yellow can make Red, Blue, and Green with saturation
> similar to the original pigments - which is a good indication that they
> are primaries.
I quite agree, but this is within the CMYK system, I was referring to RBY.
> If you want to use 3 primaries, then you can either add light (Red,
> Green, Blue), or subtract light (Cyan, Magenta, Yellow). There is no
> magical "other" set of primaries for those conditions (and the fact
> that paint is opaque doesn't enter into the problem)
>
I would suggest you purchase a tube of good quality opaque artists acrylic
in each of red, blue and yellow, a large piece of white card, a brush, a jug
for the water and a table and chair.
> If you want to use 3 primaries, then you can either add light (Red,
> Green, Blue), or subtract light (Cyan, Magenta, Yellow). There is no
> magical "other" set of primaries for those conditions (and the fact
> that paint is opaque doesn't enter into the problem)
This only refers to RGB and CMYK. You appear to be unaware of RBY. Most of
us learned this in primary school then went on to other fields. I have spent
fifty years studying colour from an artist's point of view and I can assure
you that RBY exists and you have a long way to go before telling me I am
wrong.
Alan Taylor
"Chris Cox" <cc...@mindspring.com> wrote in message
news:241020041748225831%cc...@mindspring.com...
Don Klipstein
>Sorry, you are wrong. There are three systems in use, RBY, RGB and CMYK.
>Cyan, Magenta and Yellow are the primaries in transparent printers' inks.
>Red, Blue and Green are the primaries in light and Red, Blue and Yellow are
>the primaries in paint.
>
>No one was lazy. Red, Blue and Yellow paint can produce any colour desired
>and, with the addition of white, can also produce Cyan and Magenta - tones
>of blue and red. I did a test yesterday, would you like to see a copy?
I sure would. I clearly remember that back when I was in first grade or
whatever I was unable to make a decent green with yellow and blue paint.
- Don Klipstein (d...@misty.com)
Sergey Tolstov
"Gerhard Fuernkranz" <nosp...@gmx.de> wrote in message
news:417049A1...@gmx.de...
> Jo Taylor schrieb:
> even INFINITE (e.g. if I choose an additive mixing model, and choose the
> complete CIE horseshoe as color space, then according to the above
> definition, ALL visible monochromatic spectral colors are primary colors
> for this color space, since they cannot be mixed from other colors in
> the color space, using the given mixing model).
Gerhard,
This is not true. There will be only a part of the spectral colors, which
can not be reproduced by the positive mixture of, for example, the
monochromatic r (700 nm), g (546 nm) and b (436 nm) colors.
I suppose, that those colors, which can not be reproduced by the positive r,
g, b mixture, can be reproduced by the addition of a few (two or three)
different monochromatic colors.
Thanks,
Sergey Tolstov
Robin Myers
"Jo Taylor" <jo.t...@wanadoo.fr> wrote:
> Sorry, you are wrong. There are three systems in use, RBY, RGB and CMYK.
> Cyan, Magenta and Yellow are the primaries in transparent printers' inks.
> Red, Blue and Green are the primaries in light and Red, Blue and Yellow are
> the primaries in paint.
I think this thread is getting a bit tangled. There are two methods of
producing color: additive and subtractive. Think of these as two ends of
a continuum. At the purely additive end is mixing light producers, at
the purely subtractive end is mixing light absorbers.
What everyone is confusing is that at the subtractive continuum end
there are an infinite number of light absorbers that can be used to
produce colors. The choice of absorbing primaries controls the color
gamut.
Mr. Taylor has been arguing for RBY subtractive primaries that are
completely opaque. Fine, they work! Others are suggesting CMY or CMYK
primaries that may or may not be completely opaque. They work too!
Everyone is right! You can make any of these systems produce color
mixtures. The size and extent of the GAMUTS produced will vary. Colors
achievable in one system may or may not be achievable in another.
Light absorbers also exist on a continuum of totally transparent to
totally opaque. Their position along this continuum can often be changed
by additives or their usage. Opacity is not a requirement for
subtractive color primaries, but it is one of their attributes.
Primaries are not necessarily always at one point along the production
continuum. The choice of primaries and the manner in which they are
applied will control the amount of additive and subtractive color
production. For instance, some paints can be employed in a purely
subtractive process, and yet the same paints can be subtractive at a
micro level and additive at a macro level, e.g. the pointillistic works
of Seurat and Maxwell's color disks.
The most skillful artists, engineers and scientists understand and use
these continuums to achieve their color goals.
Robin Myers
Note: Yes, I am taking license with the mathematical definition of a
continuum. Here I am using a more conceptual definition merely to
describe a line segment with an infinitely variable position between the
two ends.
Gerhard Fuernkranz
>>even INFINITE (e.g. if I choose an additive mixing model, and choose the
>>complete CIE horseshoe as color space, then according to the above
>>definition, ALL visible monochromatic spectral colors are primary colors
>>for this color space, since they cannot be mixed from other colors in
>>the color space, using the given mixing model).
>>
>
>Gerhard,
>
>This is not true. There will be only a part of the spectral colors, which
>can not be reproduced by the positive mixture of, for example, the
>monochromatic r (700 nm), g (546 nm) and b (436 nm) colors.
>
Sergey,
yes, with one prticular triple of additive primaries you can reproduce
MANY chromaticities within the horseshoe, but NOT ALL.
>I suppose, that those colors, which can not be reproduced by the positive r,
>g, b mixture, can be reproduced by the addition of a few (two or three)
>different monochromatic colors.
>
Sure, but you won't be able to find any SINGLE triple of primaries,
which can reproduce the WHOLE gamut of the horseshoe (up to the borders,
and including all visible monochromatic spectral colors), by additively
mixing (positive amounts) of these three primaries only.
Regards,
Gerhard
Sergey Tolstov
Gerhard,
I do not object to the fact that a single triple of primaries can not
reproduce the whole gammut of visible colors. I am just saying that your
example of infinite number of primaries is possibly incorrect - there should
be a FINITE set of monochromatic primaries, which, in a mixture, can
reproduce other monochromatic spectral colors.
Thanks,
Sergey
Gerhard Fuernkranz
>Gerhard,
>
>I do not object to the fact that a single triple of primaries can not
>reproduce the whole gammut of visible colors. I am just saying that your
>example of infinite number of primaries is possibly incorrect - there should
>be a FINITE set of monochromatic primaries, which, in a mixture, can
>reproduce other monochromatic spectral colors.
>
Sergey,
I would call the CIE horseshoe more or less a 2D-convex set of (x,y)
chromaticities, with a CURVED hull, do you basically agree?
IMO there does not exist any polygon with a finite number of vertices,
which can completely enclose a 2D-convex set, whose hull is curved, and
not exclusively composed of straight-line segments, given the additional
constraints that no vertex of the polygon must be outside the convex set.
But of course, if we APPROXIMATE the standard observer CMFs with
discrete samples, say at 1nm intervals, and APPROXIMATE the hull of the
horseshoe accordingly as a polygon, then indeed a finite set of
primaries can reproduce all colors in this polygon (actually, in this
case the polygon vertices implicitly become the primaries needed to
reproduce this gamut).
Regards,
Gerhard
Sergey Tolstov
Is not it, the CIE ideal observer horseshoe is a table of values? If yes,
then the table contains only limited number of rows :) Plus, this observer
is an ideal, which will match any color without an error. I have some points
on this matter, you are welcome to contribute to the new thread "primary
color" about the subject.
Thank you,
Sergey
Gerhard Fuernkranz
>Gerhard,
>
>Is not it, the CIE ideal observer horseshoe is a table of values? If yes,
>then the table contains only limited number of rows :)
>
Sergey,
right, that's the practical problem/limitation with this issue - any
table has only limited accuracy, and we won't be able to reconstruct out
the actual shape of the curve between the samples exactly :-) So in
practice, depending, on how we interpolate the table (e.g. linear vs.
spline), the hoerseshoe will either be approximated as polygon or as
curve (and we won't ever know the exact shape betwen the points).
-Gerhard
Jo Taylor
that one opaque colour cannot be overlaid on another, dry, opaque colour to
produce a third colour. In CMYK and RGB the colours are transparent and a
third colour can be made with overlay - be it subtractive or additive. The
only way a third colour can be made with opaque paint is by pre-mixing prior
to application, something not possible with a computer.
Regards, Alan Taylor
"Robin Myers" <ro...@rmimaging.com> wrote in message
news:robin-AFA558....@tribune.sj.sys.us.xo.net...
Jo Taylor
AT
"Don Klipstein" <d...@manx.misty.com> wrote in message
news:slrncnr1r...@manx.misty.com...
Erik Nikkanen
Just a thought on the practical limitations.
Which error would be larger? The error between the points of a non exact
shape and the true shape or the error between a real observer and the
standard observer?
I hope you understand what I mean. I am not sure I have stated the
question properly.
Erik
Don Klipstein
>
>"Gerhard Fuernkranz" <nosp...@gmx.de> wrote in message
>news:417049A1...@gmx.de...
>> Jo Taylor schrieb:
>
>> even INFINITE (e.g. if I choose an additive mixing model, and choose the
>> complete CIE horseshoe as color space, then according to the above
>> definition, ALL visible monochromatic spectral colors are primary colors
>> for this color space, since they cannot be mixed from other colors in
>> the color space, using the given mixing model).
>
>Gerhard,
>
>This is not true. There will be only a part of the spectral colors, which
>can not be reproduced by the positive mixture of, for example, the
>monochromatic r (700 nm), g (546 nm) and b (436 nm) colors.
A large enough majority of all colors can be reproduced by positive
combinations of these three.
An even larger majority can be reproduced if the green wavelength is
shortened slightly and the blue wavelength is lengthened slightly.
(Perhaps by having the green at 532-535 nm and the blue at 445-450 nm).
Red shortened to the mid 600's compromises little while having a
significant gain in luminous efficacy.
Although this fails to completely reproduce all spectral colors of
wavelengths other than the primaries, most can be adequately approximated.
Along with achievement of matching of a large majority of non-spectral
colors.
As for what's special about 436, 546 and 700 nm? 435.8 and 546.1 are
very strong spectral lines of mercury. And CIE has defined 700 to be as
red as red gets (reality is close enough).
- Don Klipstein (d...@misty.com)
Danny Rich
that the metre has only the resolution of the marks on the stick? The metre
is defined by a quantum mechanical function - the atomic transitions
involved in a certain LASER emission. The standard observer is defined as
the average of a number of color matching functions. These are continuous
functions which have been sampled at only a limited number of points. The
Color Matching Functions are not merely tabulated values any more than the
metre is limited by the number of marks on a rule.
The reason that a polygon must be used to approximate the continous
curvature of a chromaticity diagram is that two primary colors mix in
straight lines. Thus to match 541.05nm you need to use primaries of
wavelength 541.04nm and 541.06nm and the more significant figures one uses
the narrower the gap becomes. There is no way to place a primary on the
other side of the diagram (380nm or 780nm) and mix any wavelength of light
with it to create the match to 541.05nm.
The issue of experimental error is not relevant. While an idividual
observer may be different, the large number of studies carried out over the
years, around the world, have shown that the average of a reasonable (more
than 10) number of observers is very consistent. The CIE Color Standard
Observer is not a theoretical concept it is the results of real color
matches by real observers, averaged and transformed to an imaginary primary
set which is capable of matching all real colors - as well as a large number
of non-real colors.
Danny Rich
Jo Taylor
http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/Primary%20S
amples.p65
http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/TestTubesFr
ont.jpg
http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/TestTubeLib
raryCard2.jpg
http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/Colourwheel
Real%20Colour.jpg
Only three colous were used - primary red, primary blue and primary yellow.
The darker versions of each colour used black and the lighter versions used
white.
AT
Erik Nikkanen
Danny Rich wrote:
>
> The issue of experimental error is not relevant. While an idividual
> observer may be different, the large number of studies carried out over the
> years, around the world, have shown that the average of a reasonable (more
> than 10) number of observers is very consistent.
Danny,
So, the average is very consistent. What about the standard deviation of
all those individuals or groups of individuals? Would a group of women
have the same basic average but maybe a smaller standard deviation within
their group compared to a group of men? Would it also be true that the
average may be consistent between different cultural groups but their
preferences for specific colours could be different?
Thanks for your comments.
Erik
Gerhard Fuernkranz
>[...] The standard observer is defined as the average of a number of color matching functions. These are continuous functions which have been sampled at only a limited number of points. [...]
>
Danny,
like always, your comments are very appreciated. So it looks like you
agree, that the standard observer CMFs are actually continuous functions.
>The reason that a polygon must be used to approximate the continous curvature of a chromaticity diagram is that two primary colors mix in straight lines. Thus to match 541.05nm you need to use primaries of wavelength 541.04nm and 541.06nm and the more significant figures one uses the narrower the gap becomes. There is no way to place a primary on the other side of the diagram (380nm or 780nm) and mix any wavelength of light with it to create the match to 541.05nm.
>
I think I basicaly understand and agree, but please forgive me my
ignorance, if I still do not understand why the latter implies that a
polygon must be used to approximate the curvature, and why I could not
for instance do an alternative approximation in the following way:
- take the sampled data points of the standard observer CMFs
- transform these points to the (x,y) chromaticity space to
obtain a sampled version of the horseshoe curve
- fit a curve (for instance splines) through the mapped data points
in (x,y) space to obtain a smooth, continuous approximation for
the actual horseshoe curvature, instead of simply connecting the
points with straight line segments
or alternatively,
- fit e.g. splines through the sampled data points
of the standard observer CMFs
- transform the resulting smooth, continuous curves
to x,y space, in order to obtain a smooth approximation
of the horseshoe curvature
What exactly would not work or would be wrong, if I would approximate
the chromaticity diagram in this way, i.e. as a smooth, continuous
curve, and not as polygon?
Sure, if my goal would be to approximate the chromaticity which
corresponds to 541.05nm, by additively mixing two primaries with
541.04nm and 541.06nm, then the closest match can only lie somewhere on
the straight line segment between 541.04nm and 541.06nm.
But if I want to approximate the actual, continuous shape of the
spectral locus between 541.04nm and 541.06nm, then I guess, that splines
passing through these two points can match the actual curvature (over
the whole 0.02nm range) better than a simple straight line segment.
Where is my error in reasoning?
Thanks,
Gerhard
Mark Jackson
> I think I basicaly understand and agree, but please forgive me my
> ignorance, if I still do not understand why the latter implies that a
> polygon must be used to approximate the curvature
Because a polygon is, physically, what you *get* from color mixing with
a finite number of primaries. When Danny wrote:
> > The reason that a polygon must be used to approximate the continous
> > curvature of a chromaticity diagram is that two primary colors mix in
> > straight lines.
he was stating a characteristic of x, y space.
You can fit splines or anything else between points, but the only curve
that represents what can be achieved by actual additive mixing of the
colors represented by two points is a straight line between them.
--
Mark Jackson - http://www.alumni.caltech.edu/~mjackson
When my information changes, I change my opinion.
What do you do, Sir? - John Maynard Keynes
GraemeGill
> What everyone is confusing is that at the subtractive continuum end
> there are an infinite number of light absorbers that can be used to
> produce colors. The choice of absorbing primaries controls the color
> gamut.
Right, and because (loosely) R, G & B are the three primaries that create
the largest gamut in an additive system, their complements (C, M & Y) will
produce the largest gamut in a subtractive system.
Graeme Gill.
Mark Jackson
[quoting Don Klipstein]
> > I sure would. I clearly remember that back when I was in first grade or
> > whatever I was unable to make a decent green with yellow and blue paint.
> Here are the addresses:
>
> http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/Primary%20S
> amples.p65
>
> http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/TestTubesFr
> ont.jpg
>
> http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/TestTubeLib
> raryCard2.jpg
>
> http://groups.yahoo.com/group/BritArt/files/Alan%20-%20Technical/Colourwheel
> Real%20Colour.jpg
Can you make these available in such a way that one can see them
without petitioning for membership in the Yahoo "BritArt" group? Just
one will do, if it shows your RBY "primaries" and a decent green.
Gerhard Fuernkranz
>Gerhard Fuernkranz <nosp...@gmx.de> writes:
>
>>I think I basicaly understand and agree, but please forgive me my
>>ignorance, if I still do not understand why the latter implies that a
>>polygon must be used to approximate the curvature
>>
>Because a polygon is, physically, what you *get* from color mixing with
>a finite number of primaries.
>
This is the part which is perfeclty clear. Two primaries mix additively
in a straight line, and the gamut of N additive primaries is a poygon
(i.e. the convex hull of the primaries) in chromaticity space.
> You can fit splines or anything else between points,
That's what I'm trying to say.
> but the only curve that represents what can be achieved by
> actual additive mixing of the colors represented by two points
> is a straight line between them.
I fully agree, but I don't see why this is relevant or even a MUST, if
the goal is just to approximate the continous curvature of the horseshoe
as close as possible?
IMO the points at which the the standard observer CMFs have been sampled
do not have any special meaning like dedicated primaries, they are just
arbitrary points on the curve, and as Danny said, it just happened that
the CMFs have been sampled at these wavelengths. So why must I treat the
sampled points like primaries, and why must I stringently attempt to
approximate all missing points between two samples by *additive mixing*
of the chromaticities of the sampled points?
I could imagine that an approximation with e.g. splines can probably
match the acutual continuous curvature of the horseshoe more accurately
than a polygon, and it would even preserve the special property of the
spectral locus, that the whole gamut enclosed by the horeseshoe cannot
be covered with a finite number of additive primaries (a polygon
approximation on the other hand would not preserve this property).
Regards,
Gerhard
Danny Rich
"Erik Nikkanen" <nikk...@globalserve.net> wrote in message
news:417F93C2...@globalserve.net...
North Americ which have very diverse gene pools have a wider distribution of
"normal" observers than say only women or even some oriental or aboriginal
populations with very limited gene pools.
Mark Jackson
> Mark Jackson schrieb:
> > but the only curve that represents what can be achieved by
> > actual additive mixing of the colors represented by two points
> > is a straight line between them.
>
> I fully agree, but I don't see why this is relevant or even a MUST, if
> the goal is just to approximate the continous curvature of the horseshoe
> as close as possible?
I think we are tripping up over two uses of the verb "approximate." I
gather from the above you are using it to mean "draw, or describe
mathematically, a real continuous curve which is only tabulated, and
for that matter has only been measured experimentally, at a finite
number of points." Previously I thought you were using it in the way I
am almost certain that Danny was using it in the message you originally
quoted: "Approach as close as possible to the limits of the
chromaticity diagram with the gamut of a (necessarily finite) set of
primaries."
For the first, of course, there's no reason to limit oneself to linear
interpolation, if the set of nearby points suggests a curve would be
better.
Danny Rich
the splines. The points are connected by straight lines - connected at a
small interval. Of course, many book and web displays of the diagram have
used graphical tools like splines to make the diagram look "smooth".
gerhard.fuernkranz
Danny Rich schrieb:
> The horseshoe diagram is, in fact, exactly what you have described - without
> the splines. The points are connected by straight lines - connected at a
> small interval. Of course, many book and web displays of the diagram have
> used graphical tools like splines to make the diagram look "smooth".
Danny,
do you mean, the horseshoe diagram is just *by definition* a polygon,
connecting the sampled points, and not a smooth, continuous curve,
although the standard observer CMFs, from which the horseshoe diagram is
derived, are actually continuous functions (as you stated previously),
and although an exact transformation of these continuous standard
observer CMFs to chromaticity space would result in a smooth, continuous
horseshoe curvature?
Is there also a particular sampling interval defined, which must be used
by definition in oder to create "the" horseshoe polygon? - otherwise a
definition as polygon would IMO be ambiguous either (I know standard
observer tables at different sampling intervals, however, is there also
an "official" interval?).
Thanks,
Gerhard
Sergey Tolstov
"Danny Rich" <Dann...@softhome.net> wrote in message
news:NcEfd.26919$bz4.6...@news4.srv.hcvlny.cv.net...
Danny,
To obtain the CMFs, some procedure were applied to a set of experimental
data. They could easily represent their data as a spline function, not a
table. If they would be splines, the horseshoe would be some smooth curve
also.
Do you agree? Or you mean, that the CMFs are not smooth functions by their
nature?
Whatever they would use for representation of the CMFs, the initial
unprocessed data are measured at some interval, the data contain errors.
This may be a not relevant issue, but that means that the standardized data
are themself imprecise and sould be taken seriously only to a few digits.
The good thing with the repeatability error is, that it may help to limit
primary color number for practical application.
I believe, it possible to select a finite number of the primary colors to
cover the whole gamut in a such way, that any color, which lie outside of
the obtained gamut would be possible to match in such a way, that the color
difference would not be possible to notice.
Thank you,
Sergey
Jo Taylor
AT
"Mark Jackson" <mjac...@alumni.caltech.edu> wrote in message
news:clr8h5$t0q$1...@news.wrc.xerox.com...
Mark Jackson
> "Mark Jackson" <mjac...@alumni.caltech.edu> wrote in message
> news:clr8h5$t0q$1...@news.wrc.xerox.com...
> > Can you make these available in such a way that one can see them
> > without petitioning for membership in the Yahoo "BritArt" group? Just
> > one will do, if it shows your RBY "primaries" and a decent green.
> I could send an attachment to this group. Is this allowed, possible?
No. No binaries in non-binary newsgroups, please.
If you have no other webspace, and it's no bigger than a megabyte or
so, you can email it to me and I will make it available for a while on
my website.
Danny Rich
"Sergey Tolstov" <sa...@yahoo.com> wrote in message
news:UOvgd.34$pY6.25@trnddc04...
color matches to continuum of spectral lights. The color matching functions
are smooth functions and were fitted, not with a mathematical spline which
had not yet been invented, but with a fifth order osculatory intepolation
function. This math is more like a local Lagrangian interpolation - which
is, at best an incomplete spline function. The CMF were transformed using
linear algebra and forcing the alychne to lie coincident with the Z=0 plane
and normalized so that the spectral luminous efficacy was constant across
the diagram. In that space, by defintion, all color mixtures lie on
straight lines. Thus any attempt to match colors along the spectrum locus
will require straight lines. You can connect those points with any
functions you like but only the linear term will be realistic. The diagram
was designed to function that way.
> Whatever they would use for representation of the CMFs, the initial
> unprocessed data are measured at some interval, the data contain errors.
> This may be a not relevant issue, but that means that the standardized
> data
> are themself imprecise and sould be taken seriously only to a few digits.
>
> The good thing with the repeatability error is, that it may help to limit
> primary color number for practical application.
> I believe, it possible to select a finite number of the primary colors to
> cover the whole gamut in a such way, that any color, which lie outside of
> the obtained gamut would be possible to match in such a way, that the
> color
> difference would not be possible to notice.
>
This is also not strictly true - because the distances in the diagram are
deceptively small. The jnd for wavelength discrimination was mapped by W.
D. Wright in the 1940s and can be found in his textbook and in Color Science
by Wyszecki and Stiles. The standard deviation of color matching in the
diagram was first mapped by MacAdam and later by Brown then Wyszecki &
Fielder. The three sets of data are in very good agreement. Because of the
non-uniformity of the diagram, the size of a jnd varies with location but
the smallest values are less than 0.001 in chromaticity (x,y). Even
allowing a growth of 5x you will be hard pressed to find three widely spaced
points that will not leave regions larger than 0.005 in the diagram.
> Thank you,
> Sergey
>
>
Danny
Sergey Tolstov
I understand, that the diagram is not a CMF. I do not understand why it
should be represented by a polygon, if the CMFs are smooth functions. Is not
it, the chromaticity diagram is directly calculated from the CMFs? If you
say the CMFs are smooth, the horseshoe should be smooth also.
This is OT, but I am curious, what do you mean mathematical splines had not
been invented?
> Danny
I do not mean three points. I mean more than three and no negative color
coordinates.
Thank you,
Sergey
Gernot Hoffmann
> The horseshoe diagram is not a color-matching function. It is the locus of
> color matches to continuum of spectral lights. The color matching functions
> are smooth functions and were fitted, not with a mathematical spline which
> had not yet been invented, but with a fifth order osculatory intepolation
> function. This math is more like a local Lagrangian interpolation - which
> is, at best an incomplete spline function. The CMF were transformed using
> linear algebra and forcing the alychne to lie coincident with the Z=0 plane
> and normalized so that the spectral luminous efficacy was constant across
> the diagram. In that space, by defintion, all color mixtures lie on
> straight lines. Thus any attempt to match colors along the spectrum locus
> will require straight lines. You can connect those points with any
> functions you like but only the linear term will be realistic. The diagram
> was designed to function that way.
>
Danny,
Table I(3.3.1) in W&S shows that the original data are given in steps of
5nm typically with 6 digits + 1..3 zeros (like 0.001-368-000).
Interpolated data in steps of 1nm have 8 or 9 digits (like 0.001-368-050).
Using splines (as I do) instead of Lagrange for the 5nm table will hardly
deliver anything different, as long as we are talking about relevant
digits.
The chromaticity curve can be interpolated by splines as well (as I do),
simply based on the assumption that human vision is better described by
a continuous model instead of a discrete one (polygonal locus).
I“m always surprised about the many digits. Measuring lengths in a real
environment (ZEISS 3D coordinate measuring machine) has an uncertainty
of 2 micrometers in a working space cube of 0.5m edge length. That means
5..6 significant digits. Much worse for unfavourable geometries.
The same for operational amplifiers. 10 microvolt accuracy for the range
0..10 V cannot be achieved easily.
Measuring colors should be much more accurate ? I don“t think so.
In this sense the interpolation method is not the least of any practical
importance.
Best regards --Gernot Hoffmann
Sergey Tolstov
not
> been invented?
Gernot Hoffmann
>
> I understand, that the diagram is not a CMF. I do not understand why it
> should be represented by a polygon, if the CMFs are smooth functions. Is not
> it, the chromaticity diagram is directly calculated from the CMFs? If you
> say the CMFs are smooth, the horseshoe should be smooth also.
> This is OT, but I am curious, what do you mean mathematical splines had not
> been invented?
>
Sergey,
of course the chromaticity diagram is calculated directly from the CMFs:
For the horseshoe contour (spectral locus):
X = x-bar(lambda)
Y = y-bar(lambda)
Z = z-bar(lambda)
x = X / (X+Y+Z)
y = Y / (X+Y+Z)
About splines:
Hermite (?)
Natural spline: Rutishauser (1960)
Catmull-Rom (1972)
Kochanek-Bartels (1984)
Bézier curves (normally not called splines) (1972)
Best regards --Gernot Hoffmann
Roger Breton
Are they "splines" nevertheless?
Regards -- Roger Breton
Sergey Tolstov
Could you post your answer, mentioning 16 colors, to the newsgroup. I've got
it in my mailbox.
Do you mean those 16 primaries are enough to match all horseshoe?
Thank you,
Sergey
"Danny Rich" <Dann...@softhome.net> wrote in message
news:W9Ygd.24317$UC4.10...@news4.srv.hcvlny.cv.net...
Gernot Hoffmann
> > Bézier curves (normally not called splines) (1972)
>
> Are they "splines" nevertheless?
>
> Regards -- Roger Breton
Roger,
2D splines and Bézier curves are piecewise parametric cubic
polynomials for t=0 to t=1: x(t), y(t).
Splines are used to connect many points by pieces of curves with
C(n) continuity.
n=0: same point, n=1: also same tangent, n=2: also same curvature.
A natural spline is the most generic spline: connect all points
by an elastic rod, which results in minimal curvature (that´s
the method I´m using for color graphics).
Béziers are used to define one curve by two endpoints and two
endpoint tangents. Connecting several curves with C(1) continuity
is up to the user. In fact the smooth replacement of a polyline
doesn´t have a unique solution.
Béziers are the common tool in vector graphic drawing programs.
Some tutorials for natural splines and Béziers are on my website:
http://www.fho-emden.de/~hoffmann/howww41a.html
Best regards --Gernot Hoffmann
Danny Rich
you have enough information to match the spectral curve of the material -
that match now becomes a strictly optical physics match and not a perceptual
match. Those 16 points are located every 20nm from 400 to 700 nm and have
been shown to span the space of most colorant spectral signatures and very
many lamp light signatures. So you no longer need the horseshoe diagram.
Danny
"Sergey Tolstov" <sa...@yahoo.com> wrote in message
news:vPPhd.964$7W.891@trnddc08...
Sergey Tolstov
> No - you mis-read my post. What I said was that if you use 16 points then
> you have enough information to match the spectral curve of the material -
> that match now becomes a strictly optical physics match and not a
perceptual
> match. Those 16 points are located every 20nm from 400 to 700 nm and have
> been shown to span the space of most colorant spectral signatures and very
> many lamp light signatures. So you no longer need the horseshoe diagram.
>
> Danny
>
Is not it, there is a perseptual match, when there is the spectral match?
The diagram will be necessary to describe the color.
Thank you for the information. I'll search for the article. Could you give
the exact reference?
Sergey