Can CMY cover all colors ? If not what does it ?

[DREW WILEY]

Gosh. How did bits or pixels get into this conversation? Even they are dependent upon basic color wheel theory, and every one of those damn devices, along with how every system of dye, pigment, or colored ink is quantitatively measured is based upon 1920's color mapping models.
The only thing the has changed is how computers now make this a relatively rapid task, versus hour after hour of separate axis plots on graph
paper. But the intermediate language behind color mapping is all basically analytic geometry.

 

[DREW WILEY]

Mark, you should take a job as a dye chemist, since in one simple post you have solved problems that no dye chemist in history has been
able to achieve! But it would help to at least mention how you could also do it at the same time as making these miracles products lightfast,
environmentally and commerically viable, and a few hundred other things that Kodak, Fuji, and Agfa people had to struggle with for decades.
And if you can indeed invent a film that can even REMOTELY compare to the range and gamut of my own eyes, please please sell me a box!

 

[Bill Burk]

Mark didn't explain how the eye can be excited to the colors outside the gamut of normal human vision... Just that detection and recording of images outside the gamut of human vision has been done...

Now as to seeing pure violet being painful... My eyes hurt when I look at an ultraviolet light source... So that makes perfect sense.

 

[DREW WILEY]

Well, as usual I was being a bit sarcastic and do understand what he is saying. But given the assumption that bees and moths and great white
sharks aren't purchasing much film or printing supplies these day, I'm trying to cut to the chase about the interface of human vision and
realistic photographic products. I've taught color matching and been one of one with designers of advanced color matching computers. I leave the software to others, but am real good a throwing curve balls and exposing the weaknesses of any cumulative system. And as ever pro color matcher knows, fine tuning ALWAYS is done by eye rather than machine. The only possible kind of exception would be a DNA/protein separator based on color, like my wife used in biotech; but that cost six millions dollars and was kept behind a true bank vault door. And as far a I know, nobody has ever been able to surpass tricolor separations on black and white film and then endlessly fiddling with the specific dyes to meet the specifications, as in true Technicolor movies or the heyday of dye transfer printing. And no "ideal' dyes
have ever been discovered for either, esp if one factors in a degree of permanence. Ordinary tripack films and papers are outright pitiful
for certain hues painters have long been capable of mixing. That's a wild goose I've been personally chasing a long long time.

 

[Bill Burk]

Yes, I think you're right that the best systems are adjusted to human vision.

And as far a I know, nobody has ever been able to surpass tricolor separations on black and white film.

You mean adding a fourth color separation didn't improve rendering? Or it was too unwieldy to be commercially viable?

I once spoke to the digital photography guru in my town (this town's not big enough for two digital photography gurus so instead of skipping town I went analog).

He was telling me he saw some future technology that would amaze me. He nodded when I suggested that it might be additional spectral sensors. I'm still waiting. As far as I know that hasn't happened yet, instead we get GPS coordinates. Heck I can do that by marking my position on a map when I take the picture.

But I noticed the gamut triangles in Wikipedia are wider than the ones in my Colortron User Manual. So somewhere there has been progress, it seems to be towards purity and saturation.

 

[DREW WILEY]

Monitoring a backlit screen is a little different game than working with film and paper dyes, though all three generally share the same 3-axis CIE color
mapping theory. Opaque pigment (including inkjet) are better served with a 4-axis model where tone based upon white is plotted apart from tint based
upon hypothetical black (there is no such thing as a pure black pigment, but this is the theory). This also helps in the printing industry where CMYK needs black to enforce deeper shades which can not do so by themselves (even though in theory this is achievable). The distinction between additive RGB in color vision and projected light from CMY process colors is known to many, so no need to comment here. Inkjet systems using many different hues of ink, obviously, and must do so due to the near impossibility of finding relatively permanent pure CMYK options that are even capable of passing through those tiny nozzles like dyes can. But there are no perfect dyes either, so all even advanced dye printing machines use a complex blend. With office printers, quality of color reproduction can be compromised. Then in tripack films and papers there are all kinds of secondary issues in the way these layers interact, and in the case of color neg film, how the masking layer works - way over my head, except to note how once again, although every single one of these systems can be quantified using standard color mapping, that is no the same thing as creating consistent color devoid of metamerism. To do that, you'd have to created a separate map for every real-world lighting scenario and not just a standard lab illuminant. This is just the tip of the iceberg. It's hard enough finding an ideal dye set even in something like dye transfer printing where you need only three dyes and have 75 years of track
record of choices. Complex masking or delicate combinations of dyes are required to balance inherent spectral flaws in any of them, not to mention
how the flaws in color film itself give a biased viewpoint to begin with.

 

[Mark Antony]

Mark, you should take a job as a dye chemist, since in one simple post you have solved problems that no dye chemist in history has been
able to achieve! But it would help to at least mention how you could also do it at the same time as making these miracles products lightfast,
environmentally and commerically viable, and a few hundred other things that Kodak, Fuji, and Agfa people had to struggle with for decades.
And if you can indeed invent a film that can even REMOTELY compare to the range and gamut of my own eyes, please please sell me a box!

Well that was part of my job, I was on the development team for Velvia, and NPS. I fail to see what in my post has been 'failed to achieve' all I am pointing out is the simple fact that the human eye is an RGB device.
Films exist that EXCEED the visible spectrum and have done so for MANY years, they are lightfast too. The products being lightfast is a different issue from their spectral properties.
Most colour films use vinyl colloids for colour (synthetic) rather than long chain analine dyes, so light fastness has got progressively better over time.
Here is a film that exceeds the RGB values the human eye can record.


So yes obviously it can be done, possibly you're confusing a wider gamut than the human eye with the human perception of perfect colour response, adaption and metermeric failure which really don't have anything to do with gamut but rather with dye purity.

 

[Bill Burk]

Mark Antony,

Surely you have some stories to tell then! Drew Wiley admitted some rascally behavior for which he is well-known. Interesting point that films can receive the color outside human sense... And I believe some dyes can show colors outside of human sense too.

But I don't think there are three dyes which form a triangle of blending which can cover the whole CIE horseshoe of human visual sense. I think that multiple detectors and dyes (more than three) could form a color model that could make a more rich image for people to see.

But I don't think there is any effort outside the choice of better dyes (or LED or Lasers) to make up the largest possible triangle of RGB.

 

[Mark Antony]

Well Bill you're welcome to your opinion, but remember there are three dyes in the cones in your eyes that form your RGB response, and the model of visual response has been mapped and understood for 100 years +.
I wonder what colour you see no film has ever recorded?
It's not that hard to make a film with 100% colour response though I'm not sure luminance will be possible, so when you talk about the horseshoe possibly that's what you mean? not the hue but rather brightness.

If you are interested spectral density standards are measured by using wave lengths from cadmium (Red) and Mercury arcs spectral densities being used to describe a dye response as functions of a wavelength over the whole visual spectrum.
In practice dyes with high Q factors are rarely used because these would result in poor equivalent neutral density when printed.
Neutrality being favoured for good skin tones over high q factor (of each dye in a record separately) and analytical colour density because Maxwell's principle tells us the sensitivity should be proportional to the colour matching function corresponding to each primary. If later colour correction (printing) is used then almost any set of colour matching functions may be used.
The normal colour matching functions are 450nm in the blue, 540 green and 610 in the red.

 

[Bill Burk]

It's not that hard to make a film with 100% colour response though I'm not sure luminance will be possible, so when you talk about the horseshoe possibly that's what you mean? not the hue but rather brightness.

The middle of the CIE horseshoe model is white, the outside curve of the shape is where the highest saturation exists. I'd defer to your experience regarding the ability of film to detect all colors, but the compromise when you choose three specific colors for the output.... can only blend to make colors with less saturation between the three colors that you picked for the job.

You can probably create a single dye that has the full saturation of any color on the horseshoe curve (that's an advantage painting has over film and printing). But as soon as you pick three (or four or more) for blending, you create bounding points that all the blends happen to fall between.

At least that's how I understand it... I can't name any specific color but something in the range of periwinkle never seems to come out right. Drives me crazy. Metamerism can mess things up too. Some of the reasons I shoot black and white. I can put all that at the back of my mind.

 

[nworth]

No, it will not produce all colors properly. For instance, it will not reproduce a true violet (say 400 nm); that is because of the way the eye works. On color photos, violet comes out blue. But it does an excellent job. Manufacturer's spend millions to make sure it does. There are many causes for the errors, such as they are. Go to the digital sites and read some good discussions about color gamut to get an idea. The eye is not a particularly accurate or sensitive color measurement tool, so many small errors are unnoticeable as long as the included range is adequate in both the recording and reproducing parts of the process. Color films have quite wide and pretty accurate color ranges, as you can see by the specifications.

 

[Nodda Duma]

.
The normal colour matching functions are 450nm in the blue, 540 green and 610 in the red.

But of an aside: These are roughly the design wavelengths I would use to design optics with when it's just a straight visible spectrum lens... The actual correction would then be about 425-650. But over the past few years I've found the color comes out better when I correct 400-700 (when I can get away with using the necessarily more expensive glass). That way the actual "good" correction ends up being 375-725. Seems to provide a better result in practice.

 

[Mustafa Umut Sarac]

But of an aside: These are roughly the design wavelengths I would use to design optics with when it's just a straight visible spectrum lens... The actual correction would then be about 425-650. But over the past few years I've found the color comes out better when I correct 400-700 (when I can get away with using the necessarily more expensive glass). That way the actual "good" correction ends up being 375-725. Seems to provide a better result in practice.

Jason,

Now I understand what glass can do or not , thank you very much indeed.

Umut

 

[alanrockwood]

The following is an important distinction to be made: detection of all colors vs. accurately reproducing all colors. To take an extreme example panchromatic film may detect all colors, but it fails to accurately reproduce all colors.

 

[DREW WILEY]

My point was, that between film and any kind of print media you might choose, there are all kinds of hues that cannot be photographically
reproduced with accuracy in the sense that the human eye can see them. Every single film ever invented, and every single color paper of
dye set, has inherent limitation and idiosyncrasies which bracket the real-world usefulness. This especially applies to certain categories of
complex neutrals.

 

[amellice]

Wikipedia says that using CMYK you won't be able to get pure red, now I don't know what it means to be pure but I'm guessing that when you measure it it would result in 0G and 0B and values only in R

When certain colors cannot be expressed within a particular color model, those colors are said to be out of gamut. For example, while pure red can be expressed in the RGB color space, it cannot be expressed in the CMYK color space; pure red is out of gamut in the CMYK color space.

 

[JoJo]

Wikipedia says that using CMYK you won't be able to get pure red, now I don't know what it means to be pure but I'm guessing that when you measure it it would result in 0G and 0B and values only in R

Pure red, green or blue from RGB cannot be shown in CMYK. The RGB colors you know from your PC or TV sreen are nearly single wavelengths.
Such colors are normally printed too dark, when printed in CMYK. Red could be better, but the big problem is green.
This is because you need 2 colors C + Y to create green. But when using 100% of C + Y, density is too high making the color dark. ICC profiles match the colors best as possible by making the colors brighter but it is just an approximation to RGB.

Here is an RGB Testpicture.
When I printed this to RA4 using my homemade laser imager, you can see that R, G an B, but specially G is too dark.
You can see this at the 2nd picture. This is a scan of the print. It has some color shift but you can see the big difference in the green parts.

So if I want to print a better reproduction I have to use an ICC profile, which exposes the picture like the 3rd one. Red, blue and specially green is much brighter to be exposed to the paper.


Joachim

 

[amellice]

Thanks for the examples, what is the laser imager?

Pure red, green or blue from RGB cannot be shown in CMYK. The RGB colors you know from your PC or TV sreen are nearly single wavelengths.
Such colors are normally printed too dark, when printed in CMYK. Red could be better, but the big problem is green.
This is because you need 2 colors C + Y to create green. But when using 100% of C + Y, density is too high making the color dark. ICC profiles match the colors best as possible by making the colors brighter but it is just an approximation to RGB.

Here is an RGB Testpicture.
When I printed this to RA4 using my homemade laser imager, you can see that R, G an B, but specially G is too dark.
You can see this at the 2nd picture. This is a scan of the print. It has some color shift but you can see the big difference in the green parts.

So if I want to print a better reproduction I have to use an ICC profile, which exposes the picture like the 3rd one. Red, blue and specially green is much brighter to be exposed to the paper.


Joachim

 

[DREW WILEY]

The devil is always in the details. For example, with typical CMY enlarger colorheads, there is always a bit of white light spillover that affects the entire emulsion, so getting "pure" primary hues is impossible, even if a "perfect" color printing paper itself existed. I personally use additive RGB enlargers, which are conspicuously better in this respect, but rather complicated to make. If you use very narrow-band filters they are going to be quite dense and therefore need a lot of light. Factor all that in and it equates to more heat; and excess heat can actually cause dichroic filters to shift their spectral transmission characteristics somewhat, not to mention be hard on enlarger maintenance. So I use "trimmer" filters which are used in pairs like a sandwich to trim off the sides of the intended spectral transmission. We could go on and on and on about why color theory, though vital to understand, never ever totally equates to real-world color-reproduction issues. There are just way too many variable, and nothing in terms of supplies and equipment is every perfect.

 

[RobC]

The OP asked about CMYK filters. I presume that is because he is intending to project onto some light sensitive material. So it follows that it is how the light sensitive material responds to the light hitting it which determines what colours will be produced. So it is entirely possible that pure RGB is possible if the material is designed to respond to CMYK in that way. But since the OP doesn't appear to know and no one here is saying they do, I think we can all assume that everyone is guessing.

 

[gzinsel]

"but we are starting to get into minutia. I really don't like going there." There. . . . I was able to quote myself. I feel vindicated.

 

[Bill Burk]

The OP asked about CMYK filters. I presume that is because he is intending to project onto some light sensitive material. So it follows that it is how the light sensitive material responds to the light hitting it which determines what colours will be produced. So it is entirely possible that pure RGB is possible if the material is designed to respond to CMYK in that way. But since the OP doesn't appear to know and no one here is saying they do, I think we can all assume that everyone is guessing.

I didn't even notice he was asking about filters.

OP is into exploring theories, I thought he meant CMYK print process and assumed Umut wanted to know if print can produce all possible colors of human perception.

I think we need to talk about each step in the process and clarify whether we are talking about input or output because we've already argued about things that we don't disagree about.

Some example steps where you might talk about the range of possible colors: illumination color, subject color, film sensitivity, film dyes, enlarger light source, print paper sensitivity, paper dyes, viewing light source.

So for example, in his last post, DREW WILEY was talking about just the enlarger light source. Which I think comes closest to answering OP's original question.

 

[Bill Burk]

"but we are starting to get into minutia. I really don't like going there." There. . . . I was able to quote myself. I feel vindicated.

In a few minutes the minutia militia may give us a second opinion

 

[RobC]

the operative word is "Filter" and thats what filters do. Any filter will only pass wavelengths that it is NOT blocking. So you should already know the answer to your own question if you have done your research and know what wavelengths each of your filters block and what wavelengths each of your dyes requires to activate/create them. Your question seems to indicate you haven't done that and its not what you've asked. Do you know what you're doing?

I am able to quote myself too.

 

[gzinsel]

As Candidate "A" . ( let me give my stump speech) "I declare we should all quote ourselves as often as possible. I can not explain in words, HOW satisfying it was yesterday, to take permission and quote myself! Although, I did have problems with it looking cool, with light grey box with the bold text of who originally quoted. But I declare as Candidate A "I will WORK hard for you! and get the box thing right"