Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: ErikKaffehr on October 08, 2010, 04:05:36 pm
-
Hi,
We had a recent discussion on this forum about using Adobe RGB vs. sRGB on monitors.
As I guess we all know there are visible colors outside Adobe RGB and a digital camera has a significantly larger gamut than Adobe RGB. Similarly, a printer can print some colors outside Adobe RGB. So weather sRGB or Adobe RGB we will not able some of the colors we actually have in images and also in prints.
For the time being I have the impression that Prophoto RGB, or some derivation thereof, is the best working color space available today. Will we see monitors some day, supporting Prophoto RGB?
Would that be possible at all? The primaries are outside the "horseshoe" of the CIE 1931 xy diagram, meaning that they would be more saturated than spectral colors! Could such a monitor be built, and would it have real benefits?
Best regards
Erik
-
Will we see monitors some day, supporting Prophoto RGB?
Not likely...bigger than Adobe RGB would be likely but not PPRGB since parts of the gamut are outside of human vision.
-
"Outside"? Infra-red, Ultra-Violet?
-
"Outside"? Infra-red, Ultra-Violet?
Outside as in ProPhoto RGB's chromaticities for RGB fall outside of CIE 1931 color which represents human vision...it has nothing to do with infrared or ultraviolet which are wavelengths.
-
Will we see monitors some day, supporting Prophoto RGB?
Only when the evolution of our species allows us to see outside the current visible electromagnetic spectrum. By then, we will not need displays, we will be closer the Arthur C. Clarks super embryos <g>
-
It might need the same trick done for printing : once you can't saturate more your 3 primaries (how's about a CIE-RGB monitor? :P ), you still can extend the gamut by adding other primaries like orange or green?
-
It might need the same trick done for printing : once you can't saturate more your 3 primaries (how's about a CIE-RGB monitor? :P ), you still can extend the gamut by adding other primaries like orange or green?
It would no longer be an RGB working space (color space).
-
Hi
multi primary inaging has been about for a while but not mainstream as you will no doubt gather fro this link.
http://www.isl.titech.ac.jp/~guchi/NV/NV-IntroE.html (http://www.isl.titech.ac.jp/~guchi/NV/NV-IntroE.html)
But who knows?
The nearest you can get to "real life" imaging is the Lippmann Method
http://nobelprize.org/nobel_prizes/physics/articles/biedermann/ (http://nobelprize.org/nobel_prizes/physics/articles/biedermann/)
Very interesting stuff!
-
It would no longer be an RGB working space (color space).
I would rather have thought that the monitor space couldn't be matrix based, just as the printer spaces?
-
Is there a physical limitation? The fact that we cannot see those colors doesn't necessarily mean they cannot be achieved.
-
Hi,
The "horse shoe" in the diagram represents spectral colors. So each dot on the curve would represent a pure monochromatic spectrum color. I don't think that a physical implementation of any primary outside the horseshoe diagram would be possible. But I'm no color scientist, I might be wrong.
Best regards
Erik
Is there a physical limitation? The fact that we cannot see those colors doesn't necessarily mean they cannot be achieved.
-
Is there a physical limitation? The fact that we cannot see those colors doesn't necessarily mean they cannot be achieved.
If you can’t see them, they are not a color:
Color, is a perceptual property. So if you can't see it it's not a
color. Color is not a particular wavelength of light. It is a
cognitive perception that is the end result of the excitation of
photoreceptors followed by retinal processing and ending in the
visual cortex. We define colors based on perceptual experiments.
A coordinate in a "colorspace" outside the spectrum locus is not a
color. We often refer to these as "imaginary colors" but this is by
and large also erroneous (you can't map an imaginary color from one
colorspace to another as the math (and experimental data) for each
colorspace breaks down outside the spectrum locus.
-
So if you can't see it it's not a
color
Thanks for the responses, but I have another doubt. Suppose a full spectrum but very low intensity light, below the minimum level to excite the cones, so only Rods can see it. We will perceive a monochromatic image.
Based on the previous statement we could conclude (erroneously) that there is no color, but if you take a photo it will show the colors.
Another thought, and forgive me if this is nonsense, I'm just curious: Suppose a color blind person, we could think as him having a smaller gamut vision than a normal vision person. Would this color blind person perceive a color inside his color gamut if it was produced by primaries outside of this gamut (inside the visible spectrum for a normal vision person)?
-
Hi,
We had a recent discussion on this forum about using Adobe RGB vs. sRGB on monitors.
As I guess we all know there are visible colors outside Adobe RGB and a digital camera has a significantly larger gamut than Adobe RGB. Similarly, a printer can print some colors outside Adobe RGB. So weather sRGB or Adobe RGB we will not able some of the colors we actually have in images and also in prints.
For the time being I have the impression that Prophoto RGB, or some derivation thereof, is the best working color space available today. Will we see monitors some day, supporting Prophoto RGB?
Would that be possible at all? The primaries are outside the "horseshoe" of the CIE 1931 xy diagram, meaning that they would be more saturated than spectral colors! Could such a monitor be built, and would it have real benefits?
Best regards
Erik
many already show many more colors than adobergb covers, none are close to ProphotoRGB, not even our vision actually
-
Is there a reason that a large-gamut monitor has to adhere to some sRGB/aRGB/... standard only? Does it not make more sense that these specialized monitors come with a precise color profile that lets the color management software take advantage of whatever it offers?
-h
-
It would no longer be an RGB color space.
Regards,
Ali.
-
It would no longer be an RGB color space.
Regards,
Ali.
Does it matter what it is as long as it does what it is supposed to do well?
-h
-
If it has three primaries R,G,B and they are inside the "horseshoe" CIE xy diagram, it will indeed be an RGB Color space, regardless of being different than sRGB or AdobeRGB
-
Is there a reason that a large-gamut monitor has to adhere to some sRGB/aRGB/... standard only? Does it not make more sense that these specialized monitors come with a precise color profile that lets the color management software take advantage of whatever it offers?
-h
There is no reason. And they don't.
Most have a gamut that does not match AdobeRGB, it is usually smaller in intense yellow-greens and far larger in certain other areas.
This is why it makes no sense to use them in ADobeRGB mode when using color-managed software.
And they are certainly still RGB colorspaces, they still use R and G and B primaries.
-
No, since two of the primaries (green and blue) are not physically realizable.
-
No, since two of the primaries (green and blue) are not physically realizable.
Eric,
You make an interesting point. Ideally, we would like a monitor to reproduce all visible colors: i.e. the CIE XYZ color space. That also might be impossible, since in their color matching experiments, the investigators had to resort to negative colors. As a compromise, one could shoot to reproduce real world surface colors. These are non-emissive (reflected) colors that are encountered in nature. For an example see Page 10 of Gernot Hofmann's PDF (http://en.wikipedia.org/wiki/CIE_1931_color_space). One attempt to construct such a space is BetaRGB by Bruce Lindbloom. From his CIE plot of this space it looks like the primaries are just within the CIE horseshoe [Red = (0.6888, 0.3112), Green = (0.1986, 0.7551), Blue = (0.1265, 0.0352)]. Would this be realizable on an RGB monitor? Does adding a yellow as in Sharp's new TVs do anything for gamut?
The BetaRGB is larger than Adobe RGB and smaller than ProPhotoRGB, but it doesn't seem to have caught on. ProPhotoRGB works fine if one uses 16 bits and does not create unrealizable colors by overzealous editing. What do you think?
Regards,
Bill
-
Bill,
R.W.G. Hunt has a good explanation of the gamut limitations of trichromatic reproduction in The Reproduction of Colour. Reproducing the full gamut of human vision is theoretically impossible with three primaries because of the fact that the sensitivity curves of the different cones overlaps making it impossible to stimulate them independently.
2.5 UNWANTED STIMULATIONS
It is thus clear that the inability of any beams of red, green, and blue light to stimulate the retinal cones separately introduces a basic complication into the whole of trichromatic colour reproduction. If the ρ and β curves did not overlap in the blue-green part of the spectrum, then green light could be found that stimulated the γ-cones on their own; but, since the ρ and β curves do overlap appreciably, the γ-cones cannot be stimulated on their own. For colour vision, this overlapping provides the basis for good detection of changes in hue throughout the spectrum. But, for colour reproduction, it means that simple trichromatic methods cannot achieve correct colour reproduction of all colours.
…Thus the overlapping of the cone sensitivity curves, resulting as it does in the inability to stimulate each type of cone separately, is the reason why correct colour reproduction by simple trichromatic means is impossible to achieve for all colours.
-
Bill,
R.W.G. Hunt has a good explanation of the gamut limitations of trichromatic reproduction in The Reproduction of Colour. Reproducing the full gamut of human vision is theoretically impossible with three primaries because of the fact that the sensitivity curves of the different cones overlaps making it impossible to stimulate them independently.
Mark,
That is interesting. What additional color(s) would be needed?
Regards,
Bill
-
For a more geometric/visual way of thinking about the problem, see here:
http://en.wikipedia.org/wiki/CIE_1931_color_space
Observe the horseshoe shape (spectral locus) showing the range of colors. The salient bullet point is the one to the left of the diagrams (quoted verbatim):
It can be seen that, given three real sources, these sources cannot cover the gamut of human vision. Geometrically stated, there are no three points within the gamut that form a triangle that includes the entire gamut; or more simply, the gamut of human vision is not a triangle.
-
For a more geometric/visual way of thinking about the problem, see here:
http://en.wikipedia.org/wiki/CIE_1931_color_space
Observe the horseshoe shape (spectral locus) showing the range of colors. The salient bullet point is the one to the left of the diagrams (quoted verbatim):
To cover a curve (horseshoe) using vectors within the shape one would need an infinite amount?
-h
-
The amount is not the problem.
The problem is the shape. Additive primaries define a triangle in the space shown in the diagram. There is no way to place 3 primaries within the horseshoe that define a triangle that contains the entire horseshoe.
As a rough analogy, try drawing a circle on a piece of paper. Then pick any 3 dots within the circle. Those 3 dots define a triangle. No matter which 3 dots you pick, there's no way that triangle will contain the entire circle. The only way to make a triangle that contains the entire circle is to put those dots outside the circle.
-
The amount is not the problem.
The problem is the shape. Additive primaries define a triangle in the space shown in the diagram. There is no way to place 3 primaries within the horseshoe that define a triangle that contains the entire horseshoe.
As a rough analogy, try drawing a circle on a piece of paper. Then pick any 3 dots within the circle. Those 3 dots define a triangle. No matter which 3 dots you pick, there's no way that triangle will contain the entire circle. The only way to make a triangle that contains the entire circle is to put those dots outside the circle.
My point is that a square is going to fit better within a circle than a triangle. A penta.... is going to fit even better. As the number of corners approach infinity, you basically have a circle.
If the magic of perception and colors allows for primaries that is outside the horseshoe, even better, problem solved?
-h
-
If the magic of perception and colors allows for primaries that is outside the horseshoe, even better, problem solved?
As I said in Post #4, that issue was solved in 1968/2001ish. But only for Dave Bowman <g>
-
Yes, it is possible to construct a ProPhotoRGB Monitor that takes in 16-bit ProPhotoRGB Primaries and generates most of the equivalent color on the display. It requires 5 primaries in the display. Keep in mind that any color on the CIE 1931 diagram inside the convex hull of your primaries can be produced. Since two of the ProPhotoRGB primaries have CIE X,Y,Z coordinates outside the CIE gamut, the additional primaries would best be placed around the points where the ProPhotoRGB triangle exits the CIE diagram.
Here's how I would do it: Start with a LCD that has a very high refresh rate (240Hz) and an RGB backlight (not a white LED backlight). Replace each of the backlight's RGB LED triplets with a bank of 5 LEDs with wavelengths around 700nm, 550nm, 530nm, 485nm, and 465nm. We will likely need to increase the number of LEDs to get enough brightness. While LEDs come in a huge range of wavelengths, there are still some gaps in coverage and the above wavelengths might have to be tweaked.
Next you might need to alter the color filters in the LCD Panel a little. If you call the filter colors Fr, Fg, Fb, the Fr filter needs to pass the 550nm and 530nm LED light (and reject the others), and the Fb needs to pass the 465nm and 485nm light (and obviously block the primaries on the other side of the CIE diagram). We now time multiplex between two sets of primaries. In frame 1, the LCD display turns on the 700, 550 and 485nm primaries and in frame 2 we send across the information for the 700 (redundant, but not necessary), 530nm, and 465nm. Inside the monitor is a frame buffer to store the ProPhotoRGB image from the computer (because we're going to use it twice), and a gate array or high speed processor that converts the 16-bit ProPhotoRGB primaries to the appropriate 5-color image. For a 240Hz monitor you would get all the colors every 120/sec, so you shouldn't see any flicker. This would work best for static images. Time multiplexed color creates ugly artifacts (IMHO) for moving images.
This approach would have the least change to existing panel design and preserve the existing resolution.
-
Here's how I would do it: Start with a....
Are you planning to do this?
-
Bryan asks if I'm planning to do this.
Naw - even a highly leveraged project like this would take 3 or 4 engineers to prototype. Maybe one of the high-end monitor manufacturers will do something like this some day. One thing I've noticed in all LCD panels I've ever looked at is a small variation in color and brightness based on the angle you are are viewing. This is only a minor annoyance in an sRGB workflow, but would be more important in a ProPhotoRGB workflow/monitor, where the discriminating user is depending on being able to see subtle color differences and have those differences be consistent across the screen.
The ultimate way to do a wide-gamut display is an OLED display, with 6 primaries and wide viewing angle. For the 6th primary, I'd consider a white pixel. Here's why: When you synthesize a color in the middle of the CIE diagram with a green primary and two nearly-monochromatic primaries out near the lower corners of the CIE diagram, it is possible to sense chromatic aberration in the corners if you wear glasses. I noticed this effect on a LaCie wide gamut monitor. Here's the experiment: fill the screen with white text on a black background. Sit reasonably close to the monitor (typical 18" viewing distance) and look at the text in the center of the screen. Without moving your head, sense the text in the corners of the screen, and you can see the split of the red and blue channels. I wear low-dispersion glasses, but easily saw a 1 to 1.5 pixel split between the red and blue signals making up the white characters on the LaCie monitor in the corner. If I turn my head and look directly at the character the effect immediately goes away.
Lastly, if I was doing this for real (e.g. if I worked at Samsung, LG, LaCie, Sony, etc), I probably would just go ahead and spread the primaries out to completely cover CIEXYZ and use CIELAB directly as the signal rather than ProPhotoRGB. The CMM has to be aware of of the CIEXYZ and CIELAB color spaces anyway since these are the Profile Connection Space for converting one gamut to another, so if you have a monitor that for all practical purposes covers the entire CIEXYZ space, why not just send the signals that way rather than constraining them to strange triangle shapes. In the past, the real-time conversion would have been too compute intensive for the 200+Mpixel rates needed, but modern electronics could now feasibly put this in the graphics card or the monitor. We may still need a little faster CPUs before Adobe uses CIELAB or better yet a hyperspectral internal working space :-).
Lastly, we have to remember that real-world surfaces rarely have reflectances out near the edges of the CIE diagram, and today's displays, printers, cameras, and photo software are already capable of very pleasing reproductions without ultra-wide gamut capability, so the market may be limited. But oooooooh, would I love a display that could reproduce the Neon Light Tunnel at Chicago's O'Hare International Airport (between Terminals B and C)!!, or the structural (diffractive) color of a peacock tail, or Scarlet Tanager. Gabriel Lippmann would have been proud :-).
-
Since two of the ProPhotoRGB primaries have CIE X,Y,Z coordinates outside the CIE gamut, the additional primaries would best be placed around the points where the ProPhotoRGB triangle exits the CIE diagram.
Since that actual primaries falls outside the gamut of human vision, what good would it do us? And if you move them into the coordinates, how can we call it ProPhoto RGB?
-
Hi,
We had a recent discussion on this forum about using Adobe RGB vs. sRGB on monitors.
As I guess we all know there are visible colors outside Adobe RGB and a digital camera has a significantly larger gamut than Adobe RGB. Similarly, a printer can print some colors outside Adobe RGB. So weather sRGB or Adobe RGB we will not able some of the colors we actually have in images and also in prints.
For the time being I have the impression that Prophoto RGB, or some derivation thereof, is the best working color space available today. Will we see monitors some day, supporting Prophoto RGB?
Would that be possible at all? The primaries are outside the "horseshoe" of the CIE 1931 xy diagram, meaning that they would be more saturated than spectral colors! Could such a monitor be built, and would it have real benefits?
Best regards
Erik
Well many wide gamut monitors already show a decent amount of colors beyond AdobeRGB (although almost all also fall a little short near the extreme yellow greens) even if well short of Prophoto. Most of my photos fit into AdobeRGB but I do have plenty that look different on my monitor if I convert them down from ProphotoRGB to AdobeRGB, almost all of my deep purple flower photos are one such case.
I really don't know enough about materials, filters, backlight light sources and their properties to say what could reasonable be accomplished. I'm sure they could make them with much larger gamuts right now but maybe they would wear out too quickly or cost too much to put into commercial sets for now.
-
Is there a reason that a large-gamut monitor has to adhere to some sRGB/aRGB/... standard only? Does it not make more sense that these specialized monitors come with a precise color profile that lets the color management software take advantage of whatever it offers?
-h
that's exactly what does happen
hardly any standard gamut monitor is really sRGB and perhaps none of the wide gamuts are AdobeRGB. Most both extends past and fall short of the standard ideal gamuts that they are closest too. You use a probe and measure them and then the color aware software knows how to handle the native gamut of the monitor.
-
No, since two of the primaries (green and blue) are not physically realizable.
well you could make what would make them so in a sense they are physically realizable but since human eye wouldn't response to them properly it wouldn't do any good and it wouldn't work out so yeah I guess it would not be possible to ever make a prophotorgb monitor, with a very complex array of primaries you might be able to make one that covers all the visible parts of it though which is all that matters
-
We can call it ProPhotoRGB because the monitor could reproduce every color within the ProPhoto RGB color space that represents a color that the human visual system can perceive. It would be useful because the monitor could display many colors that humans can see and print that are not displayable by monitors that have only 3 primaries. In some ways it is loosely analogous to printers which use more than 3 inks to achieve a wider gamut.
I carefully chose my words in the first sentence - "color that the human visiual system can perceive". The explation quoted from Hunt earlier in this thread explains it well, but let me try and rephrase and see if that helps. The eye can "see" or "detect" light in the electromagnetic spectrum between about 400nm and 700nm. The huge amount of information in this spectrum gets compressed down to only 3 numbers through our S (blue) cones, M (green) cones, and L (red) cones. To reproduce the color sensation of any arbitrary collection of light from the 400-700nm spectrum, we don't need to reproduce that whole spectrum (like in Lippmann Photography), but we do need to stimulate the SML cones the same way as the original light. If there existed 3 unique colors of light that could independantly stimulate the S,M, and L cones, we'd be there - these three lights (primaries) could be combined in different amounts to reproduce any color we can perceive. However, since the spectral responses of the S,M and L cones overlap, it is physically impossible to have a light (or a primary color in a display) that individually stimulates the S,M,L cones. Since we can't independantly stimulate the SML cones, we have to come up with some affordable number of lights (primaries), less than infinity, that can produce the same SML response as real world spectrals produce. If we pick 3 primaries that are close to the peak response of the S,M, and L cones, we can cover a good chunk of what we can perceive. That is the basis of NTSC, sRGB, AdobeRGB, and many other color spaces. But we can't cover everything we can perceive, as explained earlier. As we add primaries, we can cover more and more of the possible sensations, or perceivable colors, because we have more control over the response we want from the SML cones.
Here is the concept that takes some time to absorb: even though we can't *reproduce* any arbitrary color using only 3 real-world primary lights, we can still *describe* any perceivable color with only 3 numbers. This makes sense since we only have 3 types of cones. Describing all perceivable colors with only 3 numbers is what CIEXYZ does with the X,Y,Z axes. Two of the ProPhotoRGB primaries are imaginary colors - they do not represent colors we can perceive, but these numbers still serve a purpose - they allow us to describe more colors than color spaces based on perceivable primaries.
And this is why we can call it a ProPhoto RGB monitor.
-
Since that actual primaries falls outside the gamut of human vision, what good would it do us? And if you move them into the coordinates, how can we call it ProPhoto RGB?
I think what acgoris is suggesting is a device that reproduces the visible subset of ProPhotoRGB by using five primaries as in the attached image.
I'm not sure where current tech is on affordably producing pure colors, but it seems you would need to create bright spectral (or near spectral) colors for each of the primaries for this to work.
-
Nice diagram Mark!
LEDs can be pretty close to monochromatic. Attached is a picture of a bunch of LEDs on a chart showing their locations on the CIE diagram. This is from a NIST site.
One other thing to note is that the large missing green space isn't as big as it looks. The CIE XYZ space has many nice properties in describing color, but it is not perceptually uniform - the same spacing in the upper left is not as easy to tell apart as lower in the chart.
-Andy
-
I think what acgoris is suggesting is a device that reproduces the visible subset of ProPhotoRGB by using five primaries as in the attached image.
Ah, OK but that then would not be ProPhoto RGB. Call it acgRGB if you want. Primaries that fall outside human vision are not colors. ProPhoto RGB defines such primaries. As Eric wrote, due to the shape, it has to. That doesn’t mean any real world device should or can mimic this theoretical color space.
-
This is just semantics.
The monitor could do two things:
(1) It would accept pixels in a ProPhoto RGB format.
(2) It would produce colors on the display for all regions of the ProPhoto RGB space which have meaning.
It is irrelevant how it performs these feats.
What more could anyone want from a monitor that claims to support ProPhoto RGB?
I agree that primaries that fall outside human vision (i.e., the CIE diagram) are not colors. They have no physical or perceptual meaning. I agree that ProPhoto RGB defines such primaries.
Claiming a monitor supports ProPhoto RGB is not claiming it can produce something meaningful for all the mathematical combinations of numbers within the ProPhotoRGB space, because not all combinations have meaning.
Nor is it claiming that it has *physical ProPhoto RGB primaries (which is impossible)* any more than AdobeRGB monitors contain actual, physical AdobeRGB primaries - they don't. But they use the primaries they do have to cover the majority of the Adobe RGB space.
-
this topic is quite interesting :) :)
-
Another interesting approach is a CMYK monitor like possible with electrowetting displays. Liquavista>Samsung is developing that technology but it has been quiet there the last year and a CMYK monitor must have the lowest priority of all. I do not know what a transmissive CMYK display could achieve in gamut with its subtractive color mixing but I suppose a 5 to 6 primaries OLED would make a wider gamut possible.
One thing will not change: the trend in gamut size increases and related bit depths applies to all components; camera sensors, image editors, printer output and monitors. Soft proofing will have a place as long as there are differences in size and shape of the gamuts used in the workflow. Given progress and changes in technology and the budgets to purchase equipment there probably will never be stable situations on amateur desktops and in pro studios over long periods so soft proof features should be improved first.
met vriendelijke groeten, Ernst
330+ paper white spectral plots:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
No, since two of the primaries (green and blue) are not physically realizable.
(http://upload.wikimedia.org/wikipedia/commons/thumb/8/8f/CIE_1931_XYZ_Color_Matching_Functions.svg/446px-CIE_1931_XYZ_Color_Matching_Functions.svg.png)
Just so that I am getting things here:
Is the fundamental problem that no three spectra can be designed that can excite the L, M and S cones independently? Even monochromatic ones that are placed at the wavelength extremes, and at zero-crossings?
What are the arguments for choosing broad, smooth PSD in display devices? To maximize energy efficiency? To minimize individual deviations from the CIE models?
If one could make a device that could generate any spectra with very narrow wavelength sampling, then any physical stimuli pertaining to the perceptual correlate "color" could be generated, right? If it cannot be generated physically (in nature or in the lab), then it probably is not a "color" we need to worry about?
-h
-
Q: Is the fundamental problem that no three spectra can be designed that can excite the L, M and S cones independently?
A: Exactly
Q: Even monochromatic ones that are placed at the wavelength extremes, and at zero-crossings?
A: Correct
Q: What are the arguments for choosing broad, smooth PSD in display devices?
A: They don't. They usually try to make them as narrow as possible, which moves the primaries to the edges of the CIE diagram. When you plot the CIE coordinates of the 3 primaries, you can reproduce any color inside that triangle.
Q: To maximize energy efficiency?
A: This is definitely a consideration, especially when using a white backlight. Any light that doesn't make it through a color filter is wasted energy.
Q: To minimize individual deviations from the CIE models?
A: Variation from display to display, or within a display over time is an issue, and why we calibrate displays. Display designers have to tradeoff several things - cost, brightness, size of gamut, reliability, consistency, etc. Phosphors (used in CRT's and two common backlights for LCD panels - white LEDs and CCF tubes) shift color as they warm up, and as they age. Some high-end wide-gamut LCD panels use RGB LEDs as their backlight. These also shift color with temperature and age. They also shift color with the amount of current flowing through them. All of these shifts are small, but enough that professionals in the photography business care.
Q: If one could make a device that could generate any spectra with very narrow wavelength sampling, then any physical stimuli pertaining to the perceptual correlate "color" could be generated, right?
A: Yes! You just described the Holy Grail. This has been done for film photography using an interference technique. A extremely fine-grain film is required. The pioneer of this was Gabriel Lippmann, who won the Nobel prize in 1908 for his work. Wikipedia has a good description of how this works if you look up Gabriel Lippmann.
Q: If it cannot be generated physically (in nature or in the lab), then it probably is not a "color" we need to worry about?
A: Correct! You could even change the word "probably" to "definitely" :-).
-
Q: To maximize energy efficiency?
A: This is definitely a consideration, especially when using a white backlight. Any light that doesn't make it through a color filter is wasted energy.
I have seen suggestions that "wide gamut" displays actually tend to have regular spectral selectivity in the lcd panel, while the "white" backlight is more like 3 narrow peaks centered at "L", "M" and "S" in such a ratio that it is perceived as white.
Q: To minimize individual deviations from the CIE models?
A: Variation from display to display, or within a display over time is an issue, and why we calibrate displays. Display designers have to tradeoff several things - cost, brightness, size of gamut, reliability, consistency, etc. Phosphors (used in CRT's and two common backlights for LCD panels - white LEDs and CCF tubes) shift color as they warm up, and as they age. Some high-end wide-gamut LCD panels use RGB LEDs as their backlight. These also shift color with temperature and age. They also shift color with the amount of current flowing through them. All of these shifts are small, but enough that professionals in the photography business care.
Actually, my thought was that the actual perception of color for an individual could deviate somewhat from the standardised CIE response. If this is the case, it would make sense to use display PDFs that minimized person-to-person variability. Without proof, it seems sensible to me that smooth, regular PSDs would tend to be less sensitive to variation than monochromatic light at the wavelength extremes.
Q: If one could make a device that could generate any spectra with very narrow wavelength sampling, then any physical stimuli pertaining to the perceptual correlate "color" could be generated, right?
A: Yes! You just described the Holy Grail. This has been done for film photography using an interference technique. A extremely fine-grain film is required. The pioneer of this was Gabriel Lippmann, who won the Nobel prize in 1908 for his work. Wikipedia has a good description of how this works if you look up Gabriel Lippmann.
A DLP-type projector with e.g. 256 narrow-band filters in its color wheel operating at 256x the desired framerate/rainbow-flicker threshold at 256x the normal brightness fed a HDMI input at 256/3 times the regular bandwidth should work. I have no idea how hard it is to manufacture such filters.
.h
-
As far as filters for your hyperspectral DLP projector, yes they are readily available. Image Engineering uses filters like this in their camSPECS camera calibrators.
256 bands might be overkill, but I would still like to view an implementation of this concept. If you're going this far, rather than have a bunch of filters, I'd just build a super-bright super-fast sweeping monochrometer (uses a diffraction grating and white light to make any wavelength you want). As far as human perception, there will be diminishing returns as one goes from 3 primaries to 4 primaries (like Sharp's Quattron display with RGBY) and beyond. If the goal is a tool for research into human vision and rendering, then the more the better. I would think you'd be doing pretty good with 30 bands (10-12nm spacing), but you'd need more if you wanted to truly simulate fluorescent lights. It would be nice to have pixels in the near IR and near UV as well. Young children can see much further into the UV than adults. People who've had cataract surgery and opted for non-UV blocking lenses (a friend of mine recently did this) can see to the mid-300nm's.
Here's my gut feel for the number of bands that would be be good for different applications (I'm interested in other folks opinions on this) There is diminishing returns in all of these, and keep in mind that for most applications, existing wide gamut 3-primary displays work really well, if calibrated correctly and used correctly in a color-managed workflow.
More accuracy in soft-proofing prints: 4-6 primaries
Covering PhotoRGB: 5 primaries
Reproducing highly-vivid objects, like neon lights (Neon Tunnel at O'Hare Airport), bird feathers (Peacock, Scarlet Tanager, Scarlet Macaw, Kingfisher, Painted Bunting), butterflies (Emerald Swallowtail, Blue Morpho), certain specialty paints (including fluorescent paints): 5-7 primaries
Color Vision Research: 50-200 primaries (or 'bands') that extend from 350nm to 750nm
If you divvied the primaries up across multiple DLPs and had them all point at the same screen, you could build a hyperspectral display without the temporal artifacts in time-multiplexed color displays. The latter happens as your eye moves across the screen as it refreshes each color. If you have a white line on a black background, you'll see a nice rainbow. It is fatiquing, even in 180Hz displays (60Hz/color in a 3-color system) showing a friendly (non-pathological case) picture.
-
For the 6th primary, I'd consider a white pixel. Here's why: When you synthesize a color in the middle of the CIE diagram with a green primary and two nearly-monochromatic primaries out near the lower corners of the CIE diagram, it is possible to sense chromatic aberration in the corners if you wear glasses. I noticed this effect on a LaCie wide gamut monitor. Here's the experiment: fill the screen with white text on a black background. Sit reasonably close to the monitor (typical 18" viewing distance) and look at the text in the center of the screen. Without moving your head, sense the text in the corners of the screen, and you can see the split of the red and blue channels. I wear low-dispersion glasses, but easily saw a 1 to 1.5 pixel split between the red and blue signals making up the white characters on the LaCie monitor in the corner. If I turn my head and look directly at the character the effect immediately goes away.
Why would a white primary help this any? It would be spitting out a mix of photons at different frequencies and why would they suddenly all go through glasses without any dispersion?
-
Q: Why would a white primary help this any? It would be spitting out a mix of photons at different frequencies and why would they suddenly all go through glasses without any dispersion?
A: Good question. Here was my thinking: First, I agree the components of white light would disperse just like monochromatic primaries do. However, when you have a wide mix of wavelengths (assuming the white has a broad, somewhat smooth spectrum...I should have been specific), a fewer percentage of it's photons are out at the wavelength extremes, and so the dispersion of these outer wavelengths might not be as noticeable. If the white contains some red at 700nm, that red component will disperse the same as a monochromatic red primary at 700nm, which I think is your point. However, the number of photons at 700nm emitted by a broad-spectrum white LED would be less than the number of photons emitted at 700nm by a monochromatic 700nm LED used as part of an RGB synthesis of white. Similar situation for blue.
Any benefit of a white LED would have to be tested against the benefit of another monochromatic primary, and the benefits of another monochromatic primary would probably win. The general question could be posed as this: In a display with more than 3 primaries, but less than many (8? 10?), is there any benefit to having one or more of those primaries being broad-spectrumed? With 3 or 4 primaries, obviously not. With many primaries, obviously not.
Thoughts?
To address another persons question about human-to-human variability....if you push the bluest primary very far towards 400nm, it may effect older viewers experiencing age-related change in sensitivity to shorter wavelengths due to yellowing of the lens. I recently did some testing of this cutoff using a monochrometer with some colleagues and myself, and alas my 55-year old eyes don't detect 400nm, although my resolution is tac-sharp. So I'm yellow but not fuzzy.
-
Q: Why would a white primary help this any? It would be spitting out a mix of photons at different frequencies and why would they suddenly all go through glasses without any dispersion?
A: Good question. Here was my thinking: First, I agree the components of white light would disperse just like monochromatic primaries do. However, when you have a wide mix of wavelengths (assuming the white has a broad, somewhat smooth spectrum...I should have been specific), a fewer percentage of it's photons are out at the wavelength extremes, and so the dispersion of these outer wavelengths might not be as noticeable. If the white contains some red at 700nm, that red component will disperse the same as a monochromatic red primary at 700nm, which I think is your point. However, the number of photons at 700nm emitted by a broad-spectrum white LED would be less than the number of photons emitted at 700nm by a monochromatic 700nm LED used as part of an RGB synthesis of white. Similar situation for blue.
I wonder though,if it didn't have enough energy near the ends as the middle would it not appear tinted?
OTOH with the complex response of the cones maybe a white could be mixed from energies not as far apart as the traditional red and blue primaries have most of their high energy spectral spiked at. Not sure what white LEDS output and how readily you can tune ones that work well. Maybe assumign a suitable LED can be made you could at least reduce it slightly.
To address another persons question about human-to-human variability....if you push the bluest primary very far towards 400nm, it may effect older viewers experiencing age-related change in sensitivity to shorter wavelengths due to yellowing of the lens. I recently did some testing of this cutoff using a monochrometer with some colleagues and myself, and alas my 55-year old eyes don't detect 400nm, although my resolution is tac-sharp. So I'm yellow but not fuzzy.
I just read that while the average person has a 2:1 ratio of M to L cones that they recentlyish discovered tremendous variation from person to person with some having a 1:1 ratio and others having as high as a 17:1 ratio.
-
Only when the evolution of our species allows us to see outside the current visible electromagnetic spectrum. By then, we will not need displays, we will be closer the Arthur C. Clarks super embryos <g>
http://en.wikipedia.org/wiki/Talk%3ATetrachromacy ;D
-
To address another persons question about human-to-human variability....if you push the bluest primary very far towards 400nm, it may effect older viewers experiencing age-related change in sensitivity to shorter wavelengths due to yellowing of the lens. I recently did some testing of this cutoff using a monochrometer with some colleagues and myself, and alas my 55-year old eyes don't detect 400nm, although my resolution is tac-sharp. So I'm yellow but not fuzzy.
Is the PSD (power spectral sensitivity?) of my "L", "M" and "S" cones smooth or irregular? Is it consistent from one part of my eye to the other? Is it consistent from me to the next guy?
If they are very smooth, one could essentially excite them using any narrow/irregular PSD that integrate to the desired power, or one at the wavelength extremes. One drawback of going for nearly invisible wavelength is that you might have to increase the total transmitted energy as the perceived sensitivity is so low.
If the perception spectral response and the transmission of some display are both highly irregular, even minor wavelength deviations could cause large changes in perceived "color". Then it would make a lot of sense to strive for smooth excitation PSDs?
-h
-
http://en.wikipedia.org/wiki/Talk%3ATetrachromacy ;D
wiki is down today so I can't read that but I did read elsewhere yesterday how on a HDR displays if they tune down the white point they can get people to register color more intense than the white point was predicted to allow, or something like that, I think they were implying it could make people see colors beyond what was thought possible (under very special circumstances only though) although they may merely have meant colors not expected to be able to be shown given certain considerations which wouldn't be the same thing at all.
-
Is the PSD (power spectral sensitivity?) of my "L", "M" and "S" cones smooth or irregular? Is it consistent from one part of my eye to the other? Is it consistent from me to the next guy?
If they are very smooth, one could essentially excite them using any narrow/irregular PSD that integrate to the desired power, or one at the wavelength extremes. One drawback of going for nearly invisible wavelength is that you might have to increase the total transmitted energy as the perceived sensitivity is so low.
If the perception spectral response and the transmission of some display are both highly irregular, even minor wavelength deviations could cause large changes in perceived "color". Then it would make a lot of sense to strive for smooth excitation PSDs?
-h
I believe it's been found that for the average person there is nothing that can excite just each one alone no matter what you try to do and that the best case scenario when trying to excite L, M,S as much alone as possible actually does not use light extremes out toward the edges at all, I forget the numbers used but two of them were quite close I think and even the third not way out there.
-
The human cone responses overlap so you can't stimulate one without stimulating another. (Similar to cameras.)
-
Q: Why would a white primary help this any? It would be spitting out a mix of photons at different frequencies and why would they suddenly all go through glasses without any dispersion?
A: Good question. Here was my thinking: First, I agree the components of white light would disperse just like monochromatic primaries do. However, when you have a wide mix of wavelengths (assuming the white has a broad, somewhat smooth spectrum...I should have been specific), a fewer percentage of it's photons are out at the wavelength extremes, and so the dispersion of these outer wavelengths might not be as noticeable. If the white contains some red at 700nm, that red component will disperse the same as a monochromatic red primary at 700nm, which I think is your point. However, the number of photons at 700nm emitted by a broad-spectrum white LED would be less than the number of photons emitted at 700nm by a monochromatic 700nm LED used as part of an RGB synthesis of white. Similar situation for blue.
Any benefit of a white LED would have to be tested against the benefit of another monochromatic primary, and the benefits of another monochromatic primary would probably win. The general question could be posed as this: In a display with more than 3 primaries, but less than many (8? 10?), is there any benefit to having one or more of those primaries being broad-spectrumed? With 3 or 4 primaries, obviously not. With many primaries, obviously not.
Thoughts?
yeah assuming you can make such a white LED you could surely have it excite the cones in a way to produce white using frequencies much more bunched together than the three primaries of most displays, especially compared to wider gamut ones. Of course there would still be some degree of dispersion and you'd still have some mix of the 3 primaries with full dispersion so it wouldn't come close to reducing it to zero by any means but to some extent I'm pretty sure it could help. But who knows if they can easily make such a white LED (I suppose they could break it into three monochrome subpixels though so hmm I guess it certianly should be doable). I'm not sure they care enough about the glasses wearers though to bother and how much the improvement would be and would it be worth it I don't know.