Yes, because generally we're interested about the light output of the whole system and not technical tidbits. The underlying LED itself however is still close to spectral blue (or UV or..).
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
About the color purple…
- Thread starter Ellis Vener
- Start date
LastLapPass
Forum Enthusiast
LOL, Purple Haze was high potency cannabis strain...
JJ
--
Just a Dad documenting life...
Last edited:
Jack Calypso
Senior Member
- Messages
- 1,341
- Solutions
- 4
- Reaction score
- 831
That's quite a mouthful!
No more so than complementary metal oxide semiconductor field effect transistor.
As any other color.
Indeed. I consider the linked to article worse than useless.
OpticsEngineer
Veteran Member
I have the sense that the author probably wrote a decent article but her editors considered it too difficult for the audience to understand and they turned into something disappointing. It happens all too often. People generally consider Scientific American to be decently written, but specific wavelength numbers almost never appear in it. It makes the articles on optics much less useful than they could be. I think the author ran into the same editorial barrier.
Last edited:
Dave_J_E
Forum Enthusiast
Quote “To compensate, the brain bends the spectrum into a circle, making the two extremes meet at purple”
……Classic
ProfHankD
Veteran Member
Incidentally, red + blue is usually called magenta, not purple. If you prefer, think of it as negative green. ;-)
Well, since you opened the door, I'll walk through it and say that colors are not directly related to wavelengths. Although this was known and used in the printing industry decades earlier, Edwin Land is generally credited with showing this:
Yes, that image is in full color, but it is formed by overlaid projection of two monochrome slides: one projected with red light and one with white light. The really scary thing is that the projected image can be photographed using color film and the the colors are recorded as if it really was a color image. It even can work using two monochomatic images captured and projected using a pair of narrow-band yellow filters just a few tens of nanometers different in the wavelengths they pass.
Wendy Carlos has a beautiful investigation of the two-color theorem on her website.
My take on all of this is that color does not exist for a pixel, but is an aggregate property of a region within a scene. The perception of colors becomes increasingly consistent as the number of spectral bands sampled is increased. To see colors in a single monochromatic capture, the scene conditions have to be nearly perfect. With two color bands, a much wider range of scene content shows full color -- but, for example, photographing bright line spectra does not produce a color image using two wavelength bands. I don't think that three color bands is particularly special, but for us humans, sampling three bands is enough for the vast majority of scenes (but still not all).
Form that webpage:
Yes, that image is in full color, but it is formed by overlaid projection of two monochrome slides: one projected with red light and one with white light. The really scary thing is that the projected image can be photographed using color film and the the colors are recorded as if it really was a color image. It even can work using two monochomatic images captured and projected using a pair of narrow-band yellow filters just a few tens of nanometers different in the wavelengths they pass.
Wendy Carlos has a beautiful investigation of the two-color theorem on her website.
There he was, staring at a color view, when logic said all that ought be seen right now was the red from the red projector, and white from what was the green projector! In other words, all that ought to have been visible were shades of white, pink and red, and of course, black.
Well, this logic is wrong. What he saw was what the trichromat color theory would predict, plus some visual effects created by adjacent colors and color adaptation. The image was projected with a red projector probably having a spectral curve like some of the red filters for film photography, and a white projector having roughly speaking a more uniform spectrum or at least not so skewed. The range of spectra that they can produce is a 2D space (well, a cone in it), which we project to our 3D vision and still get a 2D space. If you fix the luminosity, you would get 1D (a curve) but that does not mean that the curve would be between red and white, which is what his expectation was.
BTW, those tiny pictures which are supposed to recreate the effect do have blue pixels. I measured (138,179,179), for example, which looks blue to me even in isolation. It looks like his "white" channel was quite bluish. Then the red one just adds positive numbers to the R value. At some point, it would create gray, which was not really part of its "white" which was actually bluish. Well, you'd need to pass to a linear space first but let us not go there.
Finally, I have seen a few cathode tube TVs with one dead channel. The picture looks "full color" but horrible.
OpticsEngineer
Veteran Member
I am glad you posted this. It is one of those things that has kind of been stuck in my mind for decades, but I never knew how to follow up on. One of my college professors was on the committee that established the NTSC color TV broadcast standard. The problem they faced was how to fit color signals into a bandwidth that had been established for black-and-white TV. They didn't have the bandwidth available to do it with three different color bands. So among many things, they looked at using the effect you described.
After a lot of pestering, we got our professor to give us an extra lecture one evening on the development of NTSC. Most of the lecture was devoted to how human vision works and he discussed things like contrast sensitivity function and frame rates. Everything about TV is about making images people agree look good. One would think it would be obvious you can only do that if you understand human vision. But our professor was trying to get across to us that people naturally have a lot of intuition about human vision that is quite incorrect. A naive team will spend a lot of time chasing the wrong goals in designing things if they aren't smart enough to study a problem up front.
At one part of the lecture, he used the slide projector and demonstrated the color effect you described. With one color filter he showed an image of a bowl of fruit. Then he showed the same image with a different color filter. Then he overlaid the two and we saw a full color image. Quite unexpected.
He also added that the reason they didn't use it for broadcast TV was they just couldn't make the images look acceptable for people who were color blind.
Something he added that they got wrong was they went too much for color accuracy. The standard committee assumed the phosphors chosen for color TV's would be the ones they recommended for best color accuracy. But it turned out those phosphors were not as bright as some other ones that were available. When it came to actual sales, what sold best in a department store with bright lighting was the TV image that was brighter. As a consequence, orange was a very difficult color to appear nicely on TV. So things in a color TV studio would often be a painted a somewhat different color than what was intended to be shown in someone's living room. What looked orange on your TV screen was actually more of a brown if you saw it in real life.
After a lot of pestering, we got our professor to give us an extra lecture one evening on the development of NTSC. Most of the lecture was devoted to how human vision works and he discussed things like contrast sensitivity function and frame rates. Everything about TV is about making images people agree look good. One would think it would be obvious you can only do that if you understand human vision. But our professor was trying to get across to us that people naturally have a lot of intuition about human vision that is quite incorrect. A naive team will spend a lot of time chasing the wrong goals in designing things if they aren't smart enough to study a problem up front.
At one part of the lecture, he used the slide projector and demonstrated the color effect you described. With one color filter he showed an image of a bowl of fruit. Then he showed the same image with a different color filter. Then he overlaid the two and we saw a full color image. Quite unexpected.
He also added that the reason they didn't use it for broadcast TV was they just couldn't make the images look acceptable for people who were color blind.
Something he added that they got wrong was they went too much for color accuracy. The standard committee assumed the phosphors chosen for color TV's would be the ones they recommended for best color accuracy. But it turned out those phosphors were not as bright as some other ones that were available. When it came to actual sales, what sold best in a department store with bright lighting was the TV image that was brighter. As a consequence, orange was a very difficult color to appear nicely on TV. So things in a color TV studio would often be a painted a somewhat different color than what was intended to be shown in someone's living room. What looked orange on your TV screen was actually more of a brown if you saw it in real life.
OpticsEngineer
Veteran Member
Just thought it worth a mention that link is to the same Wendy Carlos that gave us Switched on Bach, just in case one is a music lover.
ProfHankD
Veteran Member
Note the sentence of mine above -- which I have tested multiple times. Also note that the colors can be photographed with conventional color film or sensors. This stuff is weird, but very real.
BTW, the result of Land's investigation of this (for two decades) was the multi-scale Retinex algorithm... which still doesn't explain this. ;-)
My point is that it is not weird.
ProfHankD
Veteran Member
The trichromat color theory I know of is about color matching, and Maxwell's finding was that 2 colors is not enough -- which turns out to be wrong-ish. An isolated even-color spot cannot be matched with 2 colors, but most scenes with more complex shading can be. I don't see how trichromat color theory makes this not weird...
The fact that this works with two narrow-band images (two yellows) and still produces full color that can be photographed seems particularly freaky to me. How do two pure yellows get anything to be registered on a sensor as blue, green, red, etc.? Incidentally, my wild guess is that this phenomenon probably involves quantum properties of photons and has nothing at all to do with human vision or the use of three color bands for sensing.
I don't think it's weird. Color matching experiments were designed to eliminate the effects of lots of higher order processing done in humans. There is an incredible amount of that, and to this day it is poorly understood.
My thinking is that it involves high order processing. Someone once told me that half our brain's neurons are involved with vision in one way or another.
I just asked ChatGPT about that:
- The visual cortex (occipital lobe) takes up about 30% of your cerebral cortex
- Compare that to just 8% for touch and 3% for hearing
- But the reach of visual processing extends far beyond the occipital lobe:
- Visual input is processed in parietal and temporal lobes (for motion, identity, location, etc.)
- Involves prefrontal cortex (for attention and decision-making)
- Even affects motor planning and memory centers
And it was not matched. The background is horribly wrong, for example, the yellow looks like white, etc. The logical error is that he expected hues from red to pink to white, which is not what the theory predicts. He gets more colors, completely logical, but not all, also logical.
I do not see two yellows anywhere in the demonstration, which has blue-ish (he thinks it is white/gray) and red. It looks to me that his blue changes hue depending on the luminosity.
What he has seen years ago, I can only guess. What is described as two yellows might be two very different spectra. We can even have two yellow looking sources of light which look identical to us, with very different spectra. Mixing them in different proportions would produce a range of colors to our eyes different from yellow.
ProfHankD
Veteran Member
Here's a link to a freely available copy of Land's paper; look at page 288.
I personally have had similar, highly scene-dependent, results in experiments using narrow-band filters or LED lighting (not the phosphor stuff, real LEDs). Both have nearly all energy in a band less than 30nm wide, in some cases as narrow as 10nm, around the center wavelength. My intuition (or call it wild speculation ;-) ) is that the effect has to do with the fact that the wavelength of a photon is determined by a probability density function (PDF), just like any other quantum property. Thus, even a 599nm photon actually has non-zero probabilities of being every other wavelength and interference between photon wave functions could enforce or degrade these other wavelengths. Perhaps human vision can enhance the perception of colors due to this, but other wavelengths somehow being present at all is pretty convincing evidence that something other than human vision is involved.
Similar threads
Keyboard shortcuts
- f
- Forum