Television sets and video monitors rely on tricking the visual system into believing it is seeing the full range of possible colors. In reality, they are only generating approximations of the light that would actually enter the eye if we were looking at a real object.
The problem is this: the visible spectrum actually consists of an infinite number of possible light wavelengths in the range from 380 to 700 nm; light waves from across that range enter the eye and are capable of activating photoreceptors within the eye: this corresponds to the wide range of colors we can perceive.
But TVs and other electronic displays generally only use three different wavelengths of light -- red, green, and blue. These colors are mixed in varying proportions to create all the other colors we see. But with just these three colors, only a small portion of visible light wavelengths can be approximated (the diagram on the left, taken from Wikipedia, gives you some sense of the disparity: the colored triangle represents the colors a TV can produce, while the gray arc represents all visible colors).
Particularly relevant for CogDaily readers is the fact that the human eye uses a similar trick to perceive light: there are only three (and sometimes four) different photoreceptors in the eye. Each one is activated by a certain set of wavelengths. The brain combines the information from each photoreceptor to rebuild a picture of the color. Why is the eye better than a TV monitor? Because each photoreceptor is activated by a range of wavelengths, while TV pixels can only display one color at a time.
This is where the new video technology comes in to play. The BBC is reporting on a Swiss laboratory which has developed a TV monitor which can display any wavelength from each pixel. The display uses 400 diffraction gratings -- striped membranes which work like a prisms -- to create the colors. Each grating can be independently adjusted to produce any color in the visible spectrum. While 400 pixels isn't enough to display a full picture (the best displays use millions of pixels), the researchers claim that the process can be inexpensively scaled up to full size.
What will be the impact of this new technology?
First off, truer, more realistic colors. The researchers say no current monitor can display a truly blue sky, so pictures will be more realistic. For scientists, there is also the prospect of clearer images in electronic microscopes and telescopes.
- Log in to post comments
More like this
I am underwhelmed by this. We have three different brands of color photoreceptors. If a monitor can mix different ratios of R, G and B to stimulate the different photoreceptors to different degrees, it should be able to do a pretty good job of displaying everything that can be seen.
It's like having two ears. You can get remarkably good stereo effects from a pair of headphones, you don't need to fill the walls of your room with dozens of speakers to produce the range of frequencies and directions of possible sounds.
If a monitor can mix different ratios of R, G and B to stimulate the different photoreceptors to different degrees, it should be able to do a pretty good job of displaying everything that can be seen.
No, this statement is wrong, as I have explained in the post. The ratios of RGB can only create the colors indicated in the triangle above (actually the diagram is oversimplified -- the true diagram is three-dimensional, to account for brightness). There are many visible colors which cannot be duplicated by a monitor.
re: the speaker analogy -- I'd have to defer to Shelley here, but I can say that human ears are remarkably well-equipped to detect directional sound. I'd be skeptical of a headphone's claim to offer true three-dimensional sound.
Overall, the human perceptual system is remarkably well-equipped to fill in the gaps of an incomplete stimulus, whether audio or visual, but that doesn't mean that current technology for producing audio or visual stimuli can't be improved.
I guess I'm a little skeptical of this too. Obviously an RGB monitor cannot duplicate physically many kinds of colored light. But it's not obvious to me that this carries over to perception.
As you say, each photoreceptor is sensitive to a range of wavelengths. But what does that mean as far as the signal being sent to the brain? Is there some pulse train that is simply not reproducible by RGB? Or does, say, light that is slightly off the peak wavelength for a given receptor "alias" to the same signal as weaker light at the peak wavelength?
I'm skeptical because of the dependency on other equipment.
Yeah, the new monitor can display more colors, but if the source material doesn't have that broader range of colors, the new monitor adds nothing.
Its like what will happen with the upcoming star wars theater editions - they only have a digital master from an analog source at a certain resolution (for the laserdisc) and they can't rework that digital source to better resolution to get anamorphic widescreen without revealing limitations in that digital master (e.g., a fuzzy picture). So having it on DVD is nice in that its at least watchable and more permanent since laserdiscs are gone, but you're not gaining the advantages that dvd offers beyond that.
So too with this monitor technology. If my digitizers (cameras, scanners, vidicams, and digital video systems) can't create media that uses that resolution from analog or real-world sources, then being able to display that resolution means nothing.
Contact anyone you please, but all you really need to do to find this out is to get a hold of a nice pair of headphones and put on some Pink Floyd. Dark Side of the Moon is a good choice for this.
And I think you are wrong, which I attempted to explain. The peak efficiencies of the human cone recepetors are blue (437 nm), green (533 nm) and red (564 nm) (disregarding people with colorblind variations). Suppose you want to display a color that is off those peaks; e.g. 490 nm. You don't need to put out photons at 490 nm, you just need to excite the various rods to the extent that they would be excited by color of that wavelength.
To go back to the audio analogy, if you want it to sound like a drum is being beaten 25 degrees to the left of straight ahead, you don't need to put a speaker there, you just need to control the timing of the signal reaching each ear.
somnilista, you are wrong. Pixels can only display red green or blue at one intensity, while cones can be stimulated to different intesities, as Dave explained. The pixels cannot activate cones at every color because of this.
Somnilista, you are right in that you can approximate wavelengths inbetween those ranges. The problem is that three distinct wavelengths can only simulate a region described by a triangle, as depicted above. Even if you move the three wavelengths of the RGB monitor to the boundaries of human capabilities, you still won't fill the entire gray region.
You could imagine a monitor where the wavelengths of the pixels are outside the gray region, so that the entire region of human visual capacity was covered -- but then, the human photoreceptors would be unable to process any signal, because the actual wavelengths of light entering the eye are not capable of activating the photoreceptors.
So, using RGB technology, through trial and error the people designing it chose R, G, B values that were the best compromise. However, there is no doubt that we're talking about a compromise here. A system as proposed by the scientists in the linked article would be superior.
On this issue about stereo:
Actually, stereo headphones have proven unsufficent to really recreate a 3D auditory space (or even 2D in the horizontal plane). You can vary delay and/or intensity as you want, you usually feel the sound coming from "inside" your head, indeed with some angle. This is because the auditory system not only uses ITD (interaural time difference) or IID (interaural intensity difference) to register the spatial origin of sounds but more subtle cues as the alterations of the auditory spectrum produced by the (asymetrical) external and middle ears and the tissues from the head (eg. skull).
To recreate a really 3D percept, experimenters in the field of auditory perception use what is called "head related transfer functions" (HRTF) which consist in mathematical computations of the sound alterations due to the head and the ear. These mathematical functions are then applied to the sound coming from each speaker of the headphones so that each reaches the inner ear as if it had come from a given position in space.
Thus it is only by using HRTF that you can make up a sound coming from not only the left or right hemispace but from one's front or behind (didn't you, somnilista, that with headphones on, you always feel in the middle of the band not as if they were really in front of you). HRTF also allow one to play with elevation (sounds from above or below oneself).
'Hope it helps.
Underwhelmed I am as well -- of course that company would use that colour triangle, which is known to misrepresent the _perceptual_ colour space. The large gamut missing in the top left edge is rather small in perceptual space. The "Lab" representation, for instance, shows this.
Am I the only one unimpressed by a graph whose axes are not labeled, and with no explanation given for what the graph supposedly represents?
You don't have to duplicate the actual color in order to make the eye and brain think it's the same color. It's no accident that the number of different colors in RGB is the same as the number of different color cones in the human eye - RGB was designed by and for humans.
All colors *seen by humans* are forced into a 3-dimensional manifold: red cone response, green cone response, blue cone response. If you can get those 3 all to the correct levels, the human eye won't know the difference.
The new monitor may look better to turtles with 4-color vision, but I'll believe there's a human-visible difference when I see a human pass an ABX test, and not before. (And then I'll see if that human is one of the rare people that actually does have 4-color vision.)
After thinking about it overnight, I partially conceed.
Stuart Coleman is entirely wrong to say that pixels can have only one intensity. That is so 1970s. The computer I'm on now says it can produce millions of colors; the cube root of that will indicate how many shades of intensity are possible for the R, the G and the B subpixel.
I also don't think Dave Munger has done a good job of explaining the situation. My current understanding is:
It does not clarify to talk about an infinity of wavelengths. We still only have 3 cone color receptors (ignoring color-blind people , mutants, etc). So what we see is a triplet of stimulation of our 3 photoreceptor circuits. The full range of possible colors would consist of the product of stimulating each photoreceptor in its full range, (0 - 1) multiplied by the full range of the other photoreceptors. That's a 3 dimensional range.
The reason RGB monitors cannot display all possible colors in that range is that the spectral responses of the 3 sets of photoreceptors in our eyes overlap. Lighting up a green pixel will stimulate our green cones, but will also excite the blue and red cone photoreceptors somewhat. That is why there are regions in that 3D range that cannot be reached with RGB monitors.
Notice in the article where they talk about the possibility of putting two or more colors in each pixel - this shows a weakness of the new technology. Each pixel of reality is not limited to one or even two colors. Whatever mix of colors there is will contribute to the excitation of the eye's photoreceptors, you will get a ratio of R,G and B excitation. Mixing several colors increases the odds that existing technology can suitably approximate it.
Another weakness of the new technology will be efficiency. They would use a full spectrum white light source and throw away all of it except the colors that are chosen for display.
Thinking about it leads me to this: If you could put a chip in the retina, or in the brain, to individually excite the photoreceptor circuits, you could make it possible to see colors that do not exist in nature. You could fully excite the green circuits, for example, without exciting the red or blue circuits due to spectral overlap.
Chris, the axes are labeled, but you're right in that I didn't fully explain it. Color vision, and color theory, are both immensely complex subjects. The main point of emphasis for this post was supposed to be on the new technology, not my brief color vision refresher. For more on how the color space is defined, you might want to check this Wikipedia article.
Somnilista, the issue isn't so much the overlapping colors, it's the fact that the TV can't display colors outside of a certain range -- it simply doesn't generate pixels with a wavelength below or above a certain level. That is indeed an interesting concept re: viewing colors that don't exist in nature. However, remember also that color itself is a human construct. "Color" only exists because our minds interpret electromagnetic waves in a certain way.
Notice in the article where they talk about the possibility of putting two or more colors in each pixel - this shows a weakness of the new technology. Each pixel of reality is not limited to one or even two colors. Whatever mix of colors there is will contribute to the excitation of the eye's photoreceptors, you will get a ratio of R,G and B excitation. Mixing several colors increases the odds that existing technology can suitably approximate it.
Kaylee, this is a huge improvement over existing technology, where each pixel can only display one color: red, green, or blue, in varying intensities. The new technology would allow each pixel to display any hue on the spectrum. Yes, the full range of visible colors would only be generated through mixing, but this is also true of existing technology. The new technology far exceeds the capacity of existing technology.
You seem to have missed my point that we don't see those wavelengths at all. We perceive stimulation on three receptor channels, each of which has a certain peak wavelength and range. Since it has reached the point of repetition, I'll give this a rest now.
I'm not sure what Kaylee is trying to say by repeating a paragraph of mine without comment; perhaps he/she is having trouble with the formatting?
You seem to have missed my point that we don't see those wavelengths at all. We perceive stimulation on three receptor channels, each of which has a certain peak wavelength and range.
No, I didn't miss that point. The point is that RGB monitors don't cover that entire range. There are wavelengths, and combinations of wavelengths, that we can see which aren't generated by RGB monitors.
I would have thought it would be an easier engineering approach simply to pick better RGB values (that is: spectral distributions). Maybe it would be practical to cover almost all the visible gamut satisfactorily. And if not, move to four primaries.
(Current RGB values derive from ones that must have been chosen based on various limitations for manufacturing old TVs.)
But you can't match what our eyes can detect with three or even four colors, no matter how they're combined. It's not possible to define orthogonal unit vectors in the signal space of our rod cells. Any solution that involves the combination of specific colors will leave out some of the combinations our eye can detect.
By the way, it is not really correct to say (in the original article) "TVs and other electronic displays generally only use three different wavelengths of light". Each of the primaries would have a spectral distribution of various wavelengths at different weightings. Although they may be quite spiky.
Looks like I'm coming into this conversation late, but so far it looks like the main argument has been concerning whether RGB is good enough. However, one issue that everyone seems to have missed is that even this new technology alone cannot reproduce all the colors that our eyes (or rather our brains) see. Even though it can conceivably produce any particular frequency of light, there are certain colors that we perceive that are not any particular frequency. The color brown is one example; also the color white. These colors can only be made by combining several frequencies together.