- Status
Tindytim
Distinguished
- Sep 16, 2008
- 506
- 0
- 18,930
[citation][nom]Glorian[/nom]If adding a yellow pixel is better why wouldn't adding a magenta and cyan pixel work as well? Kinda gimmicky imo. Still screwed if you are colorblind. Oh well.[/citation]
Exactly. There is no currently available format that even supports those pixels, much less any color space they would add (as I stated in the comment that got -4'd).
Exactly. There is no currently available format that even supports those pixels, much less any color space they would add (as I stated in the comment that got -4'd).
Gin Fushicho
Distinguished
- Mar 11, 2009
- 645
- 0
- 18,930
I think a little bit of thinking about the difference between theory and practice is in order here. I know in the printing industry (labels and signs etc) you are limited to only so many shades with just 4 colors. To get super realistic graphics they usually use green and orange to the CMYK mix. I can imagine it's similar with LEDs, as the article mentioned some colors especially metallics such as brass appear really muted on LED/LCD screens.
Tindytim
Distinguished
- Sep 16, 2008
- 506
- 0
- 18,930
[citation][nom]Rob Roy[/nom]I think a little bit of thinking about the difference between theory and practice is in order here. I know in the printing industry (labels and signs etc) you are limited to only so many shades with just 4 colors. To get super realistic graphics they usually use green and orange to the CMYK mix.[/citation]
The color gamut is essentially infinite. But with a display, especially when it comes to certain formats like NTSC, the color space is limited, so an increase doesn't make things any better. Read this:
http://en.wikipedia.org/wiki/Color_space
[citation][nom]Rob Roy[/nom]I can imagine it's similar with LEDs, as the article mentioned some colors especially metallics such as brass appear really muted on LED/LCD screens.[/citation]
*facepalm*
The difference between flat and glossy colors it the way the base color changes based on how it interacts with light. That has nothing to do with actually displaying that color, since you wouldn't want an actual reflective material, you'd want to convey what was being reflect in the scene, not your living room.
The color gamut is essentially infinite. But with a display, especially when it comes to certain formats like NTSC, the color space is limited, so an increase doesn't make things any better. Read this:
http://en.wikipedia.org/wiki/Color_space
[citation][nom]Rob Roy[/nom]I can imagine it's similar with LEDs, as the article mentioned some colors especially metallics such as brass appear really muted on LED/LCD screens.[/citation]
*facepalm*
The difference between flat and glossy colors it the way the base color changes based on how it interacts with light. That has nothing to do with actually displaying that color, since you wouldn't want an actual reflective material, you'd want to convey what was being reflect in the scene, not your living room.
It should be clear that adding yellow color( Wawelenght 570–590 nm )is nonsence. We cant see that kind of color and this color would be real yellow.If you want yellow pixel it must contain all three colors R G B, that we can see (wawelenghts blue 450–495 nm, green 495–570 nm, red 620–750 nm).So the only way to add yellow pixel is to add additional pixel with three light emiters (RGB)and it is hard to understand why would you need that kind of pixel.
http://en.wikipedia.org/wiki/Color_vision
http://en.wikipedia.org/wiki/Visible_spectrum
http://en.wikipedia.org/wiki/Color_vision
http://en.wikipedia.org/wiki/Visible_spectrum
Although I didn't read the article, I'm assuming that, typically, screen red and green pixels have too much blue in them. This is simply because the spectrum (absorbtive or emissive) of the materials we use isn't perfect.
For below, "eye" is what you actually see, "scr" is the intensity of the screen pixels, and "img" is the intended color.
If:
R(eye) = R(scr)
G(eye) = G(scr)
B(eye) = B(scr) + r * R(scr) + g * G(scr)
Then you can never display a color where B(img) < r * R(img) + g * G(img). Ie, you can't display less blue light than the unwanted blue that's coming from the red and green pixels.
We instead adjust the color of all the screen's pixels to have a slight blue bias, so that the screen is capable of displaying the desired image smoothly.
However, if you add yellow pixels, you can do:
R(eye) = R(scr) + Y(scr)
G(eye) = G(scr) + Y(scr)
B(eye) = B(scr) + R(scr) * r + G(scr) * g
You can create a function mapping RGB(img) to RGBY(scr) that's smooth (if not linear), and gives nicer than normal results where the smallest of R & G is much higher than B, ie colors near yellow.
The screen still only needs 3 color values per pixel, but now it uses these values to decide the intensities of the 4 screen pixel colors via mapping functions.
It doesn't seem to be what their screen is doing, however. Having information about 4 colors is a waste as there's only 3 degrees of freedom.
Of course, this could work for pixels other than yellow, depending on what spectrums are currently available from known materials.
I'm thinking that simply increasing thickness and combining different absorbtive materials should be able to yield ever-improving color distinction, although at the cost of wasted power.
Eg: If 100 um of green filter material lets 90% of green light through and 10% of blue through, then 200 um of green filter lets 81% of green light through and 1% of blue light through, although power efficiency decreases by 10%.
However, that's too simple, as there are more than 3 different wavelengths of visible light. But if you combine absorbtive materials which have a maximum permitted light intensity wavelength either side of the desired color, some combination of the two materials will give a maximum permitted light intensity wavelength of exactly the desired color. Increase the thickness of this material, and increase the backlight intensity accordingly, and your color will become nicer.
Of course, absorbed light, (in addition to being wasted power), is turned into heat, so the thicker the filter the hotter your screen will get.
I'm not really sure how thick LCD screen color filters usually are.
One other thing to mention - if you're slightly colorblind (like me) you may have slightly different absorbtion spectrums for your three cone types. You may then find the RGBY screen looks nicer. But a screen tailored specifically to your cones would look far better again. (IIRC they can grow the pigment from your DNA and know the absorbtive spectra of many common genetic colorblindness alleles).
Unfortunately, if an photo's color is perfectly tuned to a person with normal color vision, it is possible that some colorblind people might actually not get all the information from the photo that they could get from looking at the real world scene. For example, I'm looking into a way to use two different pigments that look the same to a regular person but can be distinguished by people with slightly different spectra (like me). Thus I can write signs that only people like me can see. But photos probably won't pick this information up.
For below, "eye" is what you actually see, "scr" is the intensity of the screen pixels, and "img" is the intended color.
If:
R(eye) = R(scr)
G(eye) = G(scr)
B(eye) = B(scr) + r * R(scr) + g * G(scr)
Then you can never display a color where B(img) < r * R(img) + g * G(img). Ie, you can't display less blue light than the unwanted blue that's coming from the red and green pixels.
We instead adjust the color of all the screen's pixels to have a slight blue bias, so that the screen is capable of displaying the desired image smoothly.
However, if you add yellow pixels, you can do:
R(eye) = R(scr) + Y(scr)
G(eye) = G(scr) + Y(scr)
B(eye) = B(scr) + R(scr) * r + G(scr) * g
You can create a function mapping RGB(img) to RGBY(scr) that's smooth (if not linear), and gives nicer than normal results where the smallest of R & G is much higher than B, ie colors near yellow.
The screen still only needs 3 color values per pixel, but now it uses these values to decide the intensities of the 4 screen pixel colors via mapping functions.
It doesn't seem to be what their screen is doing, however. Having information about 4 colors is a waste as there's only 3 degrees of freedom.
Of course, this could work for pixels other than yellow, depending on what spectrums are currently available from known materials.
I'm thinking that simply increasing thickness and combining different absorbtive materials should be able to yield ever-improving color distinction, although at the cost of wasted power.
Eg: If 100 um of green filter material lets 90% of green light through and 10% of blue through, then 200 um of green filter lets 81% of green light through and 1% of blue light through, although power efficiency decreases by 10%.
However, that's too simple, as there are more than 3 different wavelengths of visible light. But if you combine absorbtive materials which have a maximum permitted light intensity wavelength either side of the desired color, some combination of the two materials will give a maximum permitted light intensity wavelength of exactly the desired color. Increase the thickness of this material, and increase the backlight intensity accordingly, and your color will become nicer.
Of course, absorbed light, (in addition to being wasted power), is turned into heat, so the thicker the filter the hotter your screen will get.
I'm not really sure how thick LCD screen color filters usually are.
One other thing to mention - if you're slightly colorblind (like me) you may have slightly different absorbtion spectrums for your three cone types. You may then find the RGBY screen looks nicer. But a screen tailored specifically to your cones would look far better again. (IIRC they can grow the pigment from your DNA and know the absorbtive spectra of many common genetic colorblindness alleles).
Unfortunately, if an photo's color is perfectly tuned to a person with normal color vision, it is possible that some colorblind people might actually not get all the information from the photo that they could get from looking at the real world scene. For example, I'm looking into a way to use two different pigments that look the same to a regular person but can be distinguished by people with slightly different spectra (like me). Thus I can write signs that only people like me can see. But photos probably won't pick this information up.
Possibly slightly off the point, but I find it really quite amusing when there are online videos or tv adverts for screens or speaker systems trying to show how much detail or better picture/sound is possible. "look how good this HD picture is through your SD television" "can you hear how much better these speakers are than the ones you're listening to this advert through?"
They can show us Sulu's reaction but they can't show us the colours - if anything it backfires when you see a video and thing 'ooh, that looks nice' because looking nice means "my current display is fine for me, ta"...
They can show us Sulu's reaction but they can't show us the colours - if anything it backfires when you see a video and thing 'ooh, that looks nice' because looking nice means "my current display is fine for me, ta"...
WyomingKnott
Distinguished
- Aug 29, 2006
- 1,139
- 2
- 20,310
A thought. High-end photo printers frequently have more than three colors plus black. And the plus-black is fairly common, even though the theoretical argument can be made that black equals cyan plus magenta plus yellow.
Given that a well-established industry has determined that more colors works better than theory for subtractive colors (pigment), shouldn't we give the same idea a chance for additive colors (light)?
After all, when I did theater lighting in high school, we had more than three colors of filters. (OK, that's a specious argument, but it sounds good!)
Given that a well-established industry has determined that more colors works better than theory for subtractive colors (pigment), shouldn't we give the same idea a chance for additive colors (light)?
After all, when I did theater lighting in high school, we had more than three colors of filters. (OK, that's a specious argument, but it sounds good!)
zak_mckraken
Distinguished
- Jan 16, 2004
- 868
- 0
- 18,930
@doorspawn : You didn't read the article, yet you make a comment which is longer than the article itself. Ironic.
Our eyes don't see the same red green blue as the displays produce. The response curves of our cones (photoreceptor cells in the retina of the eye) overlap considerably and shifted closer together than the digital RGB. You could say our eyes see and interpret an approximation of colour. Some people are red-green colour-blind because 2 of the 3 types of cones have their peak sensitivity (wavelength response) too similar. Some people have 4 types of cones with 4 different response curves. The yellow pixels will still change the range of colours shown and detected by our eyes and may allow the manufactures to shift the red and green away from yellow to make room and richer colours.
I visited my local Best Buy Museum of Contemporary Culture last night and was able to compare three Sharp four-color sets against other high-end displays. I should warn that I fail one of the supposedly less-important R-G colorblindness tests, so my experience may not be typical.
As far as I could tell, Sharp's yellows looked identical to conventional displays. Where the differences appeared was on the orange to rose side of yellow, and on the yellow-green to lime side. Once I saw it, it was dramatic and unmistakable. On conventional displays, a whole range of orange and reddish-rose colors shift toward a particular "Day-Glo" fluorescent rose, and another range of yellow-greens become a particular fluorescent lime. Areas that shifted toward these two shimmering shades seemed to "bloom" a bit like overdriven reds on early color CRTs. On the four-color sets, the same areas were solid, realistic renditions of unique colors that never appeared on the conventional sets.
It makes sense to me, that "fluorescent" colors have spikes of wavelengths that play tricks on the curves of our cones' sensitivities, and that filling in other spectral bands could reduce the experience of fluorescence. (If you're trying to see this effect, beware that viewing some conventional sets off-axis seems to exaggerate the shifts of the "difficult" colors.)
Unfortunately for Sharp and the four-color concept, I fear a lot of people prefer the "Day-Glo" color rendition, and would consider the more realistic four-color image boring by comparison. But if you prefer your display calibrated instead of "electric", you owe it to yourself to stand in front of a four-color set for awhile and notice the grass and roses.
As far as I could tell, Sharp's yellows looked identical to conventional displays. Where the differences appeared was on the orange to rose side of yellow, and on the yellow-green to lime side. Once I saw it, it was dramatic and unmistakable. On conventional displays, a whole range of orange and reddish-rose colors shift toward a particular "Day-Glo" fluorescent rose, and another range of yellow-greens become a particular fluorescent lime. Areas that shifted toward these two shimmering shades seemed to "bloom" a bit like overdriven reds on early color CRTs. On the four-color sets, the same areas were solid, realistic renditions of unique colors that never appeared on the conventional sets.
It makes sense to me, that "fluorescent" colors have spikes of wavelengths that play tricks on the curves of our cones' sensitivities, and that filling in other spectral bands could reduce the experience of fluorescence. (If you're trying to see this effect, beware that viewing some conventional sets off-axis seems to exaggerate the shifts of the "difficult" colors.)
Unfortunately for Sharp and the four-color concept, I fear a lot of people prefer the "Day-Glo" color rendition, and would consider the more realistic four-color image boring by comparison. But if you prefer your display calibrated instead of "electric", you owe it to yourself to stand in front of a four-color set for awhile and notice the grass and roses.
- Status
Similar threads
TRENDING THREADS
-
Question ASUS Laptop Vivobook 17 Model X1704Za SO-DIMM RAM upgrade max size
- Started by zillah
- Replies: 7
-
-
Question Swaping two diffrent tv screens, is it possible?- Started by Nexmu
- Replies: 3
-
-
-
-
Facebook
Twitter