RGB and colors

For the more technical/geeky chat subjects!

Moderator: satrow

User avatar
mr tribute
Moon lover
Moon lover
Posts: 93
Joined: Sat, 19 Mar 2016, 23:24

RGB and colors

Unread postby mr tribute » Sat, 08 Sep 2018, 14:17

RGB is the cheapest possible way to have color reproduction, invented more than 50 years ago. It will never come close to real colors. Look at the light from a monitor/TV in a dark room from another room. You'll see that the light is cold greenish/bluish, because that is what happens when you only use RGB.

They say this monitor can reproduce 99 % of RGB, but they never tell you RGB is unsuitable for quality color reproduction. If you want to see the primary colors refracted by white light through a prism, look at the rainbow. You have 7-8 primary colors.

That being said a yellow subpixel would go a long way to compensate for the cold light from monitors/TVs. That would be RYGB. From the rainbow we are only missing the in-between colors Orange, Lime, Cyan, Magenta.

The fact that yellow can be created from RGB suggests that these are the most "primary" colors, however just using RGB gives an overall greenish/bluish picture. To hide the fact that RGB can't reproduce natural color, modern displays are often delivered with over-saturated colors and a high contrast value. People say "oh, it looks so vivid", totally forgetting that it looks nothing like real world colors.

Before someone says that RGB is the standard for computer software and in that case software has to be rewritten, no that's not necessary. Pentile displays show us that it is trivial for a display to take an RGB signal and calculate in real time (60 Hz) the appropriate way to display that RGB signal on a pentile display. So if more subpixels were added to displays, computer software would stay the same but the display firmware would calculate how to best display the RGB signal on that display.

Sorry about the rant. I have been annoyed by the RGB standard for almost a decade.

Don't get me started on those crazy white LEDs that provide the backlighting on modern displays. It's no coincidence that "low blue light" marketing bullshit became a thing when the warmer CCFL backlighting switched to LEDs. Create a problem then market "a solution". And PWM (pulse width modulation) became a much bigger problem with the much faster LEDs with no "afterglow".

adisib
Lunatic
Lunatic
Posts: 364
Joined: Sat, 13 Jun 2015, 03:34
Location: KY

Re: RGB and colors

Unread postby adisib » Sat, 08 Sep 2018, 18:34

mr tribute wrote:RGB is the cheapest possible way to have color reproduction, invented more than 50 years ago. It will never come close to real colors. Look at the light from a monitor/TV in a dark room from another room. You'll see that the light is cold greenish/bluish, because that is what happens when you only use RGB.

They say this monitor can reproduce 99 % of RGB,


It seems that you are confusing "RGB" with "sRGB". Not all displays are factory calibrated; if your display is too bluish, you may need to create an ICC profile.

mr tribute wrote: If you want to see the primary colors refracted by white light through a prism, look at the rainbow. You have 7-8 primary colors.

That being said a yellow subpixel would go a long way to compensate for the cold light from monitors/TVs. That would be RYGB. From the rainbow we are only missing the in-between colors Orange, Lime, Cyan, Magenta.

The fact that yellow can be created from RGB suggests that these are the most "primary" colors, however just using RGB gives an overall greenish/bluish picture. To hide the fact that RGB can't reproduce natural color, modern displays are often delivered with over-saturated colors and a high contrast value.



Humans, and the most similar primates to them, are trichromats. Trichromat meaning that they have eye receptors for three different wavelengths of light, and thus three components are all that is needed to reproduce the visible color spectrum for humans with normal vision.

User avatar
mr tribute
Moon lover
Moon lover
Posts: 93
Joined: Sat, 19 Mar 2016, 23:24

Re: RGB and colors

Unread postby mr tribute » Sat, 08 Sep 2018, 22:25

adisib wrote:
mr tribute wrote:RGB is the cheapest possible way to have color reproduction, invented more than 50 years ago. It will never come close to real colors. Look at the light from a monitor/TV in a dark room from another room. You'll see that the light is cold greenish/bluish, because that is what happens when you only use RGB.

They say this monitor can reproduce 99 % of RGB,


It seems that you are confusing "RGB" with "sRGB". Not all displays are factory calibrated; if your display is too bluish, you may need to create an ICC profile.


No, it's much simpler than that. Give a painter three colors let's say Red, Green and Blue plus white and black. See what kind of painting he can produce. Now give the same painter more colors and suddenly his creation will be more colorful. In theory you can mix RGB to create a full color palette, but that's only in simplified theory. Both RGB and sRGB are color models/theories. You can give any color RGB coordinates, but it won't be accurately reproduced with just RGB.

Colors are a human invention/interpretation. Visible light between 380 and 750 nm is what truly exists and humans have decided to name certain wavelengths as certain colors. Based on this we can say that 360 or 370 hues exist. Of course nanometer is also a human invention, but since the circle has been given 360 degrees, a color wheel has 360 hues.

You can't pick 3 of those hues and reproduce all other hues. That's just impossible. If you start with three you are able to reach a level when most people say: "This is good enough". But every hue or subpixel you add after that will improve the color reproduction with diminishing return. I would say we need at least 4 hues/subpixels for decent color reproduction. Decent meaning noticeably better than we have today.

Think of hues as pictures in a motion picture. 24 frames per second is enough to create the illusion of motion. The more frames per second you add the closer to the illusion of real motion you'll come. The more hues/subpixels you add the closer to the illusion of full color spectrum you'll come. Reality doesn't mix three hues. Reality mixes whatever hues and any number of hues at any given moment. 360 hues? Still not reality. 1000 frames per second? Still not reality. Only closer to reality.

I think it's enough to look at the blue/green tinted light from a screen in a dark room from another room. Doesn’t matter if it is CRT, plasma or LCD, as long as it is RGB it will be the same.

adisib wrote:Humans, and the most similar primates to them, are trichromats. Trichromat meaning that they have eye receptors for three different wavelengths of light, and thus three components are all that is needed to reproduce the visible color spectrum for humans with normal vision.


From Wikipedia:
"The normal explanation of trichromacy is that the organism's retina contains three types of color receptors (called cone cells in vertebrates) with different absorption spectra. In actuality the number of such receptor types may be greater than three, since different types may be active at different light intensities."

https://en.wikipedia.org/wiki/Trichromacy

adisib
Lunatic
Lunatic
Posts: 364
Joined: Sat, 13 Jun 2015, 03:34
Location: KY

Re: RGB and colors

Unread postby adisib » Sun, 09 Sep 2018, 02:01

mr tribute wrote:No, it's much simpler than that. Give a painter three colors let's say Red, Green and Blue plus white and black. See what kind of painting he can produce. Now give the same painter more colors and suddenly his creation will be more colorful.


Paints and digital colors are not comparable (e.g. RGB is additive, while paints are subtractive). So even if painters only used pre-mixed colors for some strange reason, it is a useless analogy as stated.

mr tribute wrote:In theory you can mix RGB to create a full color palette, but that's only in simplified theory. Both RGB and sRGB are color models/theories. You can give any color RGB coordinates, but it won't be accurately reproduced with just RGB.

...

360 hues? Still not reality. 1000 frames per second? Still not reality. Only closer to reality.



The colors RGB can produce is limited by its bit-depth. If you use 24-bit color in a "RYGB" format, it wont be able to produce more colors than RGB at the same bit-depth (edit: it would actually produce less colors, because of all of the overlap). So yes RGB will only be able to produce a limited number of colors, but that is to be expected when you have a limited amount of storage for the color and is a problem for any other color format. Why are you blaming RGB for the problems of limited bit-depth?

mr tribute wrote:From Wikipedia:
"The normal explanation of trichromacy is that the organism's retina contains three types of color receptors (called cone cells in vertebrates) with different absorption spectra. In actuality the number of such receptor types may be greater than three, since different types may be active at different light intensities."

https://en.wikipedia.org/wiki/Trichromacy


Not sure what you are trying to say here. This doesn't disagree that average humans are trichromatic as I said. On the contrary, the page affirms that reproducing human vision only needs three colors, e.g. "Hermann von Helmholtz later expanded on Young's ideas using color-matching experiments which showed that people with normal vision needed three wavelengths to create the normal range of colors. Physiological evidence for trichromatic theory was later given by Gunnar Svaetichin".
Last edited by adisib on Sun, 09 Sep 2018, 02:04, edited 1 time in total.

User avatar
mr tribute
Moon lover
Moon lover
Posts: 93
Joined: Sat, 19 Mar 2016, 23:24

Re: RGB and colors

Unread postby mr tribute » Sun, 09 Sep 2018, 13:56

The colors RGB can produce is limited by its bit-depth. If you use 24-bit color in a "RYGB" format, it wont be able to produce more colors than RGB at the same bit-depth (edit: it would actually produce less colors, because of all of the overlap). So yes RGB will only be able to produce a limited number of colors, but that is to be expected when you have a limited amount of storage for the color and is a problem for any other color format. Why are you blaming RGB for the problems of limited bit-depth?


Bit-depth would need to be higher. We are not talking building the cheapest possible tech. We are talking building better tech. ;)

I think this document contains what I want to express, but gives a better technical understanding.
(Texas Instruments's BrilliantColor augment the typical red, green, and blue channels with up to three other primaries: cyan, magenta and yellow)

https://www.embedded.com/print/4013039

"By adding yellow, cyan, and magenta colors to the rendering of the image, one can maintain bright white points while providing deeper red, green, and blue color points."

"Any color that can be displayed on the system is some combination of the red, green, and blue colors. Though this color space is suitable for many applications, it does not allow for the creation of vivid colors such as yellow and cyan. The reason for this is that the vivid yellow (or cyan) that we frequently see in nature is outside of the area bounded by the triangle. Adding additional color to the rendering engine allows us to expand the triangle into a wider polygon resulting in a greater selection of colors."

So why isn't this used everywhere? I can think of two reasons:
It's expensive to produce additional subpixels.
Digital images are RGB so additional subpixels have to be computed and if that algorithm isn't perfect it can lead to color distortion.

User avatar
Moonchild
Pale Moon guru
Pale Moon guru
Posts: 22269
Joined: Sun, 28 Aug 2011, 17:27
Location: 58.5°N 15.5°E
Contact:

Re: RGB and colors

Unread postby Moonchild » Mon, 10 Sep 2018, 11:53

I'd just like to add here that Wide-gamut monitors that display colors outside of the default RGB triangle (and then you're talking about sRGB, not RGB as a technology) use just R G and B channels to achieve this. By adding additional pixels it becomes easier to display colors that happen to lie on the "short sides" of the triangle but it's not required to do so if you simply enlarge the triangle (more depth).
Improving Mozilla code: You know you're on the right track with code changes when you spend the majority of your time deleting code.

"If you want to build a better world for yourself, you have to be willing to build one for everybody." -- Coyote Osborne

Goodydino
Lunatic
Lunatic
Posts: 359
Joined: Tue, 10 Oct 2017, 21:20

Re: RGB and colors

Unread postby Goodydino » Wed, 12 Sep 2018, 20:28

mr tribute wrote:No, it's much simpler than that. Give a painter three colors let's say Red, Green and Blue plus white and black. See what kind of painting he can produce. Now give the same painter more colors and suddenly his creation will be more colorful. In theory you can mix RGB to create a full color palette, but that's only in simplified theory. Both RGB and sRGB are color models/theories. You can give any color RGB coordinates, but it won't be accurately reproduced with just RGB.

The primary colours for paints (light absorbers) are red, yellow, and blue. Those for light emitters are red, green, and blue. Your painter would need red, yellow, blue, black and white.

Could someone tell me why the only primary colour that is different is the one in the mid-wavelength range?

adisib
Lunatic
Lunatic
Posts: 364
Joined: Sat, 13 Jun 2015, 03:34
Location: KY

Re: RGB and colors

Unread postby adisib » Wed, 12 Sep 2018, 21:15

Goodydino wrote:The primary colours for paints (light absorbers) are red, yellow, and blue. Those for light emitters are red, green, and blue. Your painter would need red, yellow, blue, black and white.

Could someone tell me why the only primary colour that is different is the one in the mid-wavelength range?


Red, yellow, blue is historical (likely chosen for psychological reasons).
For light emitters, red, green, blue are chosen for better aligning with the three color receptors in human eyes and thus having more colors in the visible range for the same bit depth.
Paints (light absorbers) typically now use magenta, cyan, yellow as primaries. This is because magenta absorbs green, cyan absorbs red, and yellow absorbs blue, so that it is still thinking in terms of RGB but in the subtractive way instead of an additive way.

Goodydino
Lunatic
Lunatic
Posts: 359
Joined: Tue, 10 Oct 2017, 21:20

Re: RGB and colors

Unread postby Goodydino » Thu, 13 Sep 2018, 21:30

By definition, primary colours are those that cannot be produced by mixing other colours together. You try producing yellow paint by mixing colours, without having yellow paint in the first place, and see how far you get. The same thing applies to green light. Try producing green light by mixing other light colours together and see what happens. If I want green paint, on the other hand, I need only to mix yellow paint with blue paint. If I want yellow light, I can mix green light with red light.

Now, why is the primary colour in the middle part of the visual spectrum the only one that is different? Conversely, why are red and blue primary for both emitters and absorbers?


Return to “Technical chat”

Who is online

Users browsing this forum: No registered users and 1 guest