4+ color display?

[PabloDons]

There is a very interesting graph of "color perception" by combining different wavelengths of light at varying intensities:

Not gonna try to explain it, there's great resources online. Search for Chromaticity diagram.

 

By selecting 3 points (colors) on this graph, you can create a triangle where the space inside is all possible colors you can make with those colors at different intensities. Most displays are rec.709 which is defined by 3 coordintes on this graph. That means there is a large space of colors you displays cannot reproduce. There are color spaces that cover more though, like rec.2020 for example, but it's still tri-color and misses a lot of the cyan:

I thought, what if we used 4 colors? Then it would cover a much bigger part of the graph. I couldn't find any displays that would do this, so I wanted to try to make one. Obviously I can't make a full on HD display, but I can make a single pixel by pointing dimmable lasers at one spot.

 

So this is a project I want to undertake and wanted to ask around for any experience y'all can share or any advice. Like what wavelengths do I choose? How I would dim lasers? Any other approach that would make more sense? I don't know much about color other than this, so any help is appreciated!

[Arika]

Curious to know what 4th colour you would use? given that all light is made up of the 3 primary colours; Red, Green, Blue. and those are the only colours the cones in our eyes can see. Any "4th" colour will just be made up of a portion of the already existing colour, making it largely pointless.

[Vangeli]

12 hours ago, comander said:

 

LG has been doing this for a while. You can "lose" resolution this way though, at least at a given marketed resolution.

 

https://www.techhive.com/article/3240976/some-lg-4k-lcd-tvs-deliver-only-28k-resolution.html

 

Printers also use an inverse process for CMYK for printing. 

Was about to say that TVs have had this already. Didn’t Sharp or Sony also have a fourth yellow pixel alongside RGB?

[WkdPaul]

* thread moved to the Displays section *

[Glenwing]

3 hours ago, Vangeli said:

Was about to say that TVs have had this already. Didn’t Sharp or Sony also have a fourth yellow pixel alongside RGB?

Yes, but it doesn't actually do anything.

https://web.archive.org/web/20190531124601/http://www.eecs.qmul.ac.uk/~tb300/pub/Appendix_Quattron.pdf

Quote

 This minima point at the intersection between the red and green primaries also suggests that yellow primary is produced by simply combining the red  and green primaries.  If yellow is implemented as red+yellow+green then a lack of an independent yellow primary will reduce a 4’th ‘yellow’ sub-pixel to equivalency with red and green sub-pixels whose surface area is increased to include that of the ‘yellow’ sub-pixel.  On this basis it may be concluded that Quattron is a “gimmick”.  It has a yellow sub-pixel but no yellow light to pass through it.  Its only function can therefore be to allow more green and red light to pass through, which is equivalent to passing the same light through green and red pixels that are 50% larger.  It therefore serves no useful function, as suggested by the critics.

 

[PabloDons]

Sorry for the late reply! I went to sleep and forgot about it. Worst short term memory ever D: Thank y'all for the replies!

On 4/23/2021 at 3:05 AM, comander said:

 

LG has been doing this for a while. You can "lose" resolution this way though, at least at a given marketed resolution.

 

https://www.techhive.com/article/3240976/some-lg-4k-lcd-tvs-deliver-only-28k-resolution.html

 

Printers also use an inverse process for CMYK for printing. 

 

I know about the white pixel thing, but that's really a brightness feature, not a color feature. I'm thinking about a 4th color pixel to increase the space of colors reproducible by the display. Whereas white isn't a "color" per-se, it's a combination of a range of colors in our visible spectrum

On 4/23/2021 at 3:35 AM, Arika S said:

Curious to know what 4th colour you would use? given that all light is made up of the 3 primary colours; Red, Green, Blue. and those are the only colours the cones in our eyes can see. Any "4th" colour will just be made up of a portion of the already existing colour, making it largely pointless.

Color in the real world isn't made up of the primary colors. It's made up of a spectrum of wavelengths of light. Displays reproduce these colors using combinations of the primary colors, which look identical to our eyes.

Also the cones in our eyes are sensitive to a range of wavelengths at different sensitivities. The red, green, blue cones idea is really just a misconception.

The actual color that you perceive is an interpretation of the signals from these cones. It's not 1:1. Notice that violet isn't really a "real" color, It's a combination of the short and long wavelength cones both firing. The reason it's on the left side of the wavelength graph is because the long wavelength cones are actually slightly sensitive to the very short wavelength light and it's interpreted by our brain as purple.

 

To answer your question, though. I would choose red and blue because they are at the very edge of the horseshoe graph in the OP. But also 540nm green and maybe somewhere in the cyan? I'm not sure because I have no idea how those colors look and what benefit in terms of saturation I would get perceptually. Though the idea is to cover as much of the chromaticity graph as possible

[PabloDons]

I'm also searching around for lasers of specific wavelength, but it appears to be a challenge. Does anyone know of an easy way of producing any coherent wavelength?

[LogicalDrm]

*** Thread cleaned ***

Quote

• Ensure a friendly atmosphere to our visitors and forum members.
• Encourage the freedom of expression and exchange of information in a mature and responsible manner.
• "Don't be a dick" - Wil Wheaton.
• "Be excellent to each other" - Bill and Ted.
• Remember your audience; both present and future.

 

[Mihle]

On 4/23/2021 at 3:35 AM, Arika S said:

Curious to know what 4th colour you would use? given that all light is made up of the 3 primary colours; Red, Green, Blue. and those are the only colours the cones in our eyes can see. Any "4th" colour will just be made up of a portion of the already existing colour, making it largely pointless.

All light is not only made from those colours, no. Light from the sun have actually all different colours.

 

Reason why RGB can look like all colours to your eyes is that the cells in the eye that detect light is basically RGB.

Or, your eyes convert light that is all kinds different colours to 3 different colours (plus some cells that are just black and white basically), so for the eyes it looks the same.

 

 But if you were to shine a light that is RGB set to white on a object that in the sun look like an in between colour of those 3, it doesn't necessarily look the same, exactly because the RGB isn't all colours of light. Even if directly looking on the light/sun or it lighting a grey scale object, looks the same. 

 

RGB is just faking other colour light, it isn't actually that.

 

EDIt: saw now that OP basically replied with this already

[PabloDons]

19 hours ago, comander said:

Having pure white as an option allows for a bit more accurate representation of a handful of colors, otherwise white is attained by color mixing, which is imperfect. 

There might be anecdotal evidence, but as far as theory goes, the perceived color should be exactly identical. Except of course when illuminating objects, which bright TVs tend to do.

 

[akio123008]

There's something else you need to consider; where is the information for your fourth color going to come from? Say hypothetically, you manage to build this display, how are you going to get the data for the fourth color? All cameras I'm aware of have RGB sensors.

 

Sure you could derive the amount of that fourth color from the RGB va-   oh wait...

 

 

[PabloDons]

21 minutes ago, akio123008 said:

There's something else you need to consider; where is the information for your fourth color going to come from? Say hypothetically, you manage to build this display, how are you going to get the data for the fourth color? All cameras I'm aware of have RGB sensors.

 

Sure you could derive the amount of that fourth color from the RGB va-   oh wait...

 

 

I'm glad you asked!

 

So like I said, I'm not looking to make a full-on grid pixel display, just a single pixel really. There's no reason to as it's not really viable anyway or even useful. The data would come from the chromaticity diagram in the OP where I would basically just select coordinates and the build would hypothetically just reproduce that color. That's what I'm interested in, really. Wide color reproduction.

 

Say hypothetically, I do build a display anyway. Well I don't have to use RGB encoded sources. I can use chromaticity diagram (ex. Digital Cinema Package) or wide gamut (ex. Dolby Vision) encoded sources. Digital art is also a use case.

 

Also to reiterate on this RGB sensor misconception. cameras (and our eyes for that matter) are not really RGB, the filters in a camera are sensitive with varying degree to a range of different wavelengths. Our color reproduction systems (monitors) are what are RGB. Any camera with RAW recording is capable of reproducing any color on the color gamut (as long as at least 2 filters are sensitive to it). Even if it's not RAW, you can just color grade anyway.

 

Take a look at the sensitivity of a D70's filters below. The "red" filter picks up some green light too. Doesn't mean all green is going to be yellow-ish. It's not 1:1. There's a bunch of convoluted math going on to determine exactly what color something was based on these 3 inputs, and then some to determine how to reproduce it using outputs (RGB (or in my case, any number of colors)). RAW simply records these 3 inputs as-is.

 

[Hairless Monkey Boy]

Given the picture that you provided in the OP, it would seem that another subpixel on the blue side of green would provide the greatest amount of additional area to the chart.

 

 

Interestingly, the fifth pixel would also be green following this logic.

 

 

[PabloDons]

17 minutes ago, HairlessMonkeyBoy said:

Given the picture that you provided in the OP, it would seem that another subpixel on the blue side of green would provide the greatest amount of additional area to the chart.

 

 

Interestingly, the fifth pixel would also be green following this logic.

 

 

It is very interesting. Clearly, the bureaucrats don't think cyan is important enough to sacrifice some of the yellow for.

 

This is the kind of advice I was looking for, though! A greedy algorithm could work. Though if I were to move the green subpixel down to say 550nm, and the blue-green one up, I'd get more saturated yellow, while keeping the area the same. But maybe 5 colors is the way to go anyway.

[PabloDons]

off-topic fun fact: while searching for a 630nm light source, I encountered "light therapy" devices, and all of them included the same 2 other wavelength lights (660nm and 850nm). Clearly a placebo hoax, though it was strangely consistent and there were so many, I had a hard time finding regular-ass LEDs. Thought I'd share

[tikker]

1 hour ago, PabloDons said:

It is very interesting. Clearly, the bureaucrats don't think cyan is important enough to sacrifice some of the yellow for.

There may perhaps partly be a physical reason to "ignore" cyan, because he distribution of cones isn't uniform over our retina. The fovea, the part of our eyes that gives us sharp vision, has a lack of blue cones and the blue cones are less senstive relative to the red and green ones. The combination might make a lack of cyan less noticeable? I also don't particularly often see cyan in real life, compared to yellow, even though I think it's a pretty colour, which may help even more.

1 hour ago, PabloDons said:

off-topic fun fact: while searching for a 630nm light source, I encountered "light therapy" devices, and all of them included the same 2 other wavelength lights (660nm and 850nm). Clearly a placebo hoax, though it was strangely consistent and there were so many, I had a hard time finding regular-ass LEDs. Thought I'd share

Well the 850 nm is obvious because it wouldn't be an infrared light if that was missing. I also wonder how monochromatic in regards they are to 630 and 660

[Glenwing]

1 hour ago, PabloDons said:

off-topic fun fact: while searching for a 630nm light source, I encountered "light therapy" devices, and all of them included the same 2 other wavelength lights (660nm and 850nm). Clearly a placebo hoax, though it was strangely consistent and there were so many, I had a hard time finding regular-ass LEDs. Thought I'd share

Scam devices don't do design work, so they're often just copied from a single source, so they're all the same. And often multiple sellers are really the same as well.

[akio123008]

12 hours ago, PabloDons said:

Also to reiterate on this RGB sensor misconception. cameras (and our eyes for that matter) are not really RGB,

Could be, but the output data from the camera is encoded as values from the pixels on the sensor. There are three kinds of color filter in front of that sensor.

 

That means if you were to generate a fourth "primary color" from this data, you'd be deriving it from those values, so you'd be doing nothing different than taking a bitmap image and deriving a fourth color from existing RGB values stored on a disk. There's no new information there.

 

Now I don't think this is in the budget, but perhaps the perfect solution that you'd be looking for would be a tunable laser: https://en.wikipedia.org/wiki/Tunable_laser

Some of these can produce light between 400-1000nm, so basically the entire spectrum. You'd have all colors covered.