High resolution visible spectrum

[Teck]

I'm trying to find a high resolution image that shows the visible electromagnetic spectrum with a fine graded scale. It should be detailed enough to pinpoint which exact colour corresponds to a particulate wavelength (integer in nanometer) of light. I find a lot of images through searches but no one that is detailed.

For example, if we say 503 nm. I want to be able to find which exact colour corresponds to that wavelength.

The importance here is the accuracy of the colour (and I know different monitors can display colours differently, but let's not get into that :D )

Please link such an image.

 

[phinds]

Well, it's a lot worse than just different monitors displaying colors differently, which of course they do (and often by a LOT), there is also the fact that color perception differs between people, so I'm pretty sure something down to the single-digit nanometer scale would not even be meaningful in any practical sense. That may be why you can't find one.

 

[A.T.]

I know different monitors can display colours differently, but let's not get into that :D

It's not just the difference between monitors. The same pixel value on the same monitor can be perceived by different people as equal or not equal to some monochromatic light with a specific wavelength.

 

[davenn]

For example, if we say 503 nm. I want to be able to find which exact colour corresponds to that wavelength.

You cannot,

just to reinforce what @phinds said ...

colour is subjective, what you see at that wavelength is likely to be different than what someone else sees
Colour is a human perception

 

[russ_watters]

I'm trying to find a high resolution image that shows the visible electromagnetic spectrum with a fine graded scale. It should be detailed enough to pinpoint which exact colour corresponds to a particulate wavelength (integer in nanometer) of light.

Others have pointed out theoretical problems, but I'm not even clear what you are actually looking for (whether it exists or not). A picture of colors like a spectrum doesn't really show readable data; are you looking for a table of rgb values?

Edit: More on that...
From this description it isn't even clear to me if you recognize how very limited this request is. Humans see and monitors attempt to reproduce a 3D color space/array whereas the electromagnetic spectrum represents a single line in that array. In other words, the ~300 steps in the spectrum could be mapped roughly to 256 RGB values in a typical 8x3 bit color representation, but that would miss almost all of the more than 16 million colors.

Nature is even more complicated. Besides being analog, there are two different ways of producing color, and an enormous number of possibilities. Consider emission lines: you could say that every emission line is its own "primary color" and describe a star's emission spectrum as an array with hundreds (thousands?) of dimensions.

 

[member 656954]

Would help to understand what you need such a diagram for, but also take into account that the spectrum does not contain all the colors we can distinguish. Some colors can only be made from a mix of multiple wavelengths. And color perception is dependent upon the interaction of all receptor cells with light, and this combination results in nearly trichromic stimulation. Plus, color sensitivity shifts with variations in light levels, so that blue colors look relatively brighter in dim light and red colors look brighter in bright light.

Unfortunately, you are asking for definitive labels for something that has not been codified into a standard.

 

[A.T.]

colour is subjective, what you see at that wavelength is likely to be different than what someone else sees

How we experience light wavelengths, colors (or anything) is not comparable because it's not quantifiable, so it's meaningless to say that it's different or that it's equal. See:
https://en.wikipedia.org/wiki/Explanatory_gap
But the issue here is if all people can agree on the mapping between monochromatic light and mixed light (e.g. RGB from a monitor) with a great spectral resolution. I don't think they can.

 

[russ_watters]

Would help to understand what you need such a diagram for, but also take into account that the spectrum does not contain all the colors we can distinguish. Some colors can only be made from a mix of multiple wavelengths.

I was thinking something similar and added a section to my post describing this issue in more detail. Maybe the OP's request is limited to the spectrum itself, but yes that would be an extremely limited description of what "color" is. The old cliche "all the colors of the rainbow" is sweet, but really misses the point of what color is and how limited a rainbow is.

 

[davenn]

How we experience light wavelengths, colors (or anything) is not comparable because it's not quantifiable, so it's meaningless to say that it's different or that it's equal. See:

exactly which is what I said :)

 

[russ_watters]

How we experience light wavelengths, colors (or anything) is not comparable because it's not quantifiable, so it's meaningless to say that it's different or that it's equal. See:

I don't think that's true. It may be difficult and lack precision, but if we couldn't compare color perceptions, we couldn't for example define/describe different types of color-blindness.

Nor do I necessarily agree that something has to be quantifiable to be comparable...or perhaps from the other direction, you can quantify pretty much anything if you want...whether that's useful in making the comparison is another story.

 

[Teck]

Although the psychological and philosophical implications of this question are surely interesting, I request that those conundrums be taken in another thread (feel free to create a new one especially for this if you want to continue that discussion).

What I'm looking for here is something practically useful, hence the best possible approximations.

So to make it more clear. Think of my request as one that maps integer wavelengths into single entries within a gamut. Note I do not wish to map in the other direction, so the image only needs to include about 360 different colours.

Kind of similar to the principle behind a plot like this: https://mathematica.stackexchange.com/questions/85990/how-to-plot-an-emission-spectrum (Note that this just an example, I'm NOT looking how to plot in Mathematica...)

 

[russ_watters]

As far as I can tell, this image from the wiki on "spectral colors" meets your required level of precision. If you want to pick off the values, you can do that with pretty much any photo editor. All caveats discussed above apply.

 

[A.T.]

... we couldn't compare color perceptions, we couldn't for example define/describe different types of color-blindness. ...

Yes, we can compare the ability to differentiate colors, but not how different people perceive those colors.

 

[A.T.]

As far as I can tell, this image from the wiki on "spectral colors" meets your required level of precision. If you want to pick off the values, you can do that with pretty much any photo editor. All caveats discussed above apply.

View attachment 243994

@Teck Note that the above is the PNG version generated by PF. The original version has smooth transitions:
https://en.wikipedia.org/wiki/File:CIE1931xy_blank.svg

 

[Teck]

As far as I can tell, this image from the wiki on "spectral colors" meets your required level of precision. If you want to pick off the values, you can do that with pretty much any photo editor. All caveats discussed above apply.

View attachment 243994

Thanks but that picture is not what I'm looking for. The picture needs to be a 1D line graph. Actually, even a 360x2 array with just integer wavelengths and corresponding color codes would do.

 

[Ibix]

Here's a fun example, from xkcd's take on a meme that did the rounds five or so years ago:

https://xkcd.com/1492/

The two girls are wearing different colour dresses (particularly obvious around the neckline, to my eye). Clearly. Unless you check the colour values with a paint program. So any answer you get will be wrong and Russ' diagram is as good as any other - just follow the line around the edge to get 1d.

What do you actually want this for? Knowing that might help us narrow down your question to something that can be answered meaningfully.

 

[russ_watters]

Thanks but that picture is not what I'm looking for. The picture needs to be a 1D line graph. Actually, even a 360x2 array with just integer wavelengths and corresponding color codes would do.

I doubt you will find this already made. You may have to spend an hour constructing it yourself from that data.

Edit:
Ehh, someone had to have made that graph. You may be able to find the data and/or equations behind it. But it still might be faster to generate it yourself.

 

[Teck]

I doubt you will find this already made. You may have to spend an hour constructing it yourself from that data.

Edit:
Ehh, someone had to have made that graph. You may be able to find the data and/or equations behind it. But it still might be faster to generate it yourself.

I'm okay with making the graphy myself, but I need the data for it. Like I said, a 360x2 table would be fine.

 

[Drakkith]

I'm okay with making the graphy myself, but I need the data for it. Like I said, a 360x2 table would be fine.

Here you go. It may take a while, but you can get RGB values from this wavelength-to-RGB value calculator: https://www.johndcook.com/wavelength_to_RGB.html

I have no idea how accurate this is unfortunately.

 

[russ_watters]

I'm okay with making the graphy myself, but I need the data for it. Like I said, a 360x2 table would be fine.

The data is on the graph. All you have to do is read it (with a photo editor) and type it into a spreadsheet/table.

 

[russ_watters]

Here you go. It may take a while, but you can get RGB values from this wavelength-to-RGB value calculator: https://www.johndcook.com/wavelength_to_RGB.html

I have no idea how accurate this is unfortunately.

I'm not sure I understand that; is the output hexadecimal? Converting their sample to decimal gives a 7 digit number. I would think the OP is looking for 3, 3-digit numbers. I guess I'm not clear on how that is stored in a bitmap and how it translates, if that's what it is...

 

[Teck]

Here you go. It may take a while, but you can get RGB values from this wavelength-to-RGB value calculator: https://www.johndcook.com/wavelength_to_RGB.html

I have no idea how accurate this is unfortunately.

Thanks, I will check and compare how accurate it is.

The data is on the graph. All you have to do is read it (with a photo editor) and type it into a spreadsheet/table.

I'm unfamiliar with this. Do you mean it's some metadata in the image file? What (free) photo editor can be used to read it and how do you copy it to a table?

 

[russ_watters]

I'm unfamiliar with this. Do you mean it's some metadata in the image file? What (free) photo editor can be used to read it and how do you copy it to a table?

No, I mean literally read off the screen with your eyes and type into a table with your fingers. E.g., I just opened the file in a 20 year old copy of Paint Shop Pro 4.12, which you should still be able to download from somewhere. But literally any photo editor worthy of the description can do this:

-Hover cursor over a point on the edge of the graph. For example, 540nm
-Read the data from the color pallet on the side of the screen: 0, 255, 118
-Type it into an Excel spreadsheet
-Repeat

Edit:
The cursor wasn't included in the screenshot, but here's what it looks like when I hover over the edge of the 540nm hash, just off the black line:

 

[Teck]

No, I mean literally read off the screen with your eyes and type into a table with your fingers. E.g., I just opened the file in a 20 year old copy of Paint Shop Pro 4.12, which you should still be able to download from somewhere. But literally any photo editor worthy of the description can do this:

-Hover cursor over a point on the edge of the graph. For example, 540nm
-Read the data from the color pallet on the side of the screen: 0, 255, 118
-Type it into an Excel spreadsheet
-Repeat

Oh, you seem to have missed that I asked for a graph that maps every visible wavelength integer to an exact colour code. "Point and click" on a low resolution image is not that. Especially not one that is uneven and squashed together at the ends.

 

[russ_watters]

Oh, you seem to have missed that I asked for a graph that maps every visible wavelength integer to an exact colour code. "Point and click" on a low resolution image is not that. Especially not one that is uneven and squashed together at the ends.

The smooshing together at the ends is a feature, not a bug; our color vision doesn't cut off cleanly and so the palette makers don't bother to assign as many values. And looking at the values, right up to 640nm you can do better than 1nm resolution. E.g, at 620 through the end of the scale, the red and green are 255 and 0. Blue is 53 at 620nm and 75 at 640; 22 steps for 20nm and you can interpolate or pick the nearest for each. The scale ends at a blue value of 83, so that's all the rest of the red end of our vision in only 7 steps of rgb values.

In most of the rest of the graph however you can do half nanometers per rgb value if you want to get that fine.

I'd be curious about why you want this level of precision too: again, the accuracy of this data is much lower than the precision you are going for, so there probably isn't much to gain from single digit nm vs, say, 5 nm precision.

 

[russ_watters]

To reiterate the accuracy problem, there are several issues:

1. Different people perceive and/or sense them differently.
2. Different monitors display them differently
3. Different color spaces are mapped differently.

That last one I'm not sure was made clear. The image I gave was one of many options:

https://en.m.wikipedia.org/wiki/Color_space#/media/File:CIE1931xy_gamut_comparison.svg

Each will give different rgb values for the same wavelength and most even cut off sooner, eliminating the smooshing problem at the ends by stopping sooner.

 

[russ_watters]

Not to belabor the accuracy problem, but...

I was wracking my brain to figure out why the color "space" was shown as a plane instead of, you know, a space. I figured it out: the color space is shown at full/flat intensity, so it doesn't represent our perception at varying brightness. This is a problem, because our perception changes completely with light intensity...and intensity sensitivity is different for different colors...and, of course, will vary between people. Dropping this spectrum (which includes that attenuation) into Photoshop:

...gives me a green value of 18 instead of 53 because the green sensitivity is attenuating. Obviously the impact of the attenuation is most pronounced at the ends of the spectrum. So again, even though I can easily get per nanometer values from this picture, I don't think it's a very useful exercise (even though I'm not sure why you are doing it...) because:

1. The values will vary from one person to another.
2. The values will vary from one monitor to another.
3. The values will vary from one color space to another.
4. The values will vary with light intensity.

By the way; the above graph may be easier to pull the data from if you think it's valid data. You could pull every 10 nm and use a spreadsheet to linearly interpolate down to 1. You should be able to do that in 10 minutes instead of an hour for trying to read every 1 directly from the previous graph.

 

[pinball1970]

I'm trying to find a high resolution image that shows the visible electromagnetic spectrum with a fine graded scale. It should be detailed enough to pinpoint which exact colour corresponds to a particulate wavelength (integer in nanometer) of light. I find a lot of images through searches but no one that is detailed.

For example, if we say 503 nm. I want to be able to find which exact colour corresponds to that wavelength.

The importance here is the accuracy of the colour (and I know different monitors can display colours differently, but let's not get into that :D )

Please link such an image.

All the guys have gone through the physics, electronics.
Colour is very subjective because every eye, retina, distribution of photo receptor cells, type of cells, proteins are different. Even one eye to the other in the same head!
Also the eye does not 'see' colour the brain does, how different is every brain?
Ten colourists all stood round the same calibrated colour screen will see the same red differently and will never be able to articulate anything about it that has any sort of meaning.
One could agree on spectral values from 400-700nm, the numbers but what would that mean?
It would be like reading a recipe for pea soup then guessing what it tastes like.