January 2017

16 171819202122

Style Credit

Expand Cut Tags

No cut tags
Monday, August 20th, 2012 10:13 am
I was thinking about color this morning (yes, I do these things), and I was struck by a disturbing thought: when did I last see violet? Not purple, which I see all the time, but actual violet. [I'm not the first person to wonder this: here's a very thorough discussion.]

Purple, as you no doubt recall, is a compound color: it's what we perceive when we see a mixture of red and blue light. But violet is a pure color: light with wavelength somewhere a little over 400nm. (Side question: anyone have a clue as to why purple and violet are perceptually similar?) So what's the problem? RGB monitors, that's what. The shortest wavelength of light produced by an RGB monitor is blue, which is probably around 460nm or so. That means that your monitor is incapable of producing a violet color. Looking at a picture of a rainbow on your computer screen is inevitably a less vibrant experience than seeing one in person.

So, fine, look at a printed photograph. Well... not so fast. I don't know how all the different types of photo printing work, but a lot of printers are RGB or CMY themselves (and I think that CMY has the same problems as RGB in this regard, or worse). I have no idea what the process is (or was) for traditional photo printing from the analog era, but I'm willing to imagine that there are photo printers today who are capable of printing with actual violet dyes. But wait: what sort of camera took the photograph? Again, I don't know how good traditional film cameras were at capturing violet, but today's digital cameras are (as far as I know) also RGB.

So: when did you last see violet? Think back: what does it look like?
Monday, August 20th, 2012 02:29 pm (UTC)
seems you are not the only one thinking about this:
Monday, August 20th, 2012 08:13 pm (UTC)
I just found in this link a discussion on how the red receptors have a secondary sensitivity a little above 400 nm. This means the wavelengths in the violet region excite both the blue and the red receptors, just like purple. So visually it should be possible to construct a purple mix which is INDISTINGUISHABLE from actual violet, awesome.
Monday, August 20th, 2012 08:37 pm (UTC)
Oh, interesting: I hadn't spotted a clear discussion of the red cones' secondary peak in the article that you (and I, actually) linked to earlier, but the discussion in this article gives a nice clear explanation (and a good graph, too). That explains why violet and purple look similar (and why we perceive a color circle rather than a color line)... I wonder to what degree it really is possible to trigger the visual system to perceive violet (or, more broadly, what hues out there we just don't see).

Also, unless digital camera sensors mimic that secondary peak (do they??), we'll still wind up missing out on violet in most photographs these days. (As [livejournal.com profile] kirinn's photo would seem to show.)
Monday, August 20th, 2012 09:04 pm (UTC)
had you found that link first, sorry, was in a rush but you had peaked my interest and that was what I found...

I am having trouble finding a good reference for this but an interesting coincidence from what I have seen before is the depth of color penetration in water. Red penetrates the least, blue the most. The coincidence is that from some graph a I remember seeing it made it lok like the color wheel was real... the penetration of orange was ~average(red, yellow) and the penetration of green was ~average(blue, yellow) which makes sense as a near linearization of the response curve. The coincidence is that because penetration depth drops off for shorter wavelengths then blue the penetration depth of violet is ~average(blue,red). The last one may have been less close then the others, but it was close enough that I simply dismissed it as the expect behavior given violet is blue plus red, false premise, strangely accurate result...

Complete speculation, but it occurs to me to wonder if this symmetric characteristic of the penetration depth has any relationship to why we would have the secondary red perception bump, causing violet to be perceived as purple. I have no idea what the causal chain would be, but if violet light behaves under enough natural conditions in a fashion similar to red mixed with blue it would then fit for this otherwise odd feature.
Monday, August 20th, 2012 02:35 pm (UTC)
I'm doing pretty well by this measure, since I happened to get to see a really bright rainbow in person just a few weeks ago.

But presumably you could also reset your clock with a decent prism...

Hmm, I wonder if true-violet LEDs are commercially available? Can you get a cellophane filter that's true violet pass-through? Are most purplish flowers violet or mixtures? I'd bet there's at least some that reflect true violet strongly, given there are some that reflect ultraviolet patterns...
Monday, August 20th, 2012 02:40 pm (UTC)
...and after reading some of the linked article, I suddenly realized I have *experimental evidence* of true-violet flowers. There's a certain kind of Iris I've found growing around here several times that looks deep purple (err, violet) to the eye but inevitably comes out deep blue on my iPhone camera. Bingo!
Monday, August 20th, 2012 03:32 pm (UTC)
Oh, awesome! Or disappointing, depending on how you look at it. (The whole time I've been thinking about this, I've had to keep fighting back the urge to say, "I'll just find a picture with Google..." or, just now, "Send me a picture of what they really look like!")
Monday, August 20th, 2012 05:58 pm (UTC)
Well, I can upload the picture of what they *don't* really look like. The one below is definitely a deep violet in person. And meanwhile, I noticed I also have several pictures of flowers on my phone that *do* look purple, meaning that unlike the Iris they're reflecting red wavelengths.

false-blue Iris
Monday, August 20th, 2012 07:52 pm (UTC)
That's really striking and beautiful even in this picture, but yeah, I'd love to see what it really looks like.
Tuesday, August 21st, 2012 11:51 pm (UTC)
Clock reset to: tonight! Awfully good rainbow weather around here lately. Not sure I've ever seen so many bright ones outside the tropics.
Tuesday, August 21st, 2012 01:31 am (UTC)
This image from wikipedia illustrates the problem from the perspective of gamuts:


Here, the inside of the triangle is what is representable by RGB space. Needless to say, the colors outside the triangle are simulated.
Tuesday, August 21st, 2012 01:32 pm (UTC)
That's a really nice illustration: thanks! (To be honest, I'm surprised that it gets as far down into the violet/purple corner as it does. But come to think of it, I realize that I don't know nearly enough about color theory to really grok the details anyway: what are the axes, and how are they scaled? I should probably avoid spending the time it would take to figure all this out right now. :) )