I’ve seen some remarks that sometimes a decent looking high CRI emitter during the day can look yellow/green at night. I notice this as well and suspect it has to do with how the luminosity function (perceived relative brightness of wavelength) changes from day (photopic) to dark adapted (scotopic) vision.
Here’s one way I can think of testing this conjecture. Let S(λ) denote the scotopic luminosity function, and P(λ) the photopic. If I understand correctly, the spectrum the human eye perceives is simply the actual spectrum weighed multiplicatively by an appropriate luminosity function.
Let E(λ) denote the spectrum of an emitter; we try to answer the question: what spectrum F(λ) under photopic vision would look the same as E(λ) under scotopic vision? We can then see whether F is greener compared to E. We could set up E(λ)S(λ)=F(λ)P(λ), and solve for F(λ)=E(λ)[S(λ)/P(λ)]. Now we just need to find out whether the ratio S(λ)/P(λ) favors green wavelengths over red/blue.
Does someone have a reliable source on the plots/values of the scotopic and photopic luminosity functions? If there are closed-form expressions that give good approximations to these functions, it would be even better. Thank you for reading!
EDIT: as jon_slider pointed out, I meant mesopic not scotopic!
Sunlight has a slightly higher duv, ie, is slightly greenish. So even a “green” light looks pretty neutral.
At night, if most lighting is incandescent (ie, warmer, redder), and that’s what you’re seeing all around you and are used to, then a greenish light will stand out like a sore green-thumb.
google says color perception is actually photopic, and the cones are involved, not the scotopic rods:
“We have three types of cones: blue, green, and red.”
that does not explain why we both noticed green tint less during the day, and more at night.
It might be that during mesopic vision, the different color receptive cones vary in their sensitivity
for example:
I wonder if the green sensitive cones are more active at lower light levels, and red sensitive cones are more active at higher light intensities?
that would suggest at night we see more green, during the day we see more red and less green?
something is happening… I just don’t know what it is… lol
I have no data / numbers / plots, but might this observation perhaps be related to this? I just happened to be reading up on this a few days ago. Yeah, slow news week:-)
@Lightbringer: when there is warmer/nicer lighting around, you're absolutely right! However, I also notice the shift to green even with no other sources of light besides the flashlight. You raise a very interesting point about sunlight, how it is objectively green but not perceived as such. I suspect part of it is due to the intensity, which puts our eyes squarely in the photopic setting. I do wonder, however, that the sunlight we see is rosier than it should be, since the blue sky contributes a lot of high CCT light, and we know CCT mixing makes things rosier.
@jon_slider: thank you for pointing out my error--I did indeed mean mesopic vision, which still allows for color perception, albeit shifted. I'm hoping to quantify this shift with the ratio of luminosity functions. Your example is exactly what I'm trying to capture.
@Desertcat: Absolutely!! I somehow forgot to mention the Purkinje effect. Perhaps a better phrasing of my suspicion is whether seeing more green is also a side effect of the Purkinje effect.
Off topic but does your name refer to quadratic reciprocity? I use qreciprocity in some other platforms. It’s the first time I’ve seen someone else use it as well.
If you’re into photography, you can find/take pix which show sunlit areas being “rosy”, whereas areas in shadow are distinctly blue. I openly mock pictures of snowscapes where the person can’t be bothered to tweak the white-balance on the camera, and you end up with “blue snow”. But some people who hate that blue also try to “correct” those sunlit/shadowed pix to try to make the in-shadow areas less blue, which gives the rest of the shot a rather eerie coloring.
One other thing to consider is the LED itself. Many many LEDs “go green” when driven lightly, then turn rosier at increasing currents. Even the ’351 and some ’219s are guilty of that. Moonlight modes tend to be quite green, and when driven harder take on a quite nice color.
So… yeah, even with no other light than the flashlight, you can get a green tinge at moonlight/firefly levels.
Oh, that also usually applies to “better” drivers that are current-regulated and not PWMed. Ironically, PWMing, which most people hate because of flicker, is a series of brief but high-current pulses which keep the LED “driven hard” in those brief periods and thus do not go green.
Here are some interesting videos about color vision
though it does not specifically answer why I see more green at night than during the day, they may help promote a better understanding.
the links are timeslots I thought particularly relevant, but the entire series is worth watching imo
agonist and antagonist sensing, may be part of the puzzle:
Opponent colors seems to be part of the puzzle too:
fwiw, the output has a rather small effect on Tint DUV. I do not think output differences day and night account for the difference in perception of green being greater at night. The difference between 519a and sw45k is much larger than any tint difference based on lumens level.
When I compare the day and night Apparent Tint, Im using the same 20 lumens level in both cases, on both lights… so Tint variation by output is not a relevant variable.
IMO the relevant variable is the ambient Tint adaptation of the Observers brain.
iow, the tint of the environment during the day, is greener than at night. So during the day, green perception is down regulated (inhibited or antagonized) by the brain. And at night, green perception is UP regulated (less inhibited)
===
in any case here are some Tint variations by output:
Opple measured:
519a at 1 lumen DUV –0.0007
519a at 20 lumens DUV –0.0005
519a at 100 lumens DUV 0.0000
==
sw45k at 1 lumen DUV –0.0092
sw45k at 20 lumen DUV –0.0131
sw45k at 100 lumen DUV –0.0137
===
both lights at 20 lumens, obviously the 519a is greener:
Current, temperature make a difference and even drive mode (current or PWM)
There is a difference even between CRI ratings, quite interesting suggestions since in practice it may vary.
3000K R8000 vs 3000K R9080 forward current and junction temperature comparison GIF
5000K R70 vs 5000K R8000 forward current and junction temperature comparison GIF
.
