That depends what you mean by “appear”.
Our brains are very good at correcting for illumination source.
Consider this image taken from Breaking Bad’s famous “Mexico filter”. There’s a very strong yellow filter over the whole image, but our brains can correct for that and recognize the car in the near view as white.
Indeed it is more natural to see it as white, it takes effort to recognize that that really is an inference.
Precisely. And this is, indeed, what you see in late afternoon/early evening when the light from the sun is passing through so much atmosphere that the sun really does appear dimmer and yellow. You can see that normally white objects look distinctly yellow. It’s what Lewis Carroll wrote about as a “golden afternoon” in his Alice in Wonderland books.
Because, as I have said, light from the sun when close to the zenith is our very definition of “white”
Here’s an article with a chart showing the relative red/blueness of various light sources.
What is Natural Lighting – Full Spectrum Solutions, Inc.
Basically noonday sun is bluish- 5000k-5500k like @Johnny_L.A says. But it varies; sunrise and sunset are considerably more yellow/red, so maybe people associate that with “sunlight”.
The fundamental problem is that our eyes are very good at doing what digital cameras calll “white balance”, meaning that regardless of the light source, we’re good at identifying things that are white and adjusting. So at night inside with incandescent bulbs, it doesn’t look overwhelmingly orange. Nor does it look bluish outdoors during the day. Nor do fluorescent bulbs look weird colors like green or magenta.
But if you’ve ever shot pictures using photographic film, you’ll recall that most film is daylight-balance film, meaning that it’s adjusted at the factory for that 5000-5500k color temperature of daylight. It also means that shots by candle light or incandescent light have a VERY orange cast to them, and fluorescent lights give weird green/magenta casts, because the film is not adjusting to the light source like your eyes would.
Basically the various sorts of LED lights and CFLs have color temperatures, and the warmer, redder ones are usually “soft white” of some stripe, and the bluer ones are “daylight”.
Daytime sun isn’t “bluish”. That chart, setting the apparent colors on a spectrum, isn’t really a good way to present it. A CIE Chromaticity diagram is better. Noontime sun and other “white” sources are clustered near the center of the horseshoe of the Spectral Locus
The way to determine the approximate spectral color best represented by any point within the CIE diagram is to shoot a line from the white point outwards to the closest point on the spectral loocaus – which is pretty much irrelevant if your color is at or very close to your decided white point; it might as well fall anywhere on the line.
Either way, it’s definitely not “yellow”.
Yes, it’s bluish compared to yellow. I can’t find a cite, but I could swear daylight films for Kodak were balanced to around 5700K, which was noonday sun over some place in Colorado, but daylight is usually regarded as the range of somewhere around 5000-5700K.
“White” is also “bluish compared to yellow”
That was the point I was trying to make.
Why Colorado? I would have just assumed noonday sun in Rochester.
Remember that I’m going on a twenty or thirty-year old memory. Rochester makes a hell of a lot more sense (though Kodak did have presence in Colorado), so maybe it was Rochester. I would not be surprised if I’m completely wrong. I can’t seem to find a cite.
I just watched a video about a balloon that was sent to the edge of the atmosphere. The sun was yellow as the balloon was released and turned white as it rose.
Yeah, I’m digging and I’m not finding anything. Perhaps I am misremembering something. Most of Kodak’s literature says 5500K, which is the usual number given for daylight, but I could swear I read a number a little bit higher, like 5600-5700K for at least some of their films, and it was tied to noontime sun at some place in the US, but I can’t dig up anything of the sort Googling.
Probably doesn’t matter much in any event.
I just imagine it’s one of those things where people assume the huge multi-billion dollar company spends tens of thousands of dollars measuring the sunlight color temperature at some specific point in the Rockies for all sorts of highly technical reasons, when the reality is far less prosaic, and the engineers doing the actual calibration just went out in the Kodak parking lot at the start of their lunch hour on some sunny day.
The obvious place to balance at least some film stock would have been outdoor Hollywood. But the reality was that Hollywood demanded and used tungsten balanced film, since studios indoors needed it. Despite Hollywood as a movie making location being initially chosen for its constant good outdoor light.
But if you want a “scientific” standard to start from, you would at least want a representative pristine atmosphere, so avoiding industrial or highly populated areas would be a good start. Somewhere with horizon to horizon blue skies a reasonable amount of time would be good too. That would tend to rule out Rochester.
You really want somewhere with the bluest of blue skies to get the baseline of the atmospheric effects. A problem is that the apparent colour temperature of sunlight changes through the day in part because the contribution of the blue sky changes as well. The effective illuminant becomes even more blue, and at noon, pretty nasty.
Film photographers especially will remember use of skylight filters - 1A and 1B, with a subtle magenta tint, used depending upon the amount of sky-blue cut you needed. Here in Oz you never went outside without one. Film was never balanced for full sky blue, but rather the more common higher latitude skies.
My memory is a bit dim here, but I though that the basic sunlight definition explicitly elided sky illumination, and took just the direct light from the sun to define it.
A flat energy spectrum (so avoiding the pedantically impossible equal energy at every wavelength definition, and replacing it with equal energy in each 1nm band) gets you very close to a perceived colour of 5455K. Clearly it can’t actually be a black body. In the CIE colour space, it is defined to get you exactly (\frac 1 3, \frac 1 3). It is defined in the CIE world as Illuminant E.
Those are the same definition, and thus have the same problem.
Its a limited total bandwidth. So is there still a problem?
That depends on whether you’re limiting the wavelength bandwidth or the frequency bandwidth (with the latter being the more common definition of the term). Equal energy in each 1 nm band, and limiting the frequency bandwidth, still gives the same problem, because 0 to 1 Hz is just as much bandwidth as 1000000 to 1000001 Hz, but 0 to 1 Hz contains an infinite number of 1 nm bands.
You can still define a spectrum that’s flat-by-wavelength or flat-by-frequency if you put boundaries on both ends, which isn’t too big a deal because realistically, there’s always going to be some limits to the portion of the spectrum you’re interested in. But it does make it clear that “Just have equal energy everywhere!” isn’t nearly as obvious a definition as it looks.
