How do I accurately and objectively capture the color of the setting sun?

Question

I'm taking a photo of the setting sun, and want to capture the color of the sun accurately and objectively.

If we're talking about reflected right, I understand that the camera can't distinguish between yellow light bouncing off a white wall, and white light bouncing off a yellow wall. Therefore, white balance is a subjective game.

But when we're talking about incident light, it should be possible to objectively and accurately measure the color temperature of the sun — the average frequency of its spectrum. Is that reasoning correct? If so, how do I accurately capture the color of the setting sun itself?

The goal is that the color of the sun in the photo should be exactly the same as it was in real life, assuming I'm viewing it on a calibrated monitor. I often find that the brilliant orangish color of the sun is replaced by a more neutral yellow (cooler temperature), destroying the effect I'm trying to capture. The goal is NOT to cancel out the color of the sunlight, say by using a gray card.

Answer 1

But there is nothing objective about perception. If the goal is to attempt to reproduce the perception, the closest will be to set the white balance from a grey card which is not directly lit with the Sun.

Answer 2

As you say, white balance is a subjective game. The only way to do this in anyway objectively would be to process your photos in conditions where all the factors affecting subjectivity, i.e. the colour temperature of the ambient light, is the same as when the photo was shot.

In my Canon 5D Mk III, for example, this could be done as follows:

• Shoot the sunset in RAW
• Choose RAW Image Processing menu item
• Adjust colour temperature in increments of 100K until what you see on the screen reflects what you see in the actual scene.

But then no doubt when you go home and view your photo on your laptop screen in a room light by your energy saving lightbulbs (you environmentally conscious person, you), your brain will tell you that your photos are more orange than they did when viewing them on your camera screen in the field.

This is the point of using white balance. White balance is to make photos "perceptually correct" not "objectively correct". An "objective" approach would dispense with white balance manipulations altogether.

The unfortunate thing about sunsets is that the sun (and even anything lit by the the combination of direct sunlight and skylight) actually appears orange to the eye, so to "correct" for it doesn't make any sense.

I know, it sucks.

Answer 3

When it comes to the sun, objectivity is harder than that. Color of the setting sun is changing while it's descending - and white balance basically means that you choose the light of the sun as white point. It differs minute-to-minute in last stages of sunset, but overall - you should decrease color temperature if you want to set white balance correctly during sunset.

tl;dr Start at 5,780 K before sunset and decrease it to your liking

And remember - after sunset the sky is starting to get blue due to Tyndall effect - so after an hour after sunset your white balance should go up.

Answer 4

I'm a little unclear what you're asking, but if you literally want to measure the color temperature of the light from the setting sun, you can take a photograph of the setting sun in raw mode, making sure not to overexpose the disk of the sun. Then, in your processing software, you can set the white balance by clicking the eyedropper on the disk of the sun. You can then read off the custom color temperature and tint that resulted. (Once you set the white balance this way, the disk of the sun will of course be gray or white.) This approach assumes that your camera has a good, accurate profile in Lightroom or whatever software you're using.

If you're asking whether there are instruments that can measure the color temperature of incident light, there are indeed. For example, the Sekonic Prodigi Color C-500 Color Meter, which works a lot like an incident light meter, but for color temperature.

Answer 5

Interesting to consider what "color temp" or "WB" causes the monitor to show the actual same color.

The same spectra would indeed be the same in a real sense. But we don't have that.

The same tristimulus RGB values should "look" the same, at least to a primitive stage in processing in the eye. But the brain interprets that based on the brain's own WB setting, so it still won't look the same even though it's truly identical in the physical meaning!

That's the whole deal with WB. If you remember when other people developed and printed film for you, and they came out orange if shot indoors, that's the thing. The print is "right" but looking at a print doesn't change the mind's current WB to what it was when you were seeing it live. In fact, the pigments on the page are interpreted using the mind's current WB, so they ideally would cancel out and the print should be made to a standard white point.

You can't capture the complete perception of the sunset. An accurate color space mapping will not evoke as much of the same feeling as a more poetic interpretation would. That's why people still take pictures of the sunset, and why it's "art". You could get a spectrometer and deliver a chart of scientific readings, but that's not what we're after here.

"I'm not worried about it changing in the few seconds between setting the white balance and taking the photo" No, don't "set" the white balance in the camera. You don't care, as the RAW data will be processed later with a more powerful computer and under your guidance. Shoot a test target immediately before, and use that to help you figure things out when you "develop" it. That takes no time or fiddling; it's just another photo.