0
\$\begingroup\$

I was trying to get a shot of a red moon, but the color was coming out wrong obviously due to a white balance problem.

Adjusting the white balance after the fact was not an option because I would have nothing to compare to and its not like I am going to remember the exact shade of red.

This same problem, I imagine, affects astrophotographers. For example, stars like Betelgeuse have a red color. How do they get the color correct? We can imagine the same problem for any light source at night. For example, neon lights, reflected light from the moon or stars, fireflies, aquatic bioluminescence, etc.

Spectrometers are supposed to measure colors, but the spectrometer I have only measures incident ambient light. It will not measure light coming from a specific point in the field of view. I guess one strategy would be to use a telescope on the object, then hold the spectrometer up to the eyepiece. Seems like a hack.

Is there any solution to this problem?


Response to the "color is subjective" non-answers: devices are sold which allows hardware stores to match the paint so when people repaint their walls the color matches with the old color. So, the argument that you can't objectively match colors is BS.

\$\endgroup\$
7
  • 5
    \$\begingroup\$ Define "correct". \$\endgroup\$
    – user29608
    Commented Aug 31, 2017 at 7:02
  • 1
    \$\begingroup\$ @fkraiem The wavelength of the light coming from the moon is the same as the wavelength of the light coming from the liveview/EVF/monitor/printer representation of the same object. \$\endgroup\$ Commented Aug 31, 2017 at 7:05
  • 2
    \$\begingroup\$ That's not how RGB displays work. There are only three wavelengths (or to be more precise three narrow bands of wavelengths) output by most display devices in various combinations. \$\endgroup\$
    – Michael C
    Commented Aug 31, 2017 at 7:35
  • 2
    \$\begingroup\$ Color do not necessarily correspond to a single wavelength and some colors cannot be reproduced by a single wavelength - brown and purple. This article on color may be useful. \$\endgroup\$ Commented Aug 31, 2017 at 8:10
  • 1
    \$\begingroup\$ The light from the moon is reflected sunlight and is relatively full spectrum. Even when it is near the horizon and more of the blue is filtered by the atmosphere than the amber it is still relatively full spectrum, just balanced towards the orange/amber end of the spectrum, much like an incandescent light bulb. \$\endgroup\$
    – Michael C
    Commented Aug 31, 2017 at 9:55

5 Answers 5

4
\$\begingroup\$

The problem with colour is that it is always a subjective perception. Look at a sheet of paper in daylight and in a room under incandecent light, it will seem white to you both times. But if you take a photo of it with a fixed WB set in the camera, they will appear very different.

Auto WB in cameras works (at a first approximation) with the assumption that the scene is on average a neutral grey. That's ok for the usual subjects, but when shooting the night sky this fails horribly.

The recommended way to treat the photos then is to set the WB so that the result matches your memory/expectation of the scene. Since your perception of the red moon was a highly subjective experience anyway, there is no objectively "correct" value that would produce anything which looks like you remember it.

Regarding WB in astrophotography, see here. Basically it says to either get a pleasing image, or to calibrate based on known objects (a galaxy can be considered to be white on average).

\$\endgroup\$
4
\$\begingroup\$

The simplest solution I've found is to shoot raw with daylight white balance, rather than auto. (Raw, so you can try other settings later if you don't like the daylight result).

After all, the moon is basically a gray rock in bright sunlight :).

If you're doing astrophotography, I'd usually use the daylight setting. If you've got a lot of light pollution, some folk recommend using the tungsten setting to tone down the light pollution background a bit (or use a light pollution reduction filter, but with the move away from yellow sodium streetlights to broad spectrum LEDs, they may not be as effective as they used to be, depending on your local light pollution mix.

For terrestrial scenes, it depends on the lighting - daylight white balance gives a warmer look to tungsten lights, the same way shooting daylight balanced slide film did. If you don't want that, then either use the appropriate preset for the predominant lighting - in which case whites should come out close to neutral - or try auto WB (or a custom WB from a white/neutral card).

But if you're taking shots of strongly coloured subjects that don't average out to gray - like neon signs - then I'd go with one of the presets. And shoot raw, so you can try other settings later.

\$\endgroup\$
3
  • \$\begingroup\$ The camera was set to daylight and it was way off. \$\endgroup\$ Commented Aug 31, 2017 at 12:52
  • 6
    \$\begingroup\$ Maybe the camera was not way off. Maybe your perception of the moon at the time was. \$\endgroup\$
    – Michael C
    Commented Aug 31, 2017 at 16:33
  • \$\begingroup\$ I agree with @JerryTheC. The moon is gray, and the correct exposure is the same as anything else that's in direct sunlight (sunny f16 rule). If you set your white balance to daylight/5500K the color will be correct, assuming your camera is working properly. \$\endgroup\$
    – BobT
    Commented Aug 31, 2017 at 23:11
2
\$\begingroup\$

If the camera was effectively in daylight white balance at the time the shots where taken, then you can deduce the color shift that happened, and apply it's reverse to the image. Specially if you have the raw files. JPEG would work too, to a lesser extent.

For simplicity lets say that every sensor has it's own deviation from "correct" color, due mainly to design, materials and manufacture. The camera software somewhat applies a correction profile to reproduce color as best as possible. It does that in any WB setting, using different "shift values" for each light source. In principle, "daylight" setting assumes "white" light (uniform intensity for all light frequencies across visible spectrum) is illuminating the scene, so it should not shift color in any way, except to compensate design limitations in the sensor. That shift would be, in any case, the best compromise in which the designers settled for you particular make and model (and firmware) of camera.

My idea is that if you take another picture with the very same settings (particularly white balance), same lens, as your moon shots and then evaluate it straight out of the camera, using a correctly calibrated monitor, you should be able to work out what changes the image needs to better represent the actual scene. These changes should, theoretically, render a good result when applied to the images that you need to correct (the moon shots).

By changes I mean, for example, + 200 points in color temperature, 7 points towards green in green/magenta, etc.

An editor like Lightroom or Adobe Camera Raw is very helpful for this as it allows you to copy and paste the color settings of the sample shot to the moon shots. In other photo editors I take note of the values in any color/contrast/hue/saturation, curves/levels/etc (Not very efficient).

A crude calibration of your camera

To further know about the color shifts that your camera may be applying, even in "daylight" white balance, I suggest the exercise of photographing different lightsources against a black non reflective (or very distant) background. They may be light bulbs (incandescent, fluorescent and/or LED). Nothing fancy, use what you have at home, even some car headlamps may do.

Take direct shots at these sources while minimizing the influence of any others. Evaluate the shots in your editor+monitor and work out whether your camera shifts towards red, green or blue. The differences may be very subtle, but hopefully you'll learn how your camera behaves. Generally, when you underexpose a direct shot of a light source, it's color cast becomes more apparent, almost like if you augmented the saturation in your editor.

Also, check if you had any special profiles in use. For example my camera has "Natural", "Vivid", "Muted" profiles. They affect the color of the image. When I import into Lightroom, by default it applies "Adobe Standard" profile. None of the profiles renders the same result as other. Most cameras have such profiles or color modes, with different names and they are often customizable and may have color shifting settings.

Environmental variables

Anything that could have been on the atmosphere between your camera and the moon can affect the perceived color of it. Besides the well known red shift that occurs near the horizon, water vapour, suspended particles, etc. may also distort the color of an image. Light pollution is in part, light from cities bouncing, reflecting, and refracting off those vapors and particles.

A couple of anecdotes on perception

However, you will be limited by your own perception at the time of capture, and your own memory. Consider the following anecdotes.

Night vision/low light sensitivity of the eye: I had been on a beach, at night for at least 4 hours, in a remote location very far from any artificial light sources. My eyes' got so adapted to darkness that kind of had my eyes in "black and white mode" (Read on cones and rods related to eye light sensitivity for information on how this happens). My surprise was when I downloaded my captures to the computer an all of them had color. I could not recall any color in the scene, I had to trust the camera's output.

Color sensitivity of the eye: I have an issue with streetlights. In my country, there are two commonly pressurized gas lamps in the streets. In photographs they show as either orange or green. I assure you, that I see only orange or blue lights, my eyes do not register them as green. I do not know why that happens but it seems to affect me only with street lights...

Consider also that our ability to recall exact colors, tonalities, light intensities from a scene is limited. Even our emotional response to the scene may cause our memory to recall a more/less colorful scene than what was recorded. And, if, at the moment we thought, for example, "this photo is less colorful than the real object", over time we recall that phrase or idea, and may end up cranking the saturation too much up, more than necessary.

\$\endgroup\$
0
\$\begingroup\$

I have been working on this problem some more and I think one possible solution may be to make sort of a color checker for a specific hue range (sunset red in this case).

The idea is to make a color checker with a hole in the center of each swatch. Then simply hold the swatch over the live view and see which swatch most closely matches the image color. So this will reveal the color wanted. Then I just need a way to determine what color is actually being produced and figure how to set the white balance so that the color shift is matched to that difference.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ The lighting difference between the emitted light and the reflective swatch is probably going to prevent this from working well. (Note I didn't DV, but that's probably why someone did.) \$\endgroup\$
    – AJ Henderson
    Commented Aug 31, 2017 at 15:50
  • 2
    \$\begingroup\$ Yep. If the light illuminating the swatch is different, the same swatch will look different while the moon on the horizon won't shift in the same way. @TylerDurden Look up the meaning of 'metamerism' or 'metameric failure'. \$\endgroup\$
    – Michael C
    Commented Aug 31, 2017 at 16:26
  • 1
    \$\begingroup\$ Also, comparing a swatch to the camera's LCD involves the added complication of having to know/calibrate the LCD's color profile, which might not be able to be calibrated correctly. The swatch, being reflected light, will look different than the LCD's band-limited RGB emission. Especially depending on the ambient light source that's illuminating the swatch. \$\endgroup\$
    – scottbb
    Commented Aug 31, 2017 at 19:22
  • \$\begingroup\$ @scottbb I would expect the live view to already be factory calibrated. Its not like a monitor that is just a generic device made by a random manufacturer. The live view screen is custom hardware, custom designed for the camera. So as long as live boost is off, I think the live view is supposed to represent the JPG exposure as closely as possible. \$\endgroup\$ Commented Aug 31, 2017 at 20:03
  • 3
    \$\begingroup\$ @TylerDurden I'm not sure if your expectation pans out or not. We have some relevant questions here: Why don't you need to calibrate a camera's LCD screen?. Also, Are the color and brightness of a camera's LCD accurate? \$\endgroup\$
    – scottbb
    Commented Aug 31, 2017 at 20:13
0
\$\begingroup\$

For night shots without flash, Use iso 400 for better print quality and try to manage exposure using shutter speed.

You will need to use shutter speed below 20 so that you can take nice shots at night

\$\endgroup\$
1
  • \$\begingroup\$ How does this answer help the OP deal with color balance issues at night (especially in this particular instance of shooting the moon)? Your answer addresses, in only the most generic and rule-of-thumb way, exposure of night shots, with no regard to the subject of the night image. \$\endgroup\$
    – scottbb
    Commented Sep 3, 2017 at 17:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.