5
\$\begingroup\$

Just as in camera menu, there is WB presets available in RAW processing. Those include daylight, cloudy, shade, flash, fluorescent, etc. And of course Auto and custom settings. When processing my photos I like to use a WB pick-up tool, but what would I pick from a photo of pitch black sky pinholed with bright stars only? I would not be asking, if I had something/anything in the foreground, like a barn or a tree, but there is only stars in the whole photo. I went through those presets, but those either did nothing or made the photo look bad. Then I tried haphazardly some custom settings out of my head, and finally surrendered to use AWB that was already there to begin with.

My question is: Is there a color temperature (zone, if not any exact temp) uniformly found good for star photography, even as a starting point for further adjusting? Or rather: Is adjustments in white balance needed at all when there's only stars showing?

\$\endgroup\$

3 Answers 3

6
\$\begingroup\$

Don't use in-camera white balance. Have the camera produce a raw file, then you take it from there.

You can measure the white balance of your sensor ahead of time, then use that correction for the star image. For something like stars, I'd use sunlight as the white reference. Put another way, sun-like stars will appear white and other stars will have colors relative to that. I have measured my sensor on a white target illuminated by direct sunlight. You can use a gray scale card to get various brightnesses, or do different exposures of the same sunlit white target. Either way you get curves for how each color in your sensor responds to light.

I've done this with several camera sensors and found them all to be quite linear. Given that, you only need to make a single white measurement since the same color balance correction applies to the whole dark to light range.

One thing to watch out for with stars is that they are point light sources and therefore could be focused so small that they hit a small number of sensels, which probably aren't balanced to red/grn/blu content. Put another way, if a star is focused on a single green sensel, then the star will appear green regardless of its actual color. The anti-aliasing filter over your sensor should help somewhat with this, but these filters still let some frequencies that will alias thru.

\$\endgroup\$
2
  • \$\begingroup\$ Oh, never thought I could do it beforehand like you described. Thank you, I will get such a reference image the very next day we have sun shining here. And of course I shoot in RAW and have camera WB set to Auto. Question was about WB settings in post process. \$\endgroup\$ Sep 18, 2013 at 22:24
  • \$\begingroup\$ @EsaP: Yes, digital sensors are much better absolute measurement devices than film. For example, sunlight registers on the sensor in my camera with the relative weights .541, 1.000, .695. I use these values to correct pictures taken in sunlight. There is no need for a separate white balance picture or to guess the white balance in those cases. Of course for other less repeatable lighting conditions, taking a separate white balance picture may still be required. \$\endgroup\$ Sep 19, 2013 at 14:06
2
\$\begingroup\$

Use 5200°K for your color temperature and you should be fine. That makes the light from the mid-day Sun appear white. If you are pointed more towards the horizon, reduce the color temperature value to compensate for the increased filtering effect of the Earth's atmosphere that makes the Sun and Moon appear orange near the horizon.

If you are taking long exposures, some of the weirdly colored "stars' you see in post are actually hot pixels. One way to tell is to periodically take a long exposure that will intentionally blur the stars. The hot pixels will still be points, and will usually be the same pixels from one frame to the next under the same conditions and exposure times. If a pixel gets hot enough, it will even spill over to adjacent pixels through the demosaicing algorithm. This will usually stay in a geometric pattern of other nearby pixels that are filtered with the same color by the Bayer layer. You can use a stamp tool to clone out the hot pixels in your blurry test shot and then copy/past to apply that clone to the other shots to get rid of the hot pixels. The longer you shoot with the sensor energized for long exposure times, the more hot pixels you will see and the brighter they will become.

To illustrate this you can take a series of shots with the lens cap on. Begin with several short shots, say 1 second each. Then move to a few consecutive 5 second shots, 10 second shots, and so on. After you have taken several 30 second shots, immediately take another 5 second shot and compare that one to the 5 second shots you took earlier. When viewing the results, turn off any noise reduction and increase the sharpening.

\$\endgroup\$
4
  • \$\begingroup\$ LENR seems to take hot pixels out pretty well, but if I'll end up doing star stacking photos this advice will surely prove very handy. I'll do your suggested test to satisfy my curiousity and then I think I'll repeat the test with LENR turned On, just to see how it performs. \$\endgroup\$ Sep 19, 2013 at 6:50
  • \$\begingroup\$ The problem with LENR is the required time gap between each image while the dark frame is taken. Sometimes that's not an issue, but sometimes it is. When shooting meteor showers, for instance, you want the shutter open for as high a percentage of the time as possible. Since you are taking consecutive frames for a long time period, the sensor tends to retain the heat almost as if you are taking a single long exposure. Same thing if you are stacking for star trails. LENR leaves a gap between each shot. \$\endgroup\$
    – Michael C
    Sep 19, 2013 at 7:19
  • \$\begingroup\$ It would be nice if Canon would let you take one dark frame at the end to apply to all of the previous shots. You can manually do this by taking a 'cap on' shot if your stacking software lets you load a dark frame. \$\endgroup\$
    – Michael C
    Sep 19, 2013 at 7:21
  • \$\begingroup\$ Yeah, that's why I said it'll become most handy if/when stacking photos, to avoid the gap between shots. \$\endgroup\$ Sep 19, 2013 at 7:29
1
\$\begingroup\$

I am using a Nikon D7000 and have found that using the "Daylight" preset produces a far too orange color cast. I have switched to using a custom temp of about 3300 that works rather well.

\$\endgroup\$
1
  • \$\begingroup\$ Actually, if the sky is orange, that sounds like excessive light pollution from city incandescent and sodium lights. Try finding a darker site much further from the cities. \$\endgroup\$
    – WayneF
    Dec 21, 2017 at 4:29

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.