0
\$\begingroup\$

I like this picture of the Pacific Ocean, taken from the ISS.

It seems as if I can see stars in the background, but similar dots are visible in the ocean. I presume some kind of artifacts? What am I seeing there, is it a result of jpeg compression, or something else? Can I distinguish between the stars and these dots, or are there no stars at all?

I have no technical background knowledge for these questions. BTW the picture has its exif data still attached.

\$\endgroup\$
7
  • \$\begingroup\$ Stars don't show as red/green/blue dots in a picture. This seems to be just noise from the camera's sensor. Some of the most faint, light grey dots might be stars... Not sure. Looking at the same image EXIF on Flickr, it was apparently shot at 1/640s, ISO 400 on a Nikon D4. So that looks like an awful lot of noise for these settings. You might want to ask the photographer on Flickr. \$\endgroup\$
    – MrUpsidown
    Jun 30, 2022 at 14:00
  • \$\begingroup\$ @MrUpsidown Noise is always present. But usually when we shoot at ISO 400, f/16, 1/640 we're in an environment where the entire frame is very bright and the noise is masked by all of the light collected by the camera. Think about it. Those exposure settings are two-thirds of a stop faster than the "Sunny 16" rule of thumb! The only time one would normally use them is for shots at a beach with very bright, white sand or on a snow covered landscape on a bright, sunny day. \$\endgroup\$
    – Michael C
    Jun 30, 2022 at 18:02
  • \$\begingroup\$ You can tell these are not stars because the color of stars obey some rules. In particular you can't have pink, green or deep blue stars. \$\endgroup\$
    – xenoid
    Jul 1, 2022 at 0:12
  • \$\begingroup\$ @MichaelC Yes, I know it is always present. It still shows a lot of noise for these settings/camera, which is exactly what you explained in your answer. Did I not get the meaning of your comment? \$\endgroup\$
    – MrUpsidown
    Jul 1, 2022 at 7:09
  • \$\begingroup\$ @xenoid Without the Earth's atmosphere in the way, you can see the colors ofstars with the naked eye. "Magenta→Red→Orange" stars are quite common, Green less so. "Blue" stars usually still look almost white. Cameras with CFA's have a harder time getting colors in stars, but it can happen for the few stars at a given exposure level that is "just right" for that magnitude. Astro cameras that take separate images filtered with various colors can capture even more stars with colors, but are still limited to narrow range of magnitude with brighter stars still blowing out. \$\endgroup\$
    – Michael C
    Jul 1, 2022 at 11:13

1 Answer 1

6
\$\begingroup\$

Those aren't stars, those are hot pixels and "shot" noise. The reason you see more against the backdrop of space than against the Earth is because the Earth is so bright it drowns out most of the noise in that part of the picture.

The image was shot at ISO 400, f/16, 1/640 seconds. That's basically the same exposure settings to use on a white sand beach or a snow covered ski slope in direct bright sunlight! It's slightly less exposure than the so-called "sunny 16" rule-of-thumb which says that when your subject is in direct mod-day sun at f/16 you should use a shutter time that is the reciprocal of the film speed (film) or ISO setting (digital).

Stars aren't bright enough to break through the noise floor at such exposures, even without the Earth's atmosphere in the way. The colored dots you see are image noise.

Image noise has two main contributors:

  • Read noise. This is electrical noise caused by the camera's operation. Though modern cameras have become very good at suppressing most read noise, one would need to cool the sensor down to absolute zero (0K or -273°C) to theoretically totally eliminate all read noise.
  • Image noise caused by the random nature of light itself will vary from frame to frame. This is often called "shot" noise or "Poisson distribution" noise after French mathematician Siméon Denis Poisson, who developed a discrete probability model that mathematically expresses the phenomenon observed in nature. How much it will affect an image will depend upon the strength of the light falling upon the sensor, how long that light is allowed to fall on the sensor, and the size and quantum efficiency of the sensor's discrete photodetectors a/k/a pixel wells, photosites, or sensels.

In the case of the Nikon D4 aboard the ISS that took this shot, there appears to be a little more read noise than would normally be expected for that camera. The D4 was introduced in 2012 and discontinued in 2014 when it was replaced as Nikon's "flagship" stills camera by the D4S in early 2014. We can guess that the camera has been aboard the ISS the better part of a decade. Outside of the protection of Earth's atmosphere it is exposed to higher levels of radiation, particularly gamma rays, than a terrestrial based camera would be. Gamma rays are known to be a cause of image sensor degradation that results in increasing numbers of noisy sensels with cumulative exposure to gamma rays and other types of short wavelength/high frequency EMR.

The Reason You Don't See Stars

If you expose bright enough to see even the brightest stars in the sky, even the dimmer Moon, which is a little more than one stop dimmer than the "Sunny 16" rule (We sometimes call it the "Looney 11") will be completely blown out. The only two differences in the following two photos that contain the Moon and much dimmer planet Jupiter is the amount of exposure and that one image is rotated about ninety degrees compared to the other.

enter image description here

enter image description here

To get even the very brightest stars (and four of Jupiter's Moon that were also within the field of view in the first image) to show up, the moon is so overexposed that light from it is bouncing around inside the lens a reflecting off the front of the sensor stack and the front of some lens elements and then off the back of other lens elements. These same reflections were also bouncing around in the lens in the first photo, but they are so dim as to not be visible at all because the first photo was exposed 12 stops dimmer than the second image, or with 1/4,096 as much light allowed into the camera.

Also note that the exposure was long enough that the stars trailed past the camera mounted to a stationary tripod.

For a couple of intermediate exposure levels between the two extremes above which illustrate that details of Jupiter's bands and Jupiter's moons can not both be properly exposed at the same time, please see this answer to a different question.

\$\endgroup\$
3
  • \$\begingroup\$ I was afraid of that. Thanks for the explanation. \$\endgroup\$
    – hurdsean
    Jul 1, 2022 at 9:45
  • \$\begingroup\$ Couldn't cosmic-ray radiation during the time of exposure also contribute to noise? \$\endgroup\$ Jul 1, 2022 at 11:51
  • \$\begingroup\$ @AndréasSundström Perhaps, but I've never seen any scientific articles about it. I have seen several that show that gamma rays can degrade CMOS sensors over time, even when the cameras are terrestrially based. Of course cosmic rays are deflected by magnetic fields, while gamma rays are not (though the result of cosmic rays colliding with certain things can produce gamma rays). \$\endgroup\$
    – Michael C
    Jul 1, 2022 at 20:55

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.