I was just looking at an image from NASA's APOD project

the antennae

I noticed that the starbursts are directly horizontal and vertical. If I recall correctly, on my DSLR I get one "ray" per blade. So does that mean that either the Hubble telescope or the Subaru NAOJ telescope just has a 4-blade aperture?

If not, what else would cause the burst pattern like that?

  • \$\begingroup\$ For what it's worth, more technical information on this image at astrobin.com/293303 (full of astronomy stuff that I don't understand, so just giving the link...) \$\endgroup\$
    – mattdm
    Commented May 4, 2017 at 12:21

2 Answers 2


What you're seeing isn't the result of an iris aperture like in a camera. The 4-point diffraction spikes in the telescope are caused by the four struts holding the reflector in the mirror telescope. This diagram from the Diffraction spike Wikipedia article shows the diffraction pattern (below) created by the corresponding strut arrangement (above):

Comparison of diffraction spike patterns of various strut arrangements
Comparison of diffraction spike patterns of various strut arrangements by Cmglee, via Wikimedia Commons. CC BY-SA 3.0

The image in your question is a composite of several images and data from the Subaru Telescope in Hawaii, and the Hubble Space Telescope. Interestingly, the Subaru Telescope has a 4-strut arrangement, but they are not 90° apart. However, in this composite image, it is likely that the data for the bright stars came from the Hubble.

Hubble's 4-strut mirror support configuration is famous for generating long, narrow diffraction spikes on bright stars. From the Hubble FAQ:

Why do stars have a cross-shaped distortion in most Hubble images? Why do galaxies not?

The cross shape visible on bright objects (such as stars) in Hubble images is a form of distortion that is visible in all telescopes that use a mirror rather than a lens to focus light rays. The crosses, known as diffraction spikes, are caused by the light’s path being disturbed slightly as it passes by the cross-shaped struts that support the telescope’s secondary mirror.

It is only noticeable for bright objects where a lot of light is concentrated on one spot, such as stars. Darker, more spread-out objects like nebulae or galaxies do not show visible levels of this distortion.

In your question, you said,

If I recall correctly, on my DSLR I get one "ray" per blade.

If by "ray", you mean a single line from the center of the star outwards, then no. You get two per blade. You get horizontally-opposed "rays" from each edge in the aperture.

In the above diagram from Wikipedia, notice that there is no difference in the number of rays between the single strut and the double strut arrangement. Similarly, there is no difference in the number of rays between the 2-strut (ell), 3-strut (tee), and 4-strut arrangements (3rd–5th arrangements): there are 4 rays.

In those cases, because of the presence of edges within the aperture that are 180° opposed, the half of the generated rays are overlaid on each other.

But in the 3-strut ("Y") arrangement on the far right, no struts are in 180° opposition, so you can clearly see the six generated rays, two from each strut.

From the same Wikipedia article, this diagram shows the diffraction spikes created by non-circular bladed iris apertures:

Comparison of diffraction spikes for apertures of different shapes and blade count
Comparison of diffraction spikes for apertures of different shapes and blade count by Cmglee, via Wikimedia Commons. CC BY-SA 3.0

In general, an aperture of N blades will create:

  • N-point stars, if N is even;
  • 2*N-point stars, if N is odd.

This is why 7- and 9-bladed aperture DLSR lenses create beautiful 14- and 18-point sun spikes at small apertures.

  • 3
    \$\begingroup\$ Fantastic answer, and exactly what I was looking for. I couldn't find any of the sunburst pictures that I took, and I knew that one point per blade didn't make sense. Great explanation \$\endgroup\$ Commented May 4, 2017 at 17:37
  • 4
    \$\begingroup\$ @HagenvonEitzen not stupid at all, interesting thought. But no, it wouldn't. Look at the 6-bladed curved aperture (2nd from right, bottom diagram in my answer). Notice the effect that curved blades (i.e., curved edges) have on the diffraction pattern. It spreads out the rays, "smudging" them. Spread out diffraction like that appears to our eyes as a loss of sharpness, or fuzziness. For deep sky images, nice clean diffraction spikes are easier to deal with (clean up or remove, if desired) than spread out fuzziness. \$\endgroup\$
    – scottbb
    Commented May 4, 2017 at 18:14
  • 2
    \$\begingroup\$ @HagenvonEitzen Also, the shortest distance between two points is a straight line. A curved path is by definition longer. Longer struts take up slightly more area of the aperture. And aperture is all about area, and maximizing it. Now, to be fair, it probably has a negligible effect on the aperture number (way down in the decimal places). The spread-out diffraction has a far greater effect on the image. \$\endgroup\$
    – scottbb
    Commented May 4, 2017 at 18:18
  • 3
    \$\begingroup\$ @HagenvonEitzen There are companies selling curved blade holders for secondary mirrors in telescopes. They're really only for visual use though because while the smears they make are generally too faint to be seen as scottbb has pointed out the total amount of light that gets difracted into the wrong location is higher and a camera/etc sensor will pick it up. AIUI unlike the curved blades from the image above 90* curved spider vanes smear the diffraction into a halo instead of wide spikes. \$\endgroup\$ Commented May 4, 2017 at 18:24
  • 2
    \$\begingroup\$ @WayneWerner For sure. Computing power and image processing is cheap and getting better all the time. But launching heavy moving parts into space (magnets and coils are just plain dense) violates 2 big no-no's in space tech: complexity that doesn't serve an absolute necessity (means stuff will break); and lots of dead mass (very expensive to launch). \$\endgroup\$
    – scottbb
    Commented May 5, 2017 at 15:10

@scottbb's excellent answer is thorough. For scientific analysis of images from Hubble, one has to account for all aberrations, diffraction, and pixilation. There is a nice writeup on this: 20 years of Hubble Space Telescope optical modeling using Tiny Tim (paywalled, also available here and see this page).

enter image description here

enter image description here

enter image description here


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.