Hacker News Reader

Comment by mysterypie | Hacker News Reader

I was wondering why the star has 6 crisp points and found an explanation here (a 3 minute video):

https://www.youtube.com/watch?v=VVAKFJ8VVp4

The most interesting part of the video explains why even your naked eye viewing the sky at night will cause this effect -- it's due to imperfections in the lens of your eye.

View on HN

I think the wikipedia explanation is much better.

https://en.wikipedia.org/wiki/Diffraction_spike

One deficiency is why apertures (effectively, the support structure in a telescope is an aperture) observe this behavior.

> No matter how fine these support rods are they diffract the incoming light from a subject star and this appears as diffraction spikes which are the Fourier transform of the support struts.

Like why is that the case?

https://en.wikipedia.org/wiki/Airy_disk

Read this if you don't immediately understand why. This is the shape of the image you see when the aperture is circular. This is the Fourier transform of a circular aperture, which, in 1D, is a sinc function.

I can't give a straightfoward answer to elucidate further, but if you've done signal processing stuff before you can probably handwavey explain that if there is a Fourier-transform relationship with one function, due to linearity and spacial invariance, you can say that it holds for all functions.

So we're not actually seeing the image of a star but only its Fourier transform?

I presume that only happens if the star is far enough away, as there are plenty of images of our Sun (which is much closer) that don't look like this.

For larger objects with a larger subtended angle, we are probably seeing the union of the Fourier transforms of each “pixel” (I’m not sure what the smallest element would be, in this case).

Good summary but I found it a bit weird how only 2 seconds were spent on the 'camera aperture' explanation - as AFAIK that's the main cause for photos of stars and lights to have 'points' - it depends on how many 'leaves' (? term) your camera/lens aperture has -> the more leaves the "more-sided" a shape the aperture makes (hexagon, octagon...) thus the more points you get when photographing lights.

The number of rays is determined by the number of aperture blades, as well as whether there's an even or odd number of aperture blades

https://phillipreeve.net/blog/best-lenses-for-sunstars/#The_...

Interestingly, some modern photography lenses have achieved aperture mechanisms with much rounder geometry, sometimes with near perfect circles at multiple apertures. This can result in a more desirable bokeh, at the cost of well defined sun stars.

The article fails to explain that, but round aperture is equivalent to infinitely many straight blades (and thus rays), so there is a light halo instead of distinct rays.

Which is essentially also why you get more diffraction limited as you stop down the aperture -- a larger and larger fraction of the light that gets through passes near the edges (and gets diffracted) rather than the central area.

yup, but when you step down the aperture a lot (which is when the sun star effect becomes more pronounced), the aperture transitions from more circular to more polygonal, so with a lot of lenses, the behavior is that when your aperture is a few steps of fully open, it is basically circular but when you need the sun starts it's definitely there.

How do stars look like in animal eyes?

To me it looks like this effect: <https://www.theclickcommunity.com/blog/creating-starbursts-i...>

I wonder what they would look like if stars were pointed objects.

The same. Because any shape to a star is too small to be detectable with these kinds of resolution.

I see 8. With the smaller horizontal ones likely coming from one of arms of the secondary mirror (check the “selfie” image). The linked video is nice, by the way.

From another article I read, it appears that most of the pattern is due to diffraction from the edges of the hexagonal mirrors.

Awesome video!

Made me laugh and learn at the same time :-)

Reply on news.ycombinator.com