star sizes in photos?

[mgoodfel]

Why are stars different sizes in astronomical photos? I assume that at interstellar distances, they are all point sources. Yet in a typical APOD picture, you'll see widely varying sizes. I thought perhaps it was due to atmospheric effects spreading the stars over a larger area ("twinkling" spread over time), but you see the same effect in Hubble pictures. So I guess I just don't understand.

[bystander]

I think the size of stars in astrophotography has more to do with their relative brightness than their actual size. Brighter stars appear larger than dimmer ones.

[geckzilla]

Oh, you know those photos of the moon where the moon just turns into a big, bright, sun-like glowing ball and doesn't look like a moon at all? And yet, you know it was just the moon and it wasn't that big.

If it was possible for a camera to hold more levels of gray past pure white or for us to even see them, maybe it wouldn't end up so nondescript. Er, well, I don't really know, but I always wondered about that.

[apodman]

"twinkling" spread over time

Close. Try Airy Diffraction or Airy Disk (google those terms for more information) spread over time. Airy was someone's name and has nothing to do with air, so it does happen with optics in space. For ground-based views, add atmospheric scattering of light spread over time.

[geckzilla]

Does that really explain the size of each dot? I thought it was an explanation for the concentric rings. I admit that I find it difficult to fully understand the Airy disk though, even with the wikipedia article.

[bystander]

Could it have something to do with pixel saturation and spillage or flooding into surrounding pixels?

[apodman]

geckzilla - If you zoom in on a star, the spaces between the concentric diffraction rings (actually not spaces but just dimmer than the rings and which may brighten in a time exposure) may contain many pixels. But as you zoom out, the distances between the rings may become equal to or smaller than one pixel (I haven't actually done the math on that, so I'm going out on a limb) in which case the rings are all combined into one blur.

bystander - I suppose in a perfect CCD there would be no spillage from one pixel to the next, and you could have a fully- or over-saturated white pixel surrounded by black pixels that would not brighten regardless of additional exposure time. I also suppose that in a real CCD there is some spillage among pixels both in the physical acquisition of the photons and in the conversion of impacting photons into a voltage, but I'm not the resident CCD expert so I can say no more regarding how much spillage there actually is.

[The Code]

I have very good eye sight.. And with the naked eye '' Beatle Juice'' looks a red-er colour to the Orion belt...Stars

I would never of known that with out this forum...

Mark

[Chris Peterson]

Why are stars different sizes in astronomical photos? I assume that at interstellar distances, they are all point sources. Yet in a typical APOD picture, you'll see widely varying sizes. I thought perhaps it was due to atmospheric effects spreading the stars over a larger area ("twinkling" spread over time), but you see the same effect in Hubble pictures. So I guess I just don't understand.

Optically, stars are not point sources. They are spread out in diameter by both atmospheric motion (seeing) and by diffraction. Scatter from optics can also increase the size, although this is not usually apparent in well made images. Internal reflections from the cover glass over the CCD frequently make bright stars look even bigger.

The effect of all this spreading of light is to produce a bunch of stars with about the same profile- close to a Gaussian, although that function doesn't exactly describe the shape. Every star has about the same full width at half maximum (FWHM), meaning the width of the profile at 50% intensity. But each star has a different apparent disc size, because brighter stars have the tails of their profile higher above the noise floor. Sometimes the effect is enhanced by processing: it is common in many images to selectively brighten dimmer areas, which will also increase the apparent size of stars in that region.

[Chris Peterson]

geckzilla - If you zoom in on a star, the spaces between the concentric diffraction rings (actually not spaces but just dimmer than the rings and which may brighten in a time exposure) may contain many pixels. But as you zoom out, the distances between the rings may become equal to or smaller than one pixel (I haven't actually done the math on that, so I'm going out on a limb) in which case the rings are all combined into one blur.

It's certainly true that undersampled images hide the diffraction effects in a single pixel, or very small group of pixels. But in reality, you almost never see diffraction rings in long exposure images made from the ground. Even under the best of seeing conditions, there is usually enough smearing from the air column to wash them out, making the star a single blob. You do see diffraction rings in all the HST images, however, since there is no atmospheric distortion.

bystander - I suppose in a perfect CCD there would be no spillage from one pixel to the next, and you could have a fully- or over-saturated white pixel surrounded by black pixels that would not brighten regardless of additional exposure time. I also suppose that in a real CCD there is some spillage among pixels both in the physical acquisition of the photons and in the conversion of impacting photons into a voltage, but I'm not the resident CCD expert so I can say no more regarding how much spillage there actually is.

It depends on the sensor. A good CCD may show blooming, which is a smear along a column from a saturated star. Some CCDs have circuitry to prevent this, at the expense of sensitivity. A CCD will not show any radial increase in a star's diameter, because rows are completely independent. Charge can only bleed along columns. And for unsaturated stars, there will be no visible spillage or carryover along the columns, either. The biggest effects at the sensor level are likely to be scatter and reflections between the CCD surface and the inside and outside surfaces of the cover slip, and scatter or reflections within the microlens array found on some CCDs. Professional astronomical CCDs do not use either microlens arrays or cover glass.

[mgoodfel]

So I think you are saying this: No, a star isn't "really" a point source, but since these telescopes aren't resolving the disk of the star, that's moot. The angle cut by the star is less than a pixel.

What is happening is that both bright and dim stars are being spread by this Airy Disk effect. Both the dim and bright stars spread over the same area, but since there's a minimum value that the camera (film or digital) can record, that chops off the base of the curve for the dim star, leaving a smaller image. And since there's a max value as well, the bright star is rendered as a solid bright circle, not a graded disk fading towards the edges.

This picture is what I think you mean. Is this right?



I was trying to simulate a star field on my computer, and of course the first thing I did is sprinkle random dots. That immediately brought me up short, since it looks nothing like an APOD picture. I'll try it again with these cropped Gaussian shapes and see if that looks more realistic.

[Chris Peterson]

So I think you are saying this: No, a star isn't "really" a point source, but since these telescopes aren't resolving the disk of the star, that's moot. The angle cut by the star is less than a pixel.

That's just the theoretical size. In reality, a star is not a point source- or even close to it- because of diffraction and various smearing effects.

What is happening is that both bright and dim stars are being spread by this Airy Disk effect. Both the dim and bright stars spread over the same area, but since there's a minimum value that the camera (film or digital) can record, that chops off the base of the curve for the dim star, leaving a smaller image. And since there's a max value as well, the bright star is rendered as a solid bright circle, not a graded disk fading towards the edges.

That's essentially correct. However, with long exposures the diffraction (Airy disc) effect is usually less significant than the spread due to atmospheric conditions. It helps to visualize what happens with the dimmest stars: they are always just one pixel in size (ignoring the possible perfect straddling of pixels). As the stars get brighter, they get bigger and bigger. In fact, each star has a disc that is infinitely large in diameter, but only the very central portion is above the noise floor.

By keeping the subexposures short enough to prevent saturation, and being careful with processing, it is possible that even the brightest stars will have proper profiles, without the top being clipped. So bright stars don't necessarily get rendered as discs, although this is quite common.

[geckzilla]

I was trying to simulate a star field on my computer, and of course the first thing I did is sprinkle random dots. That immediately brought me up short, since it looks nothing like an APOD picture. I'll try it again with these cropped Gaussian shapes and see if that looks more realistic.

I have been playing with star fields recently (I put them on my website if you want to see). I mean, I've always enjoyed painting them in my art, but lately I've been really studying what makes an aesthetic field, so I have a little advice if you want something that's pretty. The best formula I've found so far is to have thousands and thousands of dim, tiny points. It's very important to layer these up if you want to create, say, a band of milky way. If you are rendering these real time you could use less computing power by simulating these fine points of light as just a few large, blurry patches, but the more noise effect you can sneak in, the better. After you get that part down the rest is gravy, a sprinkling of moderately bright stars and then 2-5 very bright ones. That glowing bloom effect is pretty important. Having a crisp central bright point is better than having a gradual, blurry taper of light. And depending on what colors you're allowed to use, lots of the dim stars look good in warm yellows and reds with the brighter ones mostly cyan, some yellow, and a tiny bit of white.

...Sorry for going off on a rant, I just really enjoy star fields.

[apodman]

Before there was Sky & Telescope, there was TV Guide. Actually, S&T was November 1941 while TVG was April 1953, so that's not true - but on with the story anyway, as I had subscriptions to both when STNG came on the air. The TV Guide article reported that the guy who created the starfields for STNG made them by throwing darts at various velocities into cardboard and backlighting the pattern of holes to be photographed. I'm sure an old analog 525-line interlaced TV raster (with even less horizontal resolution) needed a field of more crudely rendered stars than geckzilla's pixel painting requires. Maybe geckzilla will develop the definitive process for HDTV space backgrounds and NYU will open a department dedicated to the science.

[jerbil]

Of course, long distance interferometry is invaluable in the direct measurement of stellar dimensions (so I am informed - I have never done it myself.)

A contributor to this thread mentioned Airy, a great 19th Century physicist. He is also well known for his accurate analysis of the formation of rainbows, which are considerably more complicated in structure than commonly thought.

[geckzilla]

Oh, apodman, I remember seeing some tv show about how that guy made the nebulae. I don't remember the stars or the part about the darts, but he had some way of using electricity on, I think, a flat, liquid surface. Maybe it was just a few plain paints put on and then he applied the electricity? It was really cool. Hm, I wish I could find some online article about it. Google isn't helping me this morning.

[apodman]

Google isn't helping me this morning.

Yeah, try to find a simple photo of a galaxy air-brushed on a van, and nothing. I think we've been re-routed through the Chinese filters.

[BMAONE23]

electircity used to create art
is this what you were looking for?
http://www.davearcher.com/images.html