Why can not I take pictures of stars (constellation of Orion) ?

[Pawelk198604]

I could not sleep last night so I went on my balcony, I saw that the sky is clear and you can see all the stars:-) I live in a city rarely when the stars are so beautifully displayed as to be visible tonight:-)

Unfortunately, when wanting to photograph it, the camera on my phone (Sony Xperia J) in the picture could not see anything.

A bit like it was with pictures from the Apollo missions, doubters do not believe that man was on the moon, because those stupid stars are not visible:D

[DeMatt]

Short version? Cellphone cameras are optimized (such as they are) for daylight and indoors work, where there's lots of bright light sources. Starlight is several orders of magnitudes dimmer than that.

You can try fiddling with the settings - a quick glance at the manual suggests it has both a "Night scene" setting and an adjustable "Exposure value". But if you're serious about getting into astrophotography, I think you'd do better to buy a standalone camera.

[Xaiier]

Because it's a phone.

Astrophotography requires some advanced capabilities and a long exposure time, because the light from stars isn't very bright compared to even moonlight, let alone sunlight.

[K^2]

If your hands are steady enough, or better yet, if you have a camera stand, you should be able to take a picture of the stars. You'll get much better results with a camera with a large lens, but even on most modern camera phones, you can get a faint picture where you can see some stars. You might have to play with exposure settings a bit, though. It's unlikely that you'll get anything on auto settings.

Once you can do that, the next step is to try to take a whole bunch of pictures like that and then use stacking software to put them together into a single image. Several members of this forum have shared pictures they've made this way.

[jfx]

tl;dr:

Cell phone cameras are "optimized" to be small. Small camera sensor equals not much light hitting camera sensor equals dark+grainy picture. Dark+grainy picture after camera firmware equals dark+blurry picture. And since you made the moon image comparison: If i remember correctly they used a Hasselblad Camera with a 5.6x5.6cm negative size. That's somewhere north of 300.000mm² (not really compareable but still ...). And now you know why you need at least an aps-c or 4/3 camera to shoot anything good. ... or a webcam.

[K^2]

Small camera sensor equals not much light hitting camera sensor equals dark+grainy picture.

Sensor size has nothing to do with signal-to-noise ratio. Twice the area of the sensor simple means it needs half the time to get the same exposure. You can totally compensate for small sensor size in phone cameras with longer exposure times.

On the other hand, some phones have just crappy sensors. Sony's been pretty good about it, however. Xperias usually have quite decent night capture, for phones, at any rate. And like I said above, with the right settings and a steady shot, you can get a picture with some stars visible. And that you can use with stacker to bring out some of the dimmer stars as well.

People on this forum have done that. They have posted pictures of stars taken with things like iPhone. I'd go outside and do a proof of concept, but we happen to be under dense cover right now, and doesn't look like it's going to clear out tonight.

[Pawelk198604]

I thought I could take a picture of Orion then I threw it on Facebook and I would be showcased in front of my family and friends, and my nerdness would up to next level

Unfortunately my camera phone to be completely unsuitable for the job:-(

Now soon end my contract for a mobile subscription with the Orange company , I wonder whether to sign a new contract, they have the Samsung Galaxy S3 at a decent price. I wonder if the camera is good for shooting night scenery?

Edited November 3, 2014 by Pawelk198604

[Umlüx]

no phone camera will ever be good for such pictures. they produce too much noise and lack manual settings.

[m4inbrain]

Sensor size has nothing to do with signal-to-noise ratio. Twice the area of the sensor simple means it needs half the time to get the same exposure. You can totally compensate for small sensor size in phone cameras with longer exposure times.

First part is correct, more or less. There's a difference in noise between camera sensors and for example the Canon D series, and that's the pixel-size. Larger sensors generally (not always, but usually) have bigger pixels, which reduces noise and increases the high dynamic range. And the second part of your statement is wrong. In astrophotography you can only expose for so long until you ruin your picture. In general, on a tripod, you have a maximum exposure time of roughly 25-30 seconds. Everything above that ruins the picture. My phone has a fixed aperture (not iPhone), wouldn't work for proper pictures. Now i heard the iPhone of some sort has a fixed aperture of f/2,2 - which might be good enough if the ISO is adjustable (1200+). But in general, starphotography with a phone is pretty much like eating soup with chopsticks. Works. Kinda. But not really what you'd want.

On the other hand, some phones have just crappy sensors. Sony's been pretty good about it, however. Xperias usually have quite decent night capture, for phones, at any rate. And like I said above, with the right settings and a steady shot, you can get a picture with some stars visible. And that you can use with stacker to bring out some of the dimmer stars as well.

Or you invest 100 bucks or less in a used dSLR (Nokia D3100 springs to mind), and just go for it.

A Galaxy S3 is kind of a pretty old phone, relatively speaking. I'd rather go for an Xperia Z1, which is pretty capable in the camera department.

[K^2]

In astrophotography you can only expose for so long until you ruin your picture. In general, on a tripod, you have a maximum exposure time of roughly 25-30 seconds.

That's 20th century thinking. Deconvolution can increase that threshold to several minutes, easily. That introduces a new source of noise, but one you have much more control over. Specifically, once you can capture images of some stars, sharp enough to ensure good localization, you can just keep stacking these images, and noise will go down as inverse square.

But that's an aside. The important bit is that even with exposure you can get without doing any numerical tricks, you can capture brighter stars with a decent cellphone camera. And then you can skip the deconvolution and go straight to stacking.

[cantab]

I've never heard of any processing technique to "untrail" stars on long exposures. Only physically counter-rotating the camera on an equatorial mount. Any more information on the specific deconvolution you refer to and demonstration of what it can do?

Regardless, for Pawelk's question, you need to be able to do a long exposure of 15-60 seconds. If you can do that on your phone you're in with a chance. And you need the phone rock steady, so prop it up somehow, handholding it is no good.

[m4inbrain]

It's a bit dishonest of K^2 to refer to deconvolution. He's not wrong in essence, but the tools required (the software, rather) is rarely, if ever, found in private households. It's a technique used for Hubble when the mirror was defective, for example. Or in microscopy. It's like saying that you totally can do your own gene-experiments at home. In theory you could, in real life though you won't have the tools necessary available.

In terms of deconvolution, we are talking about somebody who wants to take pictures of stars with his phone. It's simply a red herring from K^2. Yes, there's free software out there that claims to do the trick, but they don't. I tried some of them to remove startrails after only a 35 second exposure, looked worse after it was processed. edit: and i shoot with a dSLR at f/2.2 and ISO 800 roughly, depending on light pollution - best premise to make those programs work in theory.

Stacking for a cellphone is the only reasonable way to get decent pictures (not good, but decent). Single exposures will never look good.

edit:

Regarding Pawels question about the camera on the S3:

That's ISO 400 and it's already pretty noisy.

edit 2: and get something like http://www.amazon.co.uk/Tripod-Mount-Holder-Mobile-Camera/dp/B009J79NK8 , simply because it's impossible to shoot high exposures by hand - even if some people claim it. You will have trouble to keep a camera steady by hand for a 2 second exposure, let alone a 30 second exposure (or even minutes). Your heartbeat alone will ruin the picture.

Edited November 3, 2014 by m4inbrain

[PakledHostage]

I've never heard of any processing technique to "untrail" stars on long exposures. Only physically counter-rotating the camera on an equatorial mount. Any more information on the specific deconvolution you refer to and demonstration of what it can do?

You can't "untrail" the stars, but you can avoid capturing star trails in the first place by taking shorter exposures. You can then use image stacking software like the free Deep Sky Stacker to stack multiple images into an image with higher effective exposure length.

Use the "Rule of 500" to set an upper limit for your exposure length and then adjust your ISO and aperture as required for the image you're trying to shoot. Fire off a number of shots (light frames), plus some dark frames and bias frames and then load them into DSS. The software will automatically rotate each light frame to correct for the Earth's rotation by aligning the pattern of stars.

[BagelRabbit]

Fun fact: The stars you see in the sky are not tremendously large, or bright.

Also, your camera has a very small CCD chip and a very small camera aperture.

Also, it doesn't have arbitrarily long shutter speeds.

So, here's what's happening: You attempt to snap a picture of some really small and dim things. Your camera only captures a fraction of the light that they emit, and you can't leave the shutter open for long enough for more than a few photons to find their way into your camera. Furthermore, any bright bits you get are likely pixels that are, in effect, broken. Your stars are dimmer than these bright pixels. This isn't great.

Here's how to get good pictures of the stars without buying another camera, downloading software, etc.:

1) Take your cellphone camera and butt it up against a friend's telescope. Find a bright star, and it should show up reasonably well on your camera. (BONUS! You can take surprisingly good pictures of the Moon this way, and even sometimes see Jupiter's moons!)

2) Take a picture of the Sun. It is a star, and you will be able to see it on your cell-phone camera.

3) Search for 'Orion constellation' in Google Images. Find a nice, high-resolution picture and post it to all of the social media sites you want. Be sure to give credit to the person who took it.

That's really all you can do at this point, without spending a fair sum of money. (Sorry.)

Edited November 3, 2014 by UpsilonAerospace

[VirtualCLD]

Fun fact: The stars you see in the sky are not tremendously large, or bright.

Also, your camera has a very small CCD chip and a very small camera aperture.

Another fun fact, most consumer electronics cameras (smartphone, point and shoot, and even SLR) use CMOS imagers, not CCD imagers. You are right in that most astronomy cameras are CCD. CCD arrays tend to have better signal to noise than CMOS. That's because, for a given pixel size, say 2um x 2um square, most of that area is active (used to capture light) in a CCD array. In a CMOS imager, a good chunk of that area is taken up by the control and readout circuitry. Therefore, some of that pixel area is not capturing light, so you don't get as good SNR.

[cantab]

You can't "untrail" the stars, but you can avoid capturing star trails in the first place by taking shorter exposures. You can then use image stacking software like the free Deep Sky Stacker to stack multiple images into an image with higher effective exposure length.

Indeed. Stacking is pretty much universal in astro imaging these days. But K^2 seemed to be referring to something else, some processing done on the individual frames before they're stacked.

[YNM]

Simple answer : the pixels of the CMOS in your phone are small, and on night mode, is set to highest ISO. This is what limits it.

CMOS aren't like CCD that notes down just every single photons that falls on it; CMOS has a sensitivity level, called ISO. While a higher ISO means higher sensitivity, it also means that the noise generated from both the background and the sensor itself (well, they do become hot) becomes more visible - while the signal value stays the same (often referred as x/n, signal to noise ratio. Also, if you ever wonder why CCDs have a cooling fan...) . There are some ways to get around this; most often are to join up some pixels (like, I've seen such option in CCD - some phones do that too). But have you ever thought to make your 8MP phone takes only 4MP, say ? You don't want it ? Then there's the noise.

Other things are the apperture (that is, diameter of the camera opening vs the focal length) and the physical size of the CMOS. I'll talk about apperture : the view is the smaller the apperture is, the faster time needed to collect the same amount of light (or say, energy). Unluckily, this means larger FOV (and so you'd see Orion small, I guess).

The only unlucky thing here is that no one can express the ISO of a phone to the usual levels used on DSLR/SLR; so no real comparison.

Read more here : http://beta.techradar.com/news/phone-and-communications/mobile-phones/what-does-low-light-photography-in-your-smartphone-mean--1210247/1

Edited November 3, 2014 by YNM

[PakledHostage]

Indeed. Stacking is pretty much universal in astro imaging these days.

Sorry. I assumed that you weren't aware of stacking as you were going on about how "only physically counter-rotating the camera on an equatorial mount" could avoid star trails. But I do agree that a tracking mount is a nice to have. Given that stacking is as effective as it is, I am sure that a combination of a tracking mount and stacking would be awesome! I've got my eye on a tracking mount but I haven't bought it yet because there's no point when you live somewhere where it is almost always cloudy. Maybe I'll spring for one if I go on a trip to somewhere with more reliably clear skies.

[Beowolf]

Sensor size has nothing to do with signal-to-noise ratio. Twice the area of the sensor simple means it needs half the time to get the same exposure. You can totally compensate for small sensor size in phone cameras with longer exposure times.

You can compensate with longer exposure times, but the amount of random noise in an image increases in long exposures. A decade ago at least, the last time I looked into such things, the noise grew in an accelerating curve as the sensor heats up during the exposure. Beyond a second or two noise grows faster than signal so SNL is indeed impacted.

Pro-level cameras compensated by mounting the sensor on a big heat sink, but sensor-based anti-shake was new and exciting then, and incompatible with rigidly mounting the sensor to a heavy heat sink. Which I assume is why a quick search tells me those pro cameras now tend to have lens-based anti-shake. Lens-based is significantly more expensive, so I'd say they haven't solved this issue yet.

edit: Stacking dim pics in software is a new solution. I like it!

Edited November 3, 2014 by Beowolf

[K^2]

You can compensate with longer exposure times, but the amount of random noise in an image increases in long exposures. A decade ago at least, the last time I looked into such things, the noise grew in an accelerating curve as the sensor heats up during the exposure. Beyond a second or two noise grows faster than signal so SNL is indeed impacted.

Modern sensors don't seem to have that problem. At least at reasonable exposure times.

You can't "untrail" the stars

Of course you can. Trails are just a convolution, so you can deconvolute data with an educated guess of the kernel and a sensible filter.

It's a bit dishonest of K^2 to refer to deconvolution. He's not wrong in essence, but the tools required (the software, rather) is rarely, if ever, found in private households.

I wrote my own. It's a fairly trivial matter, if you understand convolution theorem and optimal filters or MEM theory. Alternatively, a lot of kernels can be deconvoluted using SVD methods. Though, I've only used that to reverse gaussian blurs. I've never tried SVD on motion trails.

[cantab]

At least at reasonable exposure times.

Astrophotography, though, demands what in other fields would be wholly unreasonable exposure times. 5 or 10 minute single exposures ("subs") aren't uncommon. Active cooling of the sensor to reduce thermal noise also isn't uncommon.

Of course you can. Trails are just a convolution, so you can deconvolute data with an educated guess of the kernel and a sensible filter.

I wrote my own. It's a fairly trivial matter, if you understand convolution theorem and optimal filters or MEM theory. Alternatively, a lot of kernels can be deconvoluted using SVD methods. Though, I've only used that to reverse gaussian blurs. I've never tried SVD on motion trails.

So you've no demonstration of the technique actually working for astro-imaging.

At this point my thought is that if it was any good it would be far more widely used. If a "trivial" computer program could replace a several-hundred-dollar tracking mount, people would be doing that. And yet in all my time reading astronomy mags and browsing forums I've never even heard of the idea before.

[Wanderfound]

Fun fact: The stars you see in the sky are not tremendously large, or bright.

Also, your camera has a very small CCD chip and a very small camera aperture.

Also, it doesn't have arbitrarily long shutter speeds.

So, here's what's happening: You attempt to snap a picture of some really small and dim things. Your camera only captures a fraction of the light that they emit, and you can't leave the shutter open for long enough for more than a few photons to find their way into your camera. Furthermore, any bright bits you get are likely pixels that are, in effect, broken. Your stars are dimmer than these bright pixels. This isn't great.

Here's how to get good pictures of the stars without buying another camera, downloading software, etc.:

1) Take your cellphone camera and butt it up against a friend's telescope. Find a bright star, and it should show up reasonably well on your camera. (BONUS! You can take surprisingly good pictures of the Moon this way, and even sometimes see Jupiter's moons!)

2) Take a picture of the Sun. It is a star, and you will be able to see it on your cell-phone camera.

3) Search for 'Orion constellation' in Google Images. Find a nice, high-resolution picture and post it to all of the social media sites you want. Be sure to give credit to the person who took it.

That's really all you can do at this point, without spending a fair sum of money. (Sorry.)

Not quite; a few years ago some guys managed to get a shot of the ISS through an iPhone camera, clear enough to see the shuttle docked to it. It was, as you say, shooting through a telescope at the time, however.