Potentially advanced - Software substitute for Equatorial Mount?

[DrunkAstronomer]

Hey all,

I'm a newbie so this goes in the newbie section, but this might be an advanced question.

I'm a photographer who has been bitten by the astronomy bug (Cosmos on Hulu!) and I've been looking to do cheap astrophotography... only, I know there's no such thing as cheap astrophotography, as told by google and the array of equipment needed to do it.

I'm also a software guy though, so I've wondered - what about a software solution? Software can be as cheap as free...

Does any software exist that can take multiple exposures taken from a tripod and combine them (with appropriate rotation and translation) into one photo with similar qualities to a long exposure taken with an equatorial mount?

If not, I could potentially write such software, but is it even plausible? I know digital cameras take the received light and turn it into a digital signal, so if the light is so faint that it registers as a zero digital value, adding multiple exposures will get you nowhere...

However, if you could take multiple exposures with a tripod, and then combine them and align them with software, do any of you think there might be some value? How long of an exposure can you manage without an equatorial mount? 1 second? 5 seconds? Longer?

You need to get enough exposure that *something* shows up in the image. It also needs to be short enough that you basically don't get any star streaking. I imagine i could rotate and add images, but not remove individual star streaks.

What is the state of software like this? I hear that observatories nowadays use cheaper alt-az mounts and correct for rotation in software, but they have very sensitive instruments. Could we do what I'm talking about (which is more complex) with our much cheaper digital cameras?

Thanks for indulging a drunk astronomer.

-Taylor

[jgs001]

Taylor, the problem with exposing from a tripod is field rotation, and even at 18mm that'll kick in around 25 to 30 seconds. If you take multiple exposures that don't show field rotation, you can combine them in Deep Sky Stacker and that will account for field rotation between frames, aligning all the images up to a point (which I didn't determine) after which you start to get smearing in the corners the more subs you add.

http://stargazerslounge.com/imaging-tips-tricks-techniques/73737-basic-widefield-camera-tripod.html

I've used a lightweight, cheap AltAz mount for deep sky, and as long as you're willing to throw a very large number of exposures at it, you can get some decent results... as long as each sub doesn't contain field rotation. This last limits the exposure lengths to up to 2 minutes, if you are low in the east or west, which isn't really ideal target locations. Overhead, I was limited to about 40 seconds. The sub length isn't affected by focal length, but obviously, the longer the scope/lens, the more accurate the motors have to be, and the more weight you are loading them with.

[themos]

Taylor, you need to learn about the basics of digital imaging. I suggest you start here

Signal to Noise: Understanding it, Measuring it, and Improving it (Part 1) - Article

and then read all the other Parts.

[ollypenrice]

Yes, the problem is that individual exposures need to be long enough to allow signal to compete with (and vanquish!!) noise. In these exposures there will be rotation. If it could be done we'd have done it. Sad but true.

Olly

[MartinB]

The problem is read noise which is the dollop of noise served up every time te chip is read and converted into digital format. Unfortunately this noise isn't eliminated by stacking unlike the noise which builds up during the exposure (shot noise). You would be restricted to images of a few seconds and noise would dominate no matter how many exposures you stacked.

Now if you could design a zero read noise camera you would be onto a winner!

[DrunkAstronomer]

Hey guys, thanks for all the responses!

Yeah, I forgot about noise - it makes sense that that would be the larger problem!

But at least there is software to try it anyway! I mostly imagined that if software existed, someone would be trying to charge a bunch for it. Glad to see some free stuff is there!

I'll follow that tutorial and give it a shot. I've taken a few single long exposure shots a night, but I've never tried combining any and I didn't know about that software.

Thanks!

-Taylor

[dph1nm]

The problem is read noise which is the dollop of noise served up every time te chip is read and converted into digital format. Unfortunately this noise isn't eliminated by stacking unlike the noise which builds up during the exposure (shot noise).

Not quite true - read noise is reduced by stacking, just like shot noise. Its just that you get more of it if you do a bunch of short exposures rather than one long one. So you have to do proportionally more exposures to reach the same signal-to-noise - but you could get the same image in the end if you tried hard.

NigelM

[themos]

Right, I'm off to hit a galaxy with the webcam :-)

[pete_l]

Taylor,

Putting aside all the objections already listed, the answer is "yes you can", with limitations. You'll take lots of reasonably wide shots of the sky from a fixed camera and stack them together (as has been mentioned already). What you'll need to do before you start is to work out what the maximum exposure time will be, in order to stop the natural rotation of the earth from moving a star off one pixel on your camera's sensor onto the next. Once that happens, you've reached the limit of sensitivity, as no matter how much longer you expose, that original pixel won't pick up any more light from that star.

Let's assume your camera has 3,600 pixels horizontally and 2,400 vertically - roughly an 8Mpixel camera (yes, I know the numbers are wrong but it makes the calculation easier to illustrate). Now, the earth rotates at 15° per hour, which is 15 arc-seconds per second. If you use a lens with a field of view of 45°, then each pixel will "see" (45/3600) degrees of the field - or 45 arc-seconds. That limits your maximum practical exposure to about 3 seconds - not much!

You can "bin" your images, which makes them smaller - with fewer pixels in each direction. So if your software does this properly, all the light from (say) each 3x3-pixel square will be added together and you'll end up with an image of 1200x800 instead of 3600x2400 - but you can expose for 3 times longer before the stars drift into the next binned pixel. An 8 or 9 second exposure still isn't very long, but if you use a higher ISO, you can start to get some decent shots.

The next trick is to take LOTS of binned photos - maybe 60 as a starting point. Save them at the higherst quality setting you can. You'll now need a piece of free sofwtare called Deep Sky Stacker which will take all these images, align them so all the drifting stars come together and add up the signal, while cancelling out to some degree the noise from the images. Save the "stacked" image and run it into photoshop or some other digital darkroom package, where you can play about with setting the black point, surves, removing light pollution and all the other tweaks that really take the time with astrophography.

If you find you don't like the results, or that it doesn't meet your expectations - well, no matter. It hasn't cost anything. However, if you do find yourself getting bitten by the bug, you can start researching and investing in a motorised mount to counteract the effect of the earth's rotation, so you can take longer duration images, without having to worry about binning.

[MartinB]

Not quite true - read noise is reduced by stacking, just like shot noise. Its just that you get more of it if you do a bunch of short exposures rather than one long one. So you have to do proportionally more exposures to reach the same signal-to-noise - but you could get the same image in the end if you tried hard.

NigelM

OK. My new year resolution is never to discuss noise on any forum ever again since in my experience there is always someone who will disagree on some fine point and the discussion ends up going way beyond the role of the orignial thread. The practical reality of read noise (be it gaussian or fixed pattern, is that your exposures need to be long enough for it to make only a very small contribution to the overall noise otherwise it will stay with you like a curse)

That's it now, new year has come early!

[themos]

LOL at MartinB!

Signal to Noise and Subexposure Calculations explains it rather well

This is a graph of SNR vs subexposure duration ("sub length") for a fixed total exposure time. Depending on the read noise, changing the sub length can make hardly any difference or all the difference. If you are way on the left (quick sub exposures) you will be swamped with noise. But you can always improve your SNR by increasing the total exposure (just take many, many more subexposures). Is anyone up to trying 10,000 2 second exposures of Andromeda? 50 Gigabytes of data to go through...ouch

[dph1nm]

The practical reality of read noise (be it gaussian or fixed pattern

Hey, I was keeping quiet about the fixed pattern element - that is a whole new can of worms!!!

NigelM

[MartinB]

I have produced similar graphs using Craig Starks sub exposure calculator and use these to determine my optimal sub exposure time. http://www.stark-labs.com/craig/articles/assets/SNR%20Calculator.xls

Using this graph I would be looking at 10 minute sub exposures by which time the read noise is making a very small contribution.

[MartinB]

Damn! resolution broken already.

[Psychobilly]

Damn! resolution broken already.

Tut Tut.. you really must try harder...

[DrunkAstronomer]

Hey Guys!

I took the advice about deep sky stacker and the exposure techniques, and went out and took some shots.

With about 15 lights and 5 darks at 30 seconds apiece, I wasn't getting any appreciable streaking, but the composited image wasn't that impressive either.

It seems to be a conversion issue though - the individual exposures actually look better than the composite, so clearly I'm just doing something wrong.

I didn't spend much time with the software (had to go to bed) so obviously I should just look at it more myself and learn it. But I was wondering if you guys had any obvious insight. The whole image was just too dark, and messing with the curves didn't seem to give me anything even close to as good as one of the single exposures. Does it have something to do with the RAW images? Do I need to process them to another format BEFORE putting into Deep Sky Stacker? It seemed like maybe it was just converting them automatically with some default (and incorrect) value.

If you have any thoughts, let me know. I mostly just need to spend more time with it though.

In other news, I bought my first telescope today! Just an Orion XT8 basic dob - not expecting to do photography with it, I just want to learn more about astronomy and have fun. I was hoping to be able to do wide field stuff with it using the techniques from this thread, but they claim I couldn't even get it to focus with a camera as it wasn't designed for it. I'll have to see if there is any way to remedy that, but it should be fun even just with my eye!

And quit arguing about noise!

-Taylor