How many images to stack in deep space photography?

[Jairo]

Hi,

I'm a newbie in deep space photography.

When I increase the exposition time, I end up with fewer images to stack, and sometimes the final result seems to have a worse signal-to-noise ratio. Does that make sense to you, or am I doing something very wrong? Is there a rule of thumb to choose the best amount of pictures for a given window of time?

Thanks.

[RickJ]

Proper stacking will always increase signal to noise ratio and result in a better image assuming all in the stack are of similar quality and exposure time. The improvement will be with the square root of the number of images in the stack. So 4 images will have twice the signal to noise ratio of 1. 9 will be 3 times better than 1 and 16 will be 4 times better.

Also if a single image has 4x the exposure time of another it will also have twice the signal to noise ratio. But you run into problems when increasing exposure time. One is sky brightness. It adds noise to the image. Once your exposure time is great enough that the sky noise is greater than the camera noise then there's no use taking longer exposures. There will be no gain. You have reached what we call the Sky Limited exposure. There are calculators on the net to help you with this calculation. Another problem with longer individual exposure time is that stars and the object itself start to saturate (over expose) if you take too long of an exposure. With many cameras and under very dark skies this can happen before you reach the Sky Limited exposure time. This is a problem I often have. Reducing exposure time means the individual subs are a bit noisier than they would be if I could expose longer. Other problems with longer exposure are how long your mount can track without tracking errors becoming a problem, longer times means a lot of time lost if some unforeseen event ruins an exposure. Also some scopes need refocusing at least every 30 minutes (mine can be a problem even faster when the night is cooling rapidly). For these and other reasons no one takes a single exposure. They use exposures that are sky limited if that is reasonable or as long as reasonable otherwise. While this may increase noise of a single image slightly but stacking a couple more this is overcome.

If you are seeing worse results after stacking similar quality and exposed images then yes you are doing something wrong. Beginners should start simple. Just calibrate the images using master darks of at least 6 darks median combined taken at the same time and temperature as the lights if your camera is one that requires darks. Many with Sony chips do not. In that case use a bad pixel map instead. Then flat field each frame again using a master of 5 or so flat frames exposed to an ADU count of 20,000 to 30,000 (all about the same) again median combined. While it is best to bias subtract the flat frames for now you can skip this step with very little loss of quality. Only a lot of math analysis of your image is likely to see a difference. Absolutely necessary for high quality photometric work however.

Once calibrated then align the images and then stack them using a basic average (mean) combine. Don't worry for now about fancy noise rejection combines as if those are set wrong they can create more issues than the correct.

You should now see a far better final image in line with the square root of the number of stacked frames. As signal to noise ratio improves the image becomes less grainy allowing finer detail to be seen. Faint features will stand out better from the background as well. If taking color data the colors will have less noise and thus can be more accurate with proper color balancing.

Note that nearly all the images I post here have too few sub frames for anything but processing as detailed above. It works very well, just leaves a few things like satellite trails and cosmic ray hits to deal with manually. Not a bit deal at all.

Rick

[Jairo]

Hi,

Thank you! Being sure of this will help me a lot. I'm taking pictures near urban lights, and now I'll concentrate on stacking more pictures with less exposure.

With my LXD75 and no autoguider nor periodic error correction, I was happy for being able to get some stable pictures with 1 min of exposition (in a VGA, ~20 arcmin field), which were not helping me much thought.

My camera is a DMK, and I saw that it does need dark frames. I was creating them right after a successfull session of light frames, but I was only creating four darks. I'll make more then.

I don't understand the difference between a dark frame and a bad pixel map. I notice my frames with an obstructed camera have pixels of different intensities, including a region which becomes lighter when I play with gama. Perhaps it's a part of the camera that becomes warmer and is near the chip.

What is ADU? The dimension of the exposure time is obvious to me, but I don't comprehend the meaning of the values of gain and brightness. Untill now, I have felt the need of flat frames only for Moon images with few frames. As I got more than 20 frames between 30 and 60 s, I was using the fancy methods just because the DeepSkyStacker automatically recommended them.

And finally, is there any free software that you recomend to correct the blur from the star trails? I will try the IRIS and already used the CCDSharp, but I think it would be nice do know more.

This was my first attempt, on Centaurus A. It is 20x60 s, 4x60 s darks, on DeepSkyStack and CCDSharp, with the lowest gain possible in the DMK configuration. Periodic errors messed with some of my frames. It also had a first-quarter Moon over the horizont yet, which sould have influenced in the sky limit a lot. I'm fully aware my CCD camera is proper for planetary imaging, so I was so happy to see my first 17 magnitude stars in this grainy picture.
Cen-A 2.jpg

I still have never been able to see it with my eyes on the telescope. So, it is like I got a new telescope!

[RickJ]

DMK cameras are fine for video. Very poor for long exposure deep sky work. Most are only 8 bit which doesn't allow for much improvement by stacking as it contains only 1/256th as much data as a 16 bit camera so not much to work with. They are made for stacking the best BRIGHT object frames such as planets and the moon. There the bright object is far above noise. Even then it is best to stack 50 or more frames. When used for deep sky work the noise so dominates it is very difficult to do much. Stick with very bright objects such as star clusters and some planetary nebula. What you have for Centaurus A is about as good as I'd expect it to be able to do. You won't begin to see what a deep sky camera can in such an object with that type of camera. Nor is a deep sky camera very good for imaging planets compared to the DMK. Each is the right tool for their particular job and poor for the other's job.

One thing I didn't mention is what happens when you stack images that are very short with a high noise camera designed for video. I didn't realize that was what you were using.

When using a high noise camera like the DMK (It is designed for ultra fast readout of video of bright objects in which the trade off of high noise in faint detail is worthwhile. Fast readout is always far noisier than slow. My camera takes 27 seconds to download what the DMK does 15 times a second for this reason.) you quickly run into a point where you just can't stack enough images to reduce noise much. In fact stacking 1000 DMK one minute images would have far more noise than my camera does in just a single one minute image when imaging galaxies. Even with my camera 60 one minute frames is nearly 3 times noisier than a stack of 6 10 minute frames. There are some cameras that can be used for 1 minute frames very effectively. All are cooled deep sky CCD cameras far more expensive than the DMK however.

DMK cameras are great for planetary work but very limited when it comes to deep sky work. Stick to star clusters and small bright nebula until you can afford a deep sky camera. The DMK then can be used as a guider camera to allow longer exposures. With guiding PEC isn't needed. In fact when I guide an image I turn off PEC.

Rick

[Jairo]

Thank you for the information! I knew cameras were specialized either to deep space or planetary, but nobody had ever told me such a complete explanation. By the way, how much worse a SLR digital camera would do, compared to a cooled, deep space CCD camera? Maybe I'll buy the former first.

[RickJ]

Today's Canon DSLR's (especially with a modified IR filter) are surprisingly good for deep sky work when run in RAW mode. My main objection is they teach some concepts that have to be unlearned when moving to a deep sky CCD camera. Like any OSC (one shot color) camera they have limitations compared to a mono CCD but have a couple minor advantages as well. There are whole forums devoted to imaging with these cameras on Yahoo and Cloudy Nights as well as other forum hosts.

They appear in the used market at some rather inviting prices. But you need to know which are the better ones for DSO work. Not following them I can't help you. See forums devoted to them for help picking the right one for you.

Their main limitations are less sensitivity than a CCD camera, more noise (but far less than a video camera), unable to do very effective narrow band imaging (3/4ths of the pixels contribute only noise to such an h alpha image) and color balance issues their users usually ignore. The latter is correctable just that few take the time to do so. Due to less sensitivity they require more total imaging time for a similar quality image than a mono deep sky CCD though again users used to higher noise seem to ignore this as well. The last two are some of the bad habits they can create. Still, used correctly they can take some very beautiful images and image surprisingly faint objects. Every year they get better. Deep sky CCD's aren't improving so quickly so they are catching up. Due to the type of dye filtration used rather than interference filtering of a mono CCD they will never be as sensitive but the real key is noise level. A cooled CCD still has a much better noise level but the gap gets smaller every year. If you live in a cold climate that will help as well.

If you are constantly imaging at temperatures above 25C then they are not all that great yet. Still far better than what you are using. Some companies now make a cooled housing for them for such climates. I have no idea if they work very well or not. The DSLR forums can help you there. http://www.cloudynights.com/ubbthrea...t/0/Board/DSLR

There are books on imaging with them: http://www.astropix.com/BGDA/BGDA.HTM and http://www.covingtoninnovations.com/dslr/ are two I found on Google. There are others. I've not read any so have no idea which are best.

Rick

[Jairo]

Hi,

This is a typical 60 s frame I used to produce the previous image, and the typical dark frame for the same configuration.
21.jpg01.jpg

Can the image still be improved with a darker sky? Does the noise of the camera still allows room for a longer explostion?

Thanks.

[RickJ]

It's impossible to tell anything from a JPG image. Try longer and see what happens.

With 60 in a stack you'd need to go to 240 to see a doubling of the signal to noise ratio. You are working with an 8 bit camera you are quite limited in what you can accomplish in DSO imaging. A 12 bit DSLR gives you 16 times more data to work with, a 16 bit camera far more. I started decades ago with an 8 bit camera far more capable than what you are trying to use but never got very good results. I did far better with film.

Of course darker skies help. I work from magnitude 6.75 skies and can get by with far less total time for the same signal to noise ratio. Since equipment has changed as well as skies I can't quantify it but the difference is very substantial.

Rick

[Jairo]

Hi,

By a magnitude 6.75 sky, do you mean a sky so dark that you can see stars of this magnitude by naked eye?

[RickJ]

Yes, at the zenith, not steady but seen to flicker in and out with averted vision.

Rick

[Jairo]

That's a feat! I can't see 4th magnitude stars with direct vision at my site.

[RickJ]

That's why you're fighting a loosing battle with that camera for galaxies. The first image you posted is about the best such a camera can do from Sao Paulo skies unless you get far away. Even then it will be severely limited by its high read noise and limited dynamic range. With that camera stick to what it can do, star clusters and a few bright planetary nebulae.

A low noise camera can do much better as you aren't fighting the high read noise of a video camera. It also has at least 16 times the dynamic range and far lower read noise meaning it can dig through the sky glow that much better. Then using narrow band filters that block nearly all your sky glow you can do amazing things with emission nebulae. Nearly on par with a very dark site in fact as 15 to 30 minute exposures will be possible depending on how narrow (and thus expensive) the filters are. Though you will always be bothered with the bright sky for wide band objects such as galaxies and reflection nebulae. Still the far higher dynamic range and low noise of a real DSO camera plus stacking can allow far better images of bright galaxies like Centaurus A than the video cameras can. You'll need many exposures but should be able to go 3 to 5 minutes before being sky limited.

Rick