Back to forums

How many subs to take

Acquisition Stacking H-alpha Deep Sky
6 replies•472 views
vikas chander avatar
vikas chander avatar
Hello fellow AstroBinners!!

I am trying to understand how many subs to take for a given target

so i ran my HA filter on the rosette nebula while the moon was up and got the following SNR results, no calibration only direct fast integration

15 frames = 38.02db
30 frames  = 40.81db
60 frames = 42.58db
111 frames = 44.83db

how shud i make sense of these figures - is the difference of 6db between 15 frames and 111 frames worth the effort

sounds crazy but the integration of 15 frames seemed better to me than the 111 frames version….

comments and thoughts of all flavors solicited….

vikas
Dave Rust avatar
Dave Rust avatar
Noise drops by 3dB each time the number of subs doubles. And noise reduction has to drop by roughly 9dB to look like half the noise of the original (single sub). The more one shoots, the better it gets, but there is a point of limited return. Each person has to decide for themselves.

1 Sub -0dB total reduction
2 Subs -3db
4 Subs -6dB
8 Subs -9dB (looks half as noisy as first sub)
16 Subs -12dB
32 Subs -15dB
64 Subs -18dB (looks half as noisy as 8 subs)
128 Subs -21dB
256 Subs -24dB
512 Subs -27dB total reduction (looks half as noisy as 64 subs)

 I am still new to the hobby after 3 years. I started out being happy with 32 subs for most targets. As is typical, I got pickier over time. Now, for noise reduction, I've made a personal goal of getting a minimum of 64 subs. The EDGE925 is 2350mm @ ƒ10, so most of my exposures are 300-600sec. I don't really want to spend more than 2 days on a target. I have gone as long as 128 subs...and man does that look nice, but it takes a week...IF the weather stays clear.

On the wider WO GT71 @ ƒ4.5, I can go for 128 more easily, as the exposure time may well be 60-90sec.

Dark nebulae are a challenge with the stronger EDGE. I have to run 10min subs. Reaching 64 of those takes a while! Fortunately, a lot of dark targets are very large, calling for the faster and wider WO anyway.

Even though noise reaches a point of limited improvement (somewhere around 64-128 subs, in my case), signal continues to improve with more subs. Serious folks will go for 20 hours or more of subs to bring out dark clouds and wisps of O2, dust, and ash. So noise is not the only consideration when deciding how many subs to shoot.

Check out a brighter target I finished just tonight. 64 subs over 2 nights. Pretty noise free!  https://astrob.in/m1197q/0/.

Of course we all have noise reduction strategies, and I've applied some to this target. But, the better the stack, the less is required, which protects the integrity of the targeted image.
Helpful Engaging
D. Jung avatar
D. Jung avatar
Mathematically speaking, you need 4 times the signal to get twice the SNR.
That is because photon signal is poisson distributed and SNR can be approximated as S/sqrt(S).
So technically after 4 times 15 frames you should gain 6db.
I would guess that not all of your frames have the same SNR which is why you don't reach that.
If you get worse image after integrating more signal it could be worth looking at your frames selection. Did you include many "bad" frames?
the question as to how many frames you need is mostly up to you and how much clear sky time you have available.

You got enough frames when YOU are happy with the result.
Helpful Insightful
Riccardo Civati avatar
Riccardo Civati avatar
Hi Vikas,
difficult to answer on a generic level because there are objects with weak halpha signals and others with intense signals.

Surely by integrating more you will have less noise in your master. Be careful that an image with a full moon is seen with a better SNR than a photo with less moon (but in reality the best photo is the one with less moon).

I do this for example: I start integrating 10 hours of halpha and create the master, then I integrate another 10 hours and if I see that the 20 hour master doesn't give better results, I stop.

I'm currently on SH2-274, I shot the first 15 hours and I saw that the weak signal was noisy so I integrated another 10 hours and it clearly improved. Now I'm at 31 hours but there's no difference with the 25 hour one so I'll stop here
Helpful
Brian Puhl avatar
Brian Puhl avatar
Ha is probably the cleanest signal you will get.  Testing your numbers on it I would think would give you inaccurate expectations.

Image stacking not only benefits your signal, but also benefits your pixel rejection… aka satellites, asteroids, etc.    So keep that in mind.   A 15 frame stack will look just fine, but if you're anywhere near the celestial equator, you're going to want alot more than 15 subs.   15 is not enough to reject satellites.

My general rule of thumb is I aim for 60 subs.   I break this rule often however.   60 IMO is the best because you get good pixel rejection and in most cases a clean signal.      I shoot much longer subs than the average person though.   Unless it's an extremely bright target, all my subs are 20 minutes for narrowband.  Honestly, I get most my signal within the first few hours.    The rest is just exposing the really faint stuff in the background which allows me to push the data farther than normal.

Modern sensors are much more sensitive than in the past, when these 'guidelines' were developed.     I have an image where I used only 6 subs per filter.   It came out just fine with strong signal (rosette actually).     I didn't have any satellites in my subs to reject, so stacking was a breeze.


TL: DR    It's about pixel rejection more than it is SNR with modern sensors.   You can shoot longer exposures and gain signal quicker, you just need to be able to reject.
Helpful Insightful Engaging Supportive
Barrington Russell avatar
Barrington Russell avatar
How ever many subs you have, the answer is you need twice as many.
Dave Rust avatar
Dave Rust avatar
TK: rejection is appreciated when standard deviation is used in stacking, you’re right. On some targets I get better results by averaging, which can make sats dimmer, but not eliminate them. And it is indeed about SNR for me. After dithering and stacking at least 64 subs, I sometimes don’t need noise reduction at all. This is especially true with the stronger signal from long subs, as you point out. Alas, I prefer medium length subs because of the time lost by having a 5 second sat transit ruin a 20 minute sub!
Helpful Insightful Respectful