Acquiring more data per sub

[Bluesboystig]

Hi everyone, I have started recently in this incredible hobby and I'm full of questions, like many others I suppose. One thing I have noticed is that my images have an incredible amount of noise and not much data in them. My latest project was the Heart Nebula: on a really good night with no moon, I took 101 subs 2 minutes each at ISO 800 with my Canon eos 2000d and WO z61 with an Optolong L-Pro filter. Very disappointingly there wasn't much in each individual raw file, I can barely see any hint of nebulosity. I was then reading Siril processing tutorial and noticed the picture on top: one single RAW (with the same filter) had way more information in it than my entire stack of 3h and 20 mins! Am I missing something here? I feel like I should be getting more data per sub but I don't really understand what I'm doing wrong. I understand the Heart Nebula is an emission nebula so L-Pro is not great, but I don't think it completely kills data.

Any tips on how to approach this?

Thank you so much for the help and apologies if this sounds all a bit stupid, I'm just looking to learn

[smashing]

Is your camera modded? Just looking at that and what I managed with a modded camera in just over 2 hours of data and your picture looks great in comparison so well done.

It has also been pretty warm lately which my DSLR hated when I was using it in the warmer months.

I think I also was using longer 4/5 min subs as well but that has its own problems.

I think, if I'm reading it correctly, that tutorial is using a 2600mc pro which yes will have substantially different to your DSLR

[Bluesboystig]

19 minutes ago, smashing said:

Is your camera modded? Just looking at that and what I managed with a modded camera in just over 2 hours of data and your picture looks great in comparison so well done.

It has also been pretty warm lately which my DSLR hated when I was using it in the warmer months.

I think I also was using longer 4/5 min subs as well but that has its own problems.

I think, if I'm reading it correctly, that tutorial is using a 2600mc pro which yes will have substantially different to your DSLR

No my camera is just the normal unmodded one. Thanks for the praise, maybe my expectations are a bit too high? I know that that longer the total integration time the better, but I thought that 3.5h would give me a better SNR. It required so much stretching to get anything out of the stacked image that it brought out all the noise as well. 

[smashing]

15 minutes ago, Bluesboystig said:

No my camera is just the normal unmodded one. Thanks for the praise, maybe my expectations are a bit too high? I know that that longer the total integration time the better, but I thought that 3.5h would give me a better SNR. It required so much stretching to get anything out of the stacked image that it brought out all the noise as well. 

Yep I had similar things and I remember starting trying to aim for 5 plus hours if I could...the 2600MC blew me away when I got my first sub out of it, totally different animal so don't compare the two too much.

[vlaiv]

Maybe put it into numbers to help you out?

There are several things that contribute to signal level on single sub:

1. Sub duration - if you want to have good SNR on single sub - you need to expose for longer time. However, in context of stacking - this becomes somewhat moot point, since it is total exposure time that dictates final SNR. What is important however, is to swamp the read noise with single exposure / background noise levels (look up how to determine single exposure here on SGL - there are numerous threads)

2. Aperture size. Large aperture gathers more light than small aperture in same amount of time - more signal, better SNR

3. Pixel scale. Pixel covers part of the sky - larger part of the sky pixel covers - more signal it will record (this is strictly speaking true for only extended light sources - but those we are interested in primarily. No one complains that the stars are too faint - mostly it is nebulosity / galaxies that are faint).

4. Quantum efficiency of you system. This includes any losses in optical train, any filters used and finally QE of your camera

There are several constraints in above things - like it does not make sense to talk about aperture increase if you don't want to swap out the scope.

There is very limited sense in which we can talk about pixel scale. It generally depends on physical pixel size in microns and focal length of telescope. You can't change first and you can slightly alter second. You can add focal reducer that will change the pixel scale. There is a way to "alter" pixel size or pixel physical dimensions - and it is called binning. It is not as complicated topic per se - but it gets complicated when you account for OSC camera and so on.

You can alter QE of your system by changing or modifying your camera (since it is not modded).

[Bluesboystig]

1 hour ago, smashing said:

Yep I had similar things and I remember starting trying to aim for 5 plus hours if I could...the 2600MC blew me away when I got my first sub out of it, totally different animal so don't compare the two too much.

I just checked the price tag of the 2600MC.. my eyes are bleeding now 😔 

Would doing longer exposures improve the situation? In other words, is it super worth investing in an ASIAIR mini and guidescope etc for guiding and extending the subs time? Not sure the GTI can handle more than 2 mins without guiding... 

[alacant]

5 hours ago, Bluesboystig said:

was then reading Siril processing tutorial

Hi

When beginning, I strongly advise processing manually with Siril. Things then become far less of a mystery and by the time you have copied the files to where the script expects them, could be most of the way through the process anyway.

Go to the scripts when you've mastered the basics. You'll almost certainly -especially with a dSLR- stick with manual anyway; you control the outcome, not the script.

Cheers

Edited September 18, 2023 by alacant

[Elp]

Even to this day I avoid scripts as I want to see the result of every operation to make sure it's doing what it's supposed to. Siril is laid out well for you to do this methodologically.

You can continue to use your camera no issue, if you can get it astro modded or replace with one that will be better. But if you're going to do that might be best to get an astro camera in the first place which usually have much better quantum efficiencies. You don't necessarily need cooled either as I use uncooled ones often for DSO.

A lot of the result from AP comes from the post processing so if you're not proficient in it start learning all aspects.

[The Lazy Astronomer]

5 hours ago, Bluesboystig said:

Hi everyone, I have started recently in this incredible hobby and I'm full of questions, like many others I suppose. One thing I have noticed is that my images have an incredible amount of noise and not much data in them. My latest project was the Heart Nebula: on a really good night with no moon, I took 101 subs 2 minutes each at ISO 800 with my Canon eos 2000d and WO z61 with an Optolong L-Pro filter. Very disappointingly there wasn't much in each individual raw file, I can barely see any hint of nebulosity. I was then reading Siril processing tutorial and noticed the picture on top: one single RAW (with the same filter) had way more information in it than my entire stack of 3h and 20 mins! Am I missing something here? I feel like I should be getting more data per sub but I don't really understand what I'm doing wrong. I understand the Heart Nebula is an emission nebula so L-Pro is not great, but I don't think it completely kills data.

Any tips on how to approach this?

Thank you so much for the help and apologies if this sounds all a bit stupid, I'm just looking to learn

Bear in mind that as the heart nebula is an emission nebula, a large proportion of the signal will be coming from Ha emission, which is severely attenuated by a stock DSLR. The example below shows the difference between a stock DSLR and a typical astro UV/IR filter, with the Ha emission line highlighted. 

Short of modding your camera, or upgrading to a dedicated astro cam (there are much cheaper options than a 2600mc!), there's not a whole lot you can do.

[alacant]

7 hours ago, Bluesboystig said:

don't think it completely kills data.

Indeed. However all a filter can do is remove signal, never create it. Do without or isolate the emission with say, a UHC. No need to spend a fortune blocking even more light;)

You can see here that there's quite a bit more to be had from the data you already have. 

HTH

[Bluesboystig]

8 minutes ago, alacant said:

Indeed. However all a filter can do is remove signal, never create it. Do without or isolate the emission with say, a UHC. No need to spend a fortune blocking even more light;)

You can see here that there's quite a bit more to be had from the data you already have. 

HTH

Agree, there is definitely more to extract from that data. However, it's the amount of noise extracted with the signal that bothers me, hence my question really. 

[alacant]

1 hour ago, Bluesboystig said:

noise

What calibration frames are you using?

On Eos it's possible that noise is being introduced by applying bias and dark frames taken using the camera. 

We find that offset subtraction only gives much cleaner calibration. Simpler too.

Cheers 

[Bluesboystig]

6 minutes ago, alacant said:

What calibration frames are you using?

On Eos it's possible that noise is being introduced by applying bias and dark frames taken using the camera. 

We find that offset subtraction only gives much cleaner calibration. Simpler too.

Cheers 

Darks biases and flats. But I'm intending on re-preprocess everything manually doing the subtraction of the offset. Something that is not clear to me on this is whether I need to subtract 2048 both for the bias and darks or just for bias? Siril's documentation seems to suggest only for bias, darks are stacked in a master. (I know I have been asking lots of questions, so thank you very much for the help, really appreciated)

[alacant]

1 hour ago, Bluesboystig said:

darks are stacked in a master

Keep it simple.

2000d so don't use dark or in camera bias frames. This is almost certainly why you're seeing extra noise.
Simply remove the offset from the flat frames then stack them. Leave the same offset in place to calibrate the light frames. For the latter, don't forget to specify the master flat frame you just stacked.

Here's what it looks like for the flat frames:

 

... and the light frames:

 

Edited September 18, 2023 by alacant

[Bluesboystig]

5 hours ago, alacant said:

Keep it simple.

2000d so don't use dark or in camera bias frames. This is almost certainly why you're seeing extra noise.
Simply remove the offset from the flat frames then stack them. Leave the same offset in place to calibrate the light frames. For the latter, don't forget to specify the master flat frame you just stacked.

Here's what it looks like for the flat frames:

 

... and the light frames:

 

So I create a sequence for the flats subtracting 2048, then calibrate the lights sequence using the master flat and I keep subtracting 2048 for biases. I don’t do anything for darks at all? Then register everything and stack? Thanks so much for the help!!

[alacant]

3 hours ago, Bluesboystig said:

I don’t do anything for darks at all?

You don't use dark frames. 

BTW: you may want to check the offset value. Later eos models use 2048 but we have seen a few rogue sensors, IIRC, lower end 24mp versions.

Take a single cr2 frame in total darkness with a shutter of 1/4000s or shorter. Now in Siril, convert to fits and load.

Right click on the image > Statistics

We don't have any 24mp eos' ATM but here is a frame taken as described from a 1200d. the offset-bias value is then self explanatory:

 

 

Edited September 19, 2023 by alacant