Jump to content

Banner.jpg.b89429c566825f6ab32bcafbada449c9.jpg

Is a stack of stacks the same as a longer stack?


Recommended Posts

I like live stacking as I can see the image develop and it saved my hard drive being cluttered up with thousands of images. 
The downside, I have found is that if a cheeky wisp of cloud comes over, it often gets included (in spite of trying to clamp down the FWHM settings). Some of my images then have a like a grey smear over them.
Then after and hour of live stacking I find it looks just like a greasy smear over the picture- which is a real pain to try and post process out.

Always seems to happen when I pop indoors to make a drink etc.

I noticed that there was a setting on SharpCap. to save the stack- save and reset. Which I have not used yet.
So Im thinking maybe I should do shorter live stacks eg. 30s for 5-10 minutes and then stack the results in A!S or DSS, so I can weed out the poorest sets, and I least I get something.
Gives me a chance to do any slight re-adjustment too.
Is a stack of stacks the same as doing a longer stack?

My rusty school maths says it should be - but thought I'd see if anyone else does this.
(I guess the setting wouldn't be there otherwise)

Thanks
 

Edited by Fraunhoffer
Link to comment
Share on other sites

A common question, much discussed.  If you search “signal stack noise” here on the forum you will get some interesting threads.

Recent similar discussion here...

Cue response from @vlaiv ...

Tony

Edited by AKB
  • Thanks 1
Link to comment
Share on other sites

If you stack from scratch, assuming each sub gets the right calibration files, you'll derive more benefit from the sigma routine. However, if you actually compare a real world re-stack with a stack of stacks, and you get the same result as I did, you may find the difference deeply underwhelming!

However, the strict answer to your question is 'no' and, if doing a restack from scratch is not too much of a faff, I'd do it that way.

Olly

Edited by ollypenrice
Link to comment
Share on other sites

17 minutes ago, AKB said:

A common question, much discussed.  If you search “signal stack noise” here on the forum you will get some interesting threads.

Recent similar discussion here...

Cue response from @vlaiv ...

Tony

Great - thank you for the link. Ive consumed that discussion with interest.
I see its a question asked by many before me.  I wonder if its a milestone on the path to better imaging met by many. 🙂
 
 

Link to comment
Share on other sites

11 minutes ago, Fraunhoffer said:

 I wonder if its a milestone on the path to better imaging met by many. 🙂
 
 

I think it is exactly that because people soon discover the value of more integration and want to add to what they already have.

Olly

  • Like 1
Link to comment
Share on other sites

5 minutes ago, ollypenrice said:

If you stack from scratch, assuming each sub gets the right calibration files, you'll derive more benefit from the sigma routine. However, if you actually compare a real world re-stack with a stack of stacks, and you get the same result as I did, you may find the difference deeply underwhelming!

However, the strict answer to your question is 'no' and, if doing a restack from scratch is not too much of a faff, I'd do it that way.

Olly

Hello Olly
Ill have a look at the sigma routine. I don't think I will often get the chance to restack. I have just about 2 1/2 - 3 hours from when the street lights turn off to the sun rising.
In that time I have to locate the object visually (no goto mount) and then set up the camera and take a picture. The picture is really there for my reference and to share if its any good - occasionally I have to use the greater sensitivity of the camera to actually 'see' the object.
I see one thing from the other discussion is to keep the number of subs the same in each stack before trying to re-stack those (assuming the snr is the same in each).
I guess my scenario might be stacks of 20x30s (10m) over an hour and one of these might be bad due to a cloud/mist/helicopter etc.
So at least the 'bad' sub-stack can be removed and that way it doesn't degrade the overall snr of the final stack.

 

Link to comment
Share on other sites

Re:   Is a stack of stacks the same as doing a longer stack?

My 2 bob's worth follows:-

Before one can answer this question one needs to develop a model of this system.

1. What does "stack" mean ?  If anyone can answer this question we may proceed from here.

If there is any form of non-linearity in this system (clipping etc.) then "hello Houston, we have a problem" .

2. If we make the grand assumption, that AFTER we have collected the original signal, the only noise in the system is white gaussian noise (WGN)  we may say that the stack of stacks (average of averages) is the same as stacking all the individual images.

We really need to know what the designer of the program has in mind and what is the science behind his/hers code. Perhaps if one is continually adding images to be "stacked"  then I would suggest we use an exponential average as ti has a "forgetting factor".

Jeremy
 

Link to comment
Share on other sites

1 hour ago, JRWASTRO said:

Re:   Is a stack of stacks the same as doing a longer stack?

My 2 bob's worth follows:-

Before one can answer this question one needs to develop a model of this system.

1. What does "stack" mean ?  If anyone can answer this question we may proceed from here.

If there is any form of non-linearity in this system (clipping etc.) then "hello Houston, we have a problem" .

2. If we make the grand assumption, that AFTER we have collected the original signal, the only noise in the system is white gaussian noise (WGN)  we may say that the stack of stacks (average of averages) is the same as stacking all the individual images.

We really need to know what the designer of the program has in mind and what is the science behind his/hers code. Perhaps if one is continually adding images to be "stacked"  then I would suggest we use an exponential average as ti has a "forgetting factor".

Jeremy
 

Point 1) As I use the term a 'stack' is the output from the combination of a set of sub exposures using one of a number of algorithms. So stacks of the same set of subs may vary depending on the algorithm chosen.

Point 2) I'd have thought that, if the stacks were made by a simple averaging of pixel values, then there would be little to choose between a re-stack and a stack of stacks properly weighted in accordance with their S/N ratio. However, if an outlier-rejection algorithlm is chosen then the outliers can be more effectively determined if the number of subs is higher.

No?

Link to comment
Share on other sites

5 minutes ago, ollypenrice said:

Point 1) As I use the term a 'stack' is the output from the combination of a set of sub exposures using one of a number of algorithms. So stacks of the same set of subs may vary depending on the algorithm chosen.

Point 2) I'd have thought that, if the stacks were made by a simple averaging of pixel values, then there would be little to choose between a re-stack and a stack of stacks properly weighted in accordance with their S/N ratio. However, if an outlier-rejection algorithlm is chosen then the outliers can be more effectively determined if the number of subs is higher.

No?

That would be assuming that I stacked all the sub-stacks.
I was imagining that lets say I can either stack 120x30s for 60 mins and have a bad 5-10 minutes somewhere in there and hope the greater number of good subs will outweigh the bad subs.
or,
I can do some 10 minute sub-stacks (say I do 20 x30s  in each, so Ive got 6), realise ive got a bad set after and so reject it, leaving me 5x(20x30s) to final stack.

or something like that - maybe 60-90 minutes total.
I don't spend days doing the same subject, whatever comes out in 60-90 minutes is it (even if I get that long) and to do the best I can with it.


🙂

 

 

Link to comment
Share on other sites

21 minutes ago, Fraunhoffer said:

That would be assuming that I stacked all the sub-stacks.
I was imagining that lets say I can either stack 120x30s for 60 mins and have a bad 5-10 minutes somewhere in there and hope the greater number of good subs will outweigh the bad subs.
or,
I can do some 10 minute sub-stacks (say I do 20 x30s  in each, so Ive got 6), realise ive got a bad set after and so reject it, leaving me 5x(20x30s) to final stack.

or something like that - maybe 60-90 minutes total.
I don't spend days doing the same subject, whatever comes out in 60-90 minutes is it (even if I get that long) and to do the best I can with it.


🙂

 

 

I wouldn't include any bad subs, ever. I would just reject them. They will only damage the final result. As for 'fewer and longer' or 'more but shorter,' I guess that depends on how risky the night is looking.

Olly

Link to comment
Share on other sites

Point 1.

Point 1) As I use the term a 'stack' is the output from the combination of a set of sub exposures using one of a number of algorithms. So stacks of the same set of subs may vary depending on the algorithm chosen.

We need to know what the particular algorithm does and the idea behind the procedure. Once we know this then we may take steps to answering the question as to whether there is gain to be made by "stacking" already stacked images. Possibly if there is some non-linearity like clipping or limiting is involved then things might get worse.

Point 2.

" simple averaging of pixel values " is the best strategy for estimating a signal in the presence of gaussian noise. It appears to be simple but the basis is very sound. It is impossible to separate signal from noise the best we may do is find the the best estimate of the signal in the presence of noise. The expected value of the noise is zero (0) so when we average the signal (signal + noise) the SNR after averaging  is higher (the noise power (variance) is reduced).

 

About good and bad subs:

Question : What is a determinant of the quality of an image frame ? We need to quantify this quality.  SNR is a start.  Is quality determined after we have removed as much of the systematic errors (Signal to Interference Ratio) as we can and look at the noise gradients in the image?  I  do not know.

Jeremy

 

Link to comment
Share on other sites

1 minute ago, vlaiv said:

Hopefully link to a thread provided by @AKB did answer original question, if not, I'm happy to participate in discussion on it and any follow up questions.

Yes thanks for the link. Its hopefully clear later on tonight so will try out a few variations.

 

Link to comment
Share on other sites

Hi

My original post on the 'Stack O Noise' subject purely related to SharpCap which can automatically live stack a user determined number of subs calibrated with a master light and dark. So I might collect, say, a stack of 10 x 60s subs then a bit later another 10. It does appear that stacking several different stacks isn't a great idea but with small numbers of lights the practical difference doesn't seem that much, especially when my greatest source of noise is skyglow/lp. I've since defaulted to saving each sub separately in addition to the live stacking so can also stack a collection of subs later :)

Louise

  • Thanks 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.