Old 08-05-2013, 09:59 PM
PRejto's Avatar
PRejto (Peter)
Registered User

PRejto is offline
Join Date: Jan 2011
Location: Sydney, Australia
Posts: 1,314
Processing Question CCDStack

I'm curious about the best method for dealing with large amounts of data. In this case I'm talking about say 50 images in luminance. Is it best to try to register and stack all 50 at once, or stack each evenings collection (say 10-15) and then combine the resulting stacks? Is the end result the same?

Reply With Quote
Old 09-05-2013, 12:18 AM
RickS's Avatar
RickS (Rick)
PI cult recruiter

RickS is offline
Join Date: Apr 2010
Location: Brisbane
Posts: 10,584
You will get different results from one big stack vs several smaller stacks due to the data rejection algorithm (just stacking without data rejection would give you the same result).

I'd definitely do one big stack in PixInsight. Experimenting with different rejection algorithms and comparing noise stats can make a significant difference to the result. Some algorithms tend to do better or worse depending on the number of subs. My experience is that the more data you have the better you can tune the algorithm and rejection parameters.

From memory, CCDStack doesn't give you as many options but I'd expect that you'd still get better results with more data. Perhaps you should try both strategies and see if you can notice a difference...

Reply With Quote
Old 09-05-2013, 07:26 AM
gregbradley's Avatar
Registered User

gregbradley is online now
Join Date: Feb 2006
Location: Sydney
Posts: 16,237
I have had this issue with CCDstack many times. I never have had 50 subs though. It would depend on your file size. CCDStack used to be very poor with memory management and large file sizes (like what you get with an 11002 camera on up) would bog it down terribly and you would get memory exception error messages if you tried to stack more than about 10 images.

CCDStack 2 seems to have improved its memory management and is less prone to that.

Even so I tend to stack no more than about 10 x 32mb images into groups.

Again it depends on your camera as well. My Proline 16803 is very clean so stacks of 5 to 10 submasters stacked together later would result in a very clean image anyway. A noisier camera may require larger numbers.

All these stacking routines come from statisical analysis maths. Basically for any of these to work effectively there needs to be a large enough sample so that outliers can be identified and removed by the maths.

In other words 2 images would not give you an outlier. Either one could be correct. But 3 now gives you an idea of what is normal. 6 for sure. 100 would be more reliable. So there would be a dimishing return on the improvement unless conditions were very variable between images.

But if you get a run of images that look much the same as you flick through them then the maths is not going to do much anyway. There is not much variability to detect and reject to get a normal sample.

I haven't used PI to do stacking and it sounds like it gives a nice report so you can see the differences but at the end the above theory is what is occurring no matter which program you are using.

You could work out a progression of more and more discriminating maths formulas. Plain average settles some differences. Median would too (mid points). A standard deviation after several samples starts to further refine what is a normal pixel value. square root of squared differences from a standard deviation is now discriminating more finely. Etc etc.

Reply With Quote
Old 09-05-2013, 08:15 PM
jase (Jason)
Registered User

jase is offline
Join Date: Sep 2006
Location: Melbourne, Victoria
Posts: 3,916
50 subs is quite large for CCDStack but it does depend on file size. Are you sure you'll use them all? Once you've got them all loaded, double click on one of the stars that doesn't look too saturated and get the FWHM measurement across the stack. You may want to remove the outliers with higher than expected FWHM values. This may cut down the stack size. The most I've had was 28 STL11k subs in a stack on my old system and found it was slow but still completely the task successfully.
Reply With Quote
Old 11-05-2013, 07:18 AM
PRejto's Avatar
PRejto (Peter)
Registered User

PRejto is offline
Join Date: Jan 2011
Location: Sydney, Australia
Posts: 1,314
Many thanks for all the answers!!

Reply With Quote


Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT +10. The time is now 07:38 PM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.
Lunatico Astronomical
Astronomy and Electronics Centre