Let´s assume, my PC has not got enough RAM to stack 1000subs of a QHY268 camera in one go. I would devide the subs into portions of 100subs and stack them separately. After that, I got 10 stacked “super-subs” which I then could easily process in e.g. PI to get the “master stack” result.
Would this be the same result as if would have been able to stack all 1000 subs in one go? Or do I have to expect differences in overall quality?
My guess is, the final result would be identical - but I am not sure - what do you think?
CS
Georg
Stacking "Super-Subs"?
Respectful Engaging
Hi Georg,
This is an excellent way to handle large data sets in PI to help prevent memory crashes. One I use often with full frame sensors due to their large file size. You just want to make sure your smallest subset has enough images for accurate pixel rejection. I believe the recommended minimum number is around 25 frames. I don’t know if the result would be identical in your question but it would certainly be a very good result.
CS,
Jeff
Helpful Respectful Concise Engaging Supportive
Georg N. Nyman · Jun 27, 2026, 08:47 AM
Let´s assume, my PC has not got enough RAM to stack 1000subs of a QHY268 camera in one go. I would devide the subs into portions of 100subs and stack them separately. After that, I got 10 stacked “super-subs” which I then could easily process in e.g. PI to get the “master stack” result.
Would this be the same result as if would have been able to stack all 1000 subs in one go? Or do I have to expect differences in overall quality?
My guess is, the final result would be identical - but I am not sure - what do you think?
CS
Georg
Yes, it is. Just stack them with no pixel rejection and scaling to SNR.
Thanks Andrea, I thought that it should be the same, but your confirmation makes me feeling safe now…
Respectful
Georg N. Nyman · Jun 27, 2026 at 08:47 AM
Let´s assume, my PC has not got enough RAM to stack 1000subs of a QHY268 camera in one go. I would devide the subs into portions of 100subs and stack them separately. After that, I got 10 stacked “super-subs” which I then could easily process in e.g. PI to get the “master stack” result.
Would this be the same result as if would have been able to stack all 1000 subs in one go? Or do I have to expect differences in overall quality?
My guess is, the final result would be identical - but I am not sure - what do you think?
CS
Georg
Georg,
Yes, this will work; however, you have to be careful to perform a weighted average using the correct parameters to get a result that has the same statistics as a single stack. Be sure to use “Additive Normalization” using SNR as the weighing factor.
This is also how to create a super-LUM master by averaging a LUM stack master a RGB(LUM) master. Since PI won’t let you average two files, you have to trick it by making a duplicate copy of each files so that you can load four files into the ImageAveraging tool. I keep meaning to contact Juan to try to convince him to fix this limitation in the averaging tool.
John
Well written Helpful Respectful Concise Engaging Supportive