Sub count

[Rossco72]

Probably been covered a hundred times before, but given we average when stacking subs, is there a minimum number needed to make it worthwhile?

If I have 60 subs, I can use something like subframe selector to assess just the very best ones or just to remove the very worst. If i just stacked the best 5 of 60 would I get a better average than if i stacked the best 50 of 60?

Cheers

Stuart

[ONIKKINEN]

You want to have the longest integration possible, but only have the best subs in it so you have to compromise. Look at the average quality of subs and deselect the ones that are clearly worse than the rest, but definitely dont reject 90% of the stuff. If you have a fair bit of variation in the data you should use some kind of weighting for the subs that are stacked, so that the worse ones have a lesser effect on the stack as the good ones. That way you can stack almost all of the data except for the clearly wind/high cloud/mount issue stricken ones.

I have seen the number 6 floated around for the minimum number of images that rejection algorithms can work with so less than this might result in satellite trails not being removed.

[Fegato]

It partly depends on your set up and how long your subs are, but you really want as many as you can use, as it's all about reducing SNR, and the more subs the better. I certainly wouldn't take the approach of only picking the "best" ones. I find I reject less and less. I use Pixinsight, and the pixel rejection algorithms allied to the improved weighting capabilities mean that it can pretty much sort it out for itself. I do like to blink through all the subs to see what I've got, and I will reject ones with very clear cloud impact or trailing stars, but I've largely stopped bothering about FWHM and eccentricity - I just let the software sort that out based on the weightings.

I have a RASA and it's a bit of an extreme instrument, so 30s is a max exposure for me in broadband. I'll normally be looking for at least 120 subs to stack, but can get good results with less that that. But I'll rarely count something as finished with less than 60 or so. As well as the need to get the best SNR, the pixel rejection algorithms just work better with lots of subs.

[Scott Badger]

I agree with Robin, plus some of the new tools like BlurXterminator widen the acceptable range given how much it can improve resolution and eccentricity. It's a dark art and everyone has their own stew recipe, but in the end knowing whether any single marginal sub is going to improve or diminish the integration, or for that matter the final image, is a guesstimate at best. Don't know if it's possible, but wouldn't it be great if for a given set of selected subs, PI's Subframe Selector could calculate what the stats of the integration would be.

For myself, it's hard not to let the blood-sweat-and-tears factor influence my decisions...... : )

Cheers,
Scott

[tomato]

I pretty much follow the methodology already outlined. I use APP which sorts the the subs on quality. I reject the obvious poor outliers (usually 3 or 4 in a stack of 100) but blink check every frame as APP can allocate a good score to a ‘knocked scope’ frame because of the double stars generated. APP assigns a weighting to each sub if in AUTO mode, so I let the software do the work.

[Rossco72]

5 hours ago, Fegato said:

It partly depends on your set up and how long your subs are, but you really want as many as you can use, as it's all about reducing SNR, and the more subs the better. I certainly wouldn't take the approach of only picking the "best" ones. I find I reject less and less. I use Pixinsight, and the pixel rejection algorithms allied to the improved weighting capabilities mean that it can pretty much sort it out for itself. I do like to blink through all the subs to see what I've got, and I will reject ones with very clear cloud impact or trailing stars, but I've largely stopped bothering about FWHM and eccentricity - I just let the software sort that out based on the weightings.

I have a RASA and it's a bit of an extreme instrument, so 30s is a max exposure for me in broadband. I'll normally be looking for at least 120 subs to stack, but can get good results with less that that. But I'll rarely count something as finished with less than 60 or so. As well as the need to get the best SNR, the pixel rejection algorithms just work better with lots of subs.

Interesting. Mainly i find I am doing Narrowband and either 5 minute or increasingly 10 minute exposures. Age old question I suppose, your 60 x 30s is only 3 of my 10m exposures in terms of photons, but I am not expert enough yet on understanding the impact on the noise. I only get 1 or 2 nights a month clear skies and I am loathe to spend one testing things out!

[Fegato]

4 minutes ago, Rossco72 said:

Interesting. Mainly i find I am doing Narrowband and either 5 minute or increasingly 10 minute exposures. Age old question I suppose, your 60 x 30s is only 3 of my 10m exposures in terms of photons, but I am not expert enough yet on understanding the impact on the noise. I only get 1 or 2 nights a month clear skies and I am loathe to spend one testing things out!

Yes, but the speed of the scope (focal ratio essentially) has a big impact. My RASA 11 is F/2.2.  This has roughly a 5x exposure "advantage" over F/5 for example and 10x over F/7. Obviously camera, target, sky conditions, and filters all come into play as well. Certainly narrowband will generally need longer exposures - I'm usually around 60s -180s when using a dual narrowband filter, depending on conditions.

[Rossco72]

15 hours ago, Fegato said:

Yes, but the speed of the scope (focal ratio essentially) has a big impact. My RASA 11 is F/2.2.  This has roughly a 5x exposure "advantage" over F/5 for example and 10x over F/7. Obviously camera, target, sky conditions, and filters all come into play as well. Certainly narrowband will generally need longer exposures - I'm usually around 60s -180s when using a dual narrowband filter, depending on conditions.

All true, I am at f4.7 so quick rather than fast lol. That RASA11 must suck in the light mind you...