| Which one? | |
|---|---|
| Just add time amigo! Only remove trailing frames and bad focussed ones | |
| Keep only the best, even if that means to cut your total time in half! | |
| Login to vote and view results. |
0.90
#...
·
 |
|---|
|
Hello to all! Coming back to the topic of what is the optimal zone of data quality VS quantity. Getting as much frames as you can even if that means being less than 20o from the horizon or keep only the good frames. Where is the Goldilocks zone? Adding an additional 30 frames to the final stack that look like this:  or ditch all of the above and keep only the frames that look like this:  In a nutshell which stack will be better at the end? Will it be the one that has 8 hours with a portion of average frames or the one which has 6 hours but with a higher signal amount? Clear skies to all! |
8.59
#...
·
·
1
like
|
|---|
Omiros Angelidis: Is this just due to atmospheric extinction? Or is this something else, like thin clouds passing through? If it is just extinction, then I'd say keep these subs, as they are going to be adding signal. Extinction should be pretty consistent across the field, and isn't really adding light (i.e. its not adding some kind of pollutant signal), so I think these subs could add value. If the issue is say passing clouds, then you might want to experiment and see whether including these subs or not improves the result. Passing clouds can completely change the nature of the signal throughout the frame, and can do so differently in each frame. Clouds can obscure the object and introduce their own, completely different signal. In that case, rejecting such frames is probably better than trying to include them. Often the "replacement" signal is not strong enough to just easily get rejected by an outlier rejection algorithm, and it will just muck with the overall quality of your object signals. If this is just LP, again, you might want to do some test integrations. I don't just see LP as a "gradient"...that gradient can move throughout the field, change dramatically with meridian flips, etc. As such, its not just a simple linear gradient that can very easily be modeled and removed...often, LP is an additional, structured signal that can be hard to remove. (Granted, modern gradient removal tools have improved significantly, but, they still avoid messing with any identifiable object signal whenever they can...meaning strong LP signal that is directly mixed with object signal can be irreparable.) If your LP is limited and consistent, it may be easy to remove. Once removed, the key factor would then be...does including the pollted frames actually IMPROVE the quality of your object signal, or detract from it. That quality should be evaluated on more than simply the amount of noise....color, structural accuracy, etc. should also be considered. Sometimes, taking a little less SNR for better structural differentiation and color is worth the tradeoff. When I first moved to imaging from a dark site, I tried to combine light polluted data. I ended up never doing that, as the polluted data often had a devastating impact to the overall quality (notably the contrast) of the dark site data. The dark site data alone was vastly superior to any version combined with light polluted data. So I would just use the dark site data only. |
4.02
#...
·
|
|---|
|
If I were you, I would stack everything together and only discard frames that have deformed stars (like due to bad guiding). Looks similar to what happened with me and my Orion nebula; I was taking 60 subs, and the hazier/crappier frames started appearing once the target got a little low on the horizon, but, I still stacked everything, and I think those last 20/30 hazier frames still benefitted my final image, but not by a very noticeable amount. My final image needed 3 background extractions though  |
