2023-09-16: APP 2.0.0-beta23 has been released !
Improved performance again, CMD-A now works in macOS File Chooser, big improvement for bad column cosmetic correction, solved several bugs
We are very close now to releasing APP 2.0.0 stable with a complete printable manual...
[Sticky] Light integration settings with large number of subs
Most of tutorials that I found are dealing with small set of subs.
I have 80 subs and I'm wondering what are the good starting point settings for the light integration.
Good question 😉 I have made this a sticky
First of all, to get the fastest integration, try to work on the fastest harddrive that you have available.
For a large number of subs, let's say more than 40 light frames, I would advise the following settings in 6) INTEGRATE from top to bottom :
Set the weights on quality, this will give you (in most cases) an integration that is the most optimal for noise and sharpness.
Regarding the integration weights and their effects on integration , check this link:
Always use "average" integration mode. Only use median with small stacks. Average is superior for a large number of subs.
The composition mode will not have influence on the quality of integration, choose whichever you need, in most cases you want to use "full" to integrate the whole field of view of all your subs combined.
Local Normalization Correction: first make an integration without LNC. Then one with LNC at degree 1 and 1-2 iterations. This will give you an indication if your data needs this. If you have clearly changing gradients in your data, your integration will strongly benefit if you enable LNC. Be aware, LNC can take a long time to complete with a lot of frames. The progress of LNC can be clearly followed in the console panel.
I would suggest to always use MultiBandBlending iwth 5-10% to reduce stack artefacts at the borders of your integration field of view.
For outlier rejection, use sigma or winsorized sigma clipping, only 1 iteration with kappa of 3 to start with. This setting will not be destructive to your SNR and with a large amount of subs will easily remove all outliers.
Don't use linear fit clipping if you use LNC, linear fit clipping is redundant if you apply LNC.
The current implementation of linear fit clipping is really slow (needs improvement), so I wouldn't suggest to use it. Use LNC with sigma or winsorized sigma clipping.
You can enable output maps. These are usefull to get a better grip on what is happening with outlier rejection, MBB and the quality of normalization.
Pixel interpolation: use Lanczos 3 without no under /over shoot with a lot of frames. But if you were to enable it, the difference will be hard to see. It will prevent under and overshoot of the lanczos3 algorithm.
And finally start your integration.
Let me know if all of this is clear 😉 and if the result(s) are pleasing.
Thank you very much Mabula.
I used 110 frames. All setting like in post by Mabula. No LNC. Also trying 200 frames - it`s OK.
Thank you Roman for your feedback 😉