15th Feb 2024: Astro Pixel Processor 2.0.0-beta29 released - macOS native File Chooser, macOS CMD-Q fixed, read-only Fits on network fixed and other bug fixes
7th December 2023: added payment option Alipay to purchase Astro Pixel Processor from China, Hong Kong, Macau, Taiwan, Korea, Japan and other countries where Alipay is used.
I was just wondering, instead of saving individual subs, can I just save the final stacked .fits file? If I want to add more subs at a later time for the same subject, can I then stack the new subs with the previously stacked file? My thinking is that this should work, especially since in the integration phase, you have the option to weigh by integration time. The previously stacked file would have a very large integration time (8 hours, for example). Combining it with tonight's 30 minutes worth of data should obviously give more weight to the 8 hour stack than the 30 minute one.
Would that work? I'd like to not have to save terabytes of subs, if I can just save the final stack, and still be able to go back later and add more subs.
I think it should work, although I'm not 100% sure if it will work exactly like you would want (reducing the noise further and such) as now you use a few subs (with more average noise) and add that to 1 other "sub" (being the stack). What does work at least is two seperate, fully integrated results and combining those later. So two 8 hour sessions (with comparable noise levels) for instance will definitely improve it.
So, long story short, yes you can do that, but I'm not sure if it improves it with just a few subs added to it like that (@mabula-admin ?).
That was a bit of an extreme example. In actual practice, it would probably be more like adding another 4 or 8 hours at a later date.
But that's where the weight comes in. If one "sub" is 8 hours of data, and I wish to combine with another 4 hours worth of data, the by setting weight to "exposure", it will give the 8 hour sub twice the weight than the 4 hours worth.
At least that is my thinking, and just looking for some confirmation that it will actually work. 🙂
It will work, I would just test both situations and judge the quality. APP has been amazing in combining all sorts of data from all sorts of sources to produce realllly good results then any of those integrations separately. So it improves the quality without a doubt, I was just wondering about the noise levels when you combine an integration with more noise, indeed quality will then have a weight to the better stack, but it will still add signal from the "lesser" one.
I'll have to do as you say, and try it both ways and compare. If the results are just as good, then I can delete the old subs and just keep the integrated .fits file. If the results aren't just as good, then I have to buy more hard drives.
Hi @kross & @vincent-mod,
Let me give you some more insight in how this would work and what problems you might face if you are adding single subs to an already long integration of many subs.
Kross, you are correct that the exposure weights are the way to go in this case. The integration contains the cumulative exposure time in the meta data so that is compared to the exposure time of the individual subs.
Now, the main part here is how the weighting is actually done. This has to do with noise and the Signal To Noise ratio, or SNR. Theoretically, the noise drops with the square root of the exposure time:
An exposure that is 2x longer than another exposure (with all else being equal), will have the square root of 2 = 1.4 x less noise.
An exposure that is 4x longer than another exposure (with all else being equal), will have the square root of 4 = 2 x less noise.
So the weighting factors between all frames are relatively calculated and use the inverse square root of the exposure time = 1 / sq.rt(exposure time) ) . So that ensures that the weights are created properly to further reduce noise in your end result.
Now, with respect to adding frames to an integration, the biggest problem would be, I think, the use of Outlier Rejection Filters to remove all outlier signals like airplans, satellites, hot pixels etc.. from the frames that you add to the integration.
These outlier rejection filters work better if you use more frames. So integrating all data of all individual sub is always preferred, at least theoretically, practically it can be very cumbersome for 100s or even 1000s of frames. My suggestion would be to add at least 40 frames to the already existing integration to increase the quality of the resulting integration without introducing possible outliers that are present in the individual frames that you add. Then I think you will be fine actually 😉
Please let me know if this is clear and makes sense to you 😉 and if you have additional questions about this.
Kind regards,
Mabula
Thank you Mabula for your detailed response.
Just to be clear, I would never add just a single sub to a previous integration. If it was a single file, then it would be because it was a new integration of tonight's data, that I could then combine with previous data. But it wouldn't be just a single subexposure. So I don't think outlier rejection would be a problem.
